diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,92635 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.9997084831406085,
+  "global_step": 15435,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.620689655172415e-07,
+      "loss": 10.1021,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.724137931034483e-06,
+      "loss": 10.0237,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.586206896551724e-06,
+      "loss": 8.6429,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.448275862068966e-06,
+      "loss": 8.4444,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.310344827586207e-06,
+      "loss": 7.8156,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.172413793103448e-06,
+      "loss": 7.2054,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.03448275862069e-06,
+      "loss": 6.8732,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.896551724137932e-06,
+      "loss": 6.2516,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.758620689655173e-06,
+      "loss": 5.8313,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.620689655172414e-06,
+      "loss": 5.374,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.482758620689655e-06,
+      "loss": 4.8527,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0344827586206897e-05,
+      "loss": 4.5063,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1206896551724138e-05,
+      "loss": 4.2828,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.206896551724138e-05,
+      "loss": 4.1085,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2931034482758622e-05,
+      "loss": 3.8599,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3793103448275863e-05,
+      "loss": 3.83,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4655172413793103e-05,
+      "loss": 3.6496,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5517241379310346e-05,
+      "loss": 3.4078,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6379310344827585e-05,
+      "loss": 3.2816,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7241379310344828e-05,
+      "loss": 2.9963,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.810344827586207e-05,
+      "loss": 2.9643,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.896551724137931e-05,
+      "loss": 2.8638,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9827586206896554e-05,
+      "loss": 2.8962,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0689655172413793e-05,
+      "loss": 2.9144,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1551724137931033e-05,
+      "loss": 2.8342,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2413793103448276e-05,
+      "loss": 2.7362,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.327586206896552e-05,
+      "loss": 2.6953,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.413793103448276e-05,
+      "loss": 2.7252,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5e-05,
+      "loss": 2.6563,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5862068965517244e-05,
+      "loss": 2.6083,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.672413793103448e-05,
+      "loss": 2.7005,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7586206896551727e-05,
+      "loss": 2.6455,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.844827586206897e-05,
+      "loss": 2.5143,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9310344827586206e-05,
+      "loss": 2.4283,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.017241379310345e-05,
+      "loss": 2.4716,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.103448275862069e-05,
+      "loss": 2.389,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1896551724137935e-05,
+      "loss": 2.4337,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.275862068965517e-05,
+      "loss": 2.3634,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3620689655172414e-05,
+      "loss": 2.2679,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4482758620689657e-05,
+      "loss": 2.2321,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.53448275862069e-05,
+      "loss": 2.2504,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.620689655172414e-05,
+      "loss": 2.1807,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7068965517241385e-05,
+      "loss": 2.2052,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.793103448275862e-05,
+      "loss": 2.2559,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8793103448275865e-05,
+      "loss": 2.1965,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.965517241379311e-05,
+      "loss": 2.1569,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0517241379310344e-05,
+      "loss": 2.165,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1379310344827587e-05,
+      "loss": 2.0582,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.224137931034483e-05,
+      "loss": 2.1486,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3103448275862066e-05,
+      "loss": 2.1625,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.396551724137931e-05,
+      "loss": 2.1665,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.482758620689655e-05,
+      "loss": 2.1538,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5689655172413794e-05,
+      "loss": 2.1776,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.655172413793104e-05,
+      "loss": 2.0997,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.741379310344828e-05,
+      "loss": 2.0254,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.827586206896552e-05,
+      "loss": 2.0228,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.913793103448276e-05,
+      "loss": 1.9735,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-05,
+      "loss": 1.9684,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.086206896551724e-05,
+      "loss": 1.9822,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.172413793103449e-05,
+      "loss": 1.9668,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.2586206896551724e-05,
+      "loss": 2.0435,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.344827586206896e-05,
+      "loss": 1.9674,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.431034482758621e-05,
+      "loss": 1.922,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.517241379310345e-05,
+      "loss": 1.9616,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.603448275862069e-05,
+      "loss": 1.9572,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.689655172413794e-05,
+      "loss": 1.952,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.7758620689655175e-05,
+      "loss": 1.9545,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.862068965517241e-05,
+      "loss": 1.8653,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.948275862068966e-05,
+      "loss": 1.9058,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.03448275862069e-05,
+      "loss": 1.9922,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.120689655172413e-05,
+      "loss": 1.9083,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.206896551724138e-05,
+      "loss": 1.7585,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.293103448275862e-05,
+      "loss": 1.9061,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.379310344827587e-05,
+      "loss": 1.7501,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.46551724137931e-05,
+      "loss": 1.8271,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.551724137931034e-05,
+      "loss": 1.907,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.637931034482759e-05,
+      "loss": 1.8452,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.724137931034483e-05,
+      "loss": 1.7608,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.810344827586206e-05,
+      "loss": 1.8126,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.896551724137931e-05,
+      "loss": 1.8631,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.982758620689655e-05,
+      "loss": 1.8006,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.06896551724138e-05,
+      "loss": 1.8557,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.155172413793105e-05,
+      "loss": 1.8234,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.241379310344828e-05,
+      "loss": 1.7501,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.327586206896552e-05,
+      "loss": 1.8465,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.413793103448277e-05,
+      "loss": 1.7556,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 1.7828,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.586206896551724e-05,
+      "loss": 1.692,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.672413793103449e-05,
+      "loss": 1.7522,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.758620689655173e-05,
+      "loss": 1.7137,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.844827586206897e-05,
+      "loss": 1.6936,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.931034482758621e-05,
+      "loss": 1.7395,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.017241379310345e-05,
+      "loss": 1.7509,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.103448275862069e-05,
+      "loss": 1.7144,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.189655172413794e-05,
+      "loss": 1.6705,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.275862068965517e-05,
+      "loss": 1.7377,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.362068965517241e-05,
+      "loss": 1.7549,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.448275862068966e-05,
+      "loss": 1.6985,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.53448275862069e-05,
+      "loss": 1.7757,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.620689655172413e-05,
+      "loss": 1.7281,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.706896551724138e-05,
+      "loss": 1.6173,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.793103448275862e-05,
+      "loss": 1.7423,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.879310344827587e-05,
+      "loss": 1.7343,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.96551724137931e-05,
+      "loss": 1.6745,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.051724137931035e-05,
+      "loss": 1.7545,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.137931034482759e-05,
+      "loss": 1.6913,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.224137931034484e-05,
+      "loss": 1.7183,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.310344827586207e-05,
+      "loss": 1.6556,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.396551724137932e-05,
+      "loss": 1.6365,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.482758620689656e-05,
+      "loss": 1.7009,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.56896551724138e-05,
+      "loss": 1.6219,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.655172413793105e-05,
+      "loss": 1.6165,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.741379310344828e-05,
+      "loss": 1.6315,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.827586206896552e-05,
+      "loss": 1.6966,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.913793103448277e-05,
+      "loss": 1.648,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001,
+      "loss": 1.7087,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010086206896551724,
+      "loss": 1.6146,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010172413793103448,
+      "loss": 1.6942,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010258620689655174,
+      "loss": 1.7341,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010344827586206898,
+      "loss": 1.6456,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010431034482758621,
+      "loss": 1.6423,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010517241379310345,
+      "loss": 1.6095,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010603448275862069,
+      "loss": 1.7078,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010689655172413792,
+      "loss": 1.6474,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010775862068965518,
+      "loss": 1.6246,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010862068965517242,
+      "loss": 1.6735,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010948275862068966,
+      "loss": 1.7111,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001103448275862069,
+      "loss": 1.7256,
+      "step": 128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00011120689655172414,
+      "loss": 1.6489,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00011206896551724138,
+      "loss": 1.7141,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00011293103448275863,
+      "loss": 1.6431,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00011379310344827588,
+      "loss": 1.677,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00011465517241379311,
+      "loss": 1.6421,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00011551724137931035,
+      "loss": 1.6806,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00011637931034482759,
+      "loss": 1.6997,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00011724137931034482,
+      "loss": 1.5746,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00011810344827586209,
+      "loss": 1.575,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00011896551724137932,
+      "loss": 1.6685,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00011982758620689656,
+      "loss": 1.6454,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001206896551724138,
+      "loss": 1.5749,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00012155172413793103,
+      "loss": 1.6312,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00012241379310344827,
+      "loss": 1.6024,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00012327586206896553,
+      "loss": 1.6089,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00012413793103448277,
+      "loss": 1.6202,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000125,
+      "loss": 1.6033,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00012586206896551724,
+      "loss": 1.6386,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00012672413793103447,
+      "loss": 1.5965,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00012758620689655174,
+      "loss": 1.6152,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00012844827586206897,
+      "loss": 1.6454,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001293103448275862,
+      "loss": 1.7019,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00013017241379310345,
+      "loss": 1.5759,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00013103448275862068,
+      "loss": 1.6081,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00013189655172413792,
+      "loss": 1.6078,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00013275862068965518,
+      "loss": 1.6437,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00013362068965517242,
+      "loss": 1.6094,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00013448275862068965,
+      "loss": 1.6496,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001353448275862069,
+      "loss": 1.6625,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00013620689655172413,
+      "loss": 1.5625,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001370689655172414,
+      "loss": 1.6133,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00013793103448275863,
+      "loss": 1.5758,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00013879310344827586,
+      "loss": 1.6048,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001396551724137931,
+      "loss": 1.6069,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00014051724137931036,
+      "loss": 1.681,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001413793103448276,
+      "loss": 1.6427,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00014224137931034483,
+      "loss": 1.6563,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001431034482758621,
+      "loss": 1.6111,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00014396551724137933,
+      "loss": 1.6053,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00014482758620689657,
+      "loss": 1.6009,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001456896551724138,
+      "loss": 1.6304,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00014655172413793104,
+      "loss": 1.6432,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00014741379310344828,
+      "loss": 1.4854,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00014827586206896554,
+      "loss": 1.5659,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00014913793103448278,
+      "loss": 1.7268,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 1.8055,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00015086206896551725,
+      "loss": 1.5363,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00015172413793103449,
+      "loss": 1.5984,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00015258620689655175,
+      "loss": 1.6348,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00015344827586206899,
+      "loss": 1.6465,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00015431034482758622,
+      "loss": 1.6507,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00015517241379310346,
+      "loss": 1.6425,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001560344827586207,
+      "loss": 1.6651,
+      "step": 181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00015689655172413793,
+      "loss": 1.6279,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001577586206896552,
+      "loss": 1.5545,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00015862068965517243,
+      "loss": 1.5307,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00015948275862068967,
+      "loss": 1.6141,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001603448275862069,
+      "loss": 1.5631,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00016120689655172414,
+      "loss": 1.5747,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00016206896551724137,
+      "loss": 1.5665,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00016293103448275864,
+      "loss": 1.6169,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00016379310344827587,
+      "loss": 1.6975,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001646551724137931,
+      "loss": 1.6448,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00016551724137931035,
+      "loss": 1.5855,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00016637931034482758,
+      "loss": 1.5628,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00016724137931034482,
+      "loss": 1.6557,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00016810344827586208,
+      "loss": 1.5999,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00016896551724137932,
+      "loss": 1.5408,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00016982758620689655,
+      "loss": 1.6,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001706896551724138,
+      "loss": 1.4942,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00017155172413793103,
+      "loss": 1.5596,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00017241379310344826,
+      "loss": 1.596,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00017327586206896553,
+      "loss": 1.5497,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00017413793103448276,
+      "loss": 1.5903,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000175,
+      "loss": 1.5636,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00017586206896551723,
+      "loss": 1.6283,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00017672413793103447,
+      "loss": 1.5519,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00017758620689655173,
+      "loss": 1.5798,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00017844827586206897,
+      "loss": 1.4785,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001793103448275862,
+      "loss": 1.5783,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018017241379310347,
+      "loss": 1.6066,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001810344827586207,
+      "loss": 1.5518,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018189655172413794,
+      "loss": 1.5337,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018275862068965518,
+      "loss": 1.6136,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018362068965517244,
+      "loss": 1.5838,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018448275862068968,
+      "loss": 1.5419,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001853448275862069,
+      "loss": 1.5636,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018620689655172415,
+      "loss": 1.6437,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018706896551724139,
+      "loss": 1.5425,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018793103448275865,
+      "loss": 1.4652,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018879310344827589,
+      "loss": 1.586,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018965517241379312,
+      "loss": 1.5467,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019051724137931036,
+      "loss": 1.5762,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001913793103448276,
+      "loss": 1.5289,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019224137931034483,
+      "loss": 1.5981,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001931034482758621,
+      "loss": 1.5816,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019396551724137933,
+      "loss": 1.6323,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019482758620689657,
+      "loss": 1.5943,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001956896551724138,
+      "loss": 1.553,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019655172413793104,
+      "loss": 1.6126,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019741379310344827,
+      "loss": 1.5756,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019827586206896554,
+      "loss": 1.5366,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019913793103448277,
+      "loss": 1.5089,
+      "step": 231
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 1.5453,
+      "step": 232
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00020086206896551727,
+      "loss": 1.5644,
+      "step": 233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00020172413793103448,
+      "loss": 1.5668,
+      "step": 234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00020258620689655175,
+      "loss": 1.5551,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00020344827586206895,
+      "loss": 1.5641,
+      "step": 236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00020431034482758622,
+      "loss": 1.6024,
+      "step": 237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00020517241379310348,
+      "loss": 1.5417,
+      "step": 238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002060344827586207,
+      "loss": 1.5861,
+      "step": 239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00020689655172413795,
+      "loss": 1.6076,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00020775862068965516,
+      "loss": 1.5059,
+      "step": 241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00020862068965517243,
+      "loss": 1.5454,
+      "step": 242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00020948275862068963,
+      "loss": 1.499,
+      "step": 243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002103448275862069,
+      "loss": 1.551,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00021120689655172416,
+      "loss": 1.5652,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00021206896551724137,
+      "loss": 1.5537,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00021293103448275863,
+      "loss": 1.5791,
+      "step": 247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00021379310344827584,
+      "loss": 1.6343,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002146551724137931,
+      "loss": 1.4551,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00021551724137931037,
+      "loss": 1.51,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00021637931034482758,
+      "loss": 1.6092,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00021724137931034484,
+      "loss": 1.6216,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00021810344827586208,
+      "loss": 1.4501,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00021896551724137931,
+      "loss": 1.5284,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00021982758620689655,
+      "loss": 1.4881,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002206896551724138,
+      "loss": 1.5206,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00022155172413793105,
+      "loss": 1.4672,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00022241379310344829,
+      "loss": 1.5472,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00022327586206896555,
+      "loss": 1.5942,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00022413793103448276,
+      "loss": 1.5317,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00022500000000000002,
+      "loss": 1.545,
+      "step": 261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00022586206896551726,
+      "loss": 1.547,
+      "step": 262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002267241379310345,
+      "loss": 1.532,
+      "step": 263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00022758620689655176,
+      "loss": 1.5304,
+      "step": 264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00022844827586206897,
+      "loss": 1.6273,
+      "step": 265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00022931034482758623,
+      "loss": 1.5407,
+      "step": 266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002301724137931035,
+      "loss": 1.4822,
+      "step": 267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002310344827586207,
+      "loss": 1.5061,
+      "step": 268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00023189655172413796,
+      "loss": 1.5009,
+      "step": 269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00023275862068965517,
+      "loss": 1.5273,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00023362068965517244,
+      "loss": 1.5032,
+      "step": 271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00023448275862068965,
+      "loss": 1.4956,
+      "step": 272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002353448275862069,
+      "loss": 1.417,
+      "step": 273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00023620689655172417,
+      "loss": 1.5564,
+      "step": 274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00023706896551724138,
+      "loss": 1.5144,
+      "step": 275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00023793103448275864,
+      "loss": 1.567,
+      "step": 276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00023879310344827585,
+      "loss": 1.4862,
+      "step": 277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00023965517241379312,
+      "loss": 1.5194,
+      "step": 278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00024051724137931038,
+      "loss": 1.4619,
+      "step": 279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002413793103448276,
+      "loss": 1.5442,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00024224137931034485,
+      "loss": 1.5151,
+      "step": 281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00024310344827586206,
+      "loss": 1.5774,
+      "step": 282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00024396551724137932,
+      "loss": 1.5138,
+      "step": 283
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00024482758620689653,
+      "loss": 1.555,
+      "step": 284
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002456896551724138,
+      "loss": 1.5544,
+      "step": 285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00024655172413793106,
+      "loss": 1.5001,
+      "step": 286
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002474137931034483,
+      "loss": 1.5863,
+      "step": 287
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00024827586206896553,
+      "loss": 1.5289,
+      "step": 288
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00024913793103448277,
+      "loss": 1.5529,
+      "step": 289
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00025,
+      "loss": 1.5301,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002508620689655173,
+      "loss": 1.536,
+      "step": 291
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002517241379310345,
+      "loss": 1.5673,
+      "step": 292
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00025258620689655177,
+      "loss": 1.5651,
+      "step": 293
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00025344827586206895,
+      "loss": 1.4524,
+      "step": 294
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00025431034482758624,
+      "loss": 1.4755,
+      "step": 295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002551724137931035,
+      "loss": 1.4398,
+      "step": 296
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002560344827586207,
+      "loss": 1.5474,
+      "step": 297
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00025689655172413795,
+      "loss": 1.5277,
+      "step": 298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002577586206896552,
+      "loss": 1.4263,
+      "step": 299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002586206896551724,
+      "loss": 1.5159,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00025948275862068966,
+      "loss": 1.5155,
+      "step": 301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002603448275862069,
+      "loss": 1.4738,
+      "step": 302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002612068965517242,
+      "loss": 1.5373,
+      "step": 303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00026206896551724137,
+      "loss": 1.5479,
+      "step": 304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00026293103448275866,
+      "loss": 1.5483,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00026379310344827584,
+      "loss": 1.5175,
+      "step": 306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00026465517241379313,
+      "loss": 1.4866,
+      "step": 307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00026551724137931036,
+      "loss": 1.5088,
+      "step": 308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002663793103448276,
+      "loss": 1.4895,
+      "step": 309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00026724137931034484,
+      "loss": 1.4882,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002681034482758621,
+      "loss": 1.5734,
+      "step": 311
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002689655172413793,
+      "loss": 1.5364,
+      "step": 312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00026982758620689655,
+      "loss": 1.5774,
+      "step": 313
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002706896551724138,
+      "loss": 1.5291,
+      "step": 314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00027155172413793107,
+      "loss": 1.4976,
+      "step": 315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00027241379310344825,
+      "loss": 1.4762,
+      "step": 316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00027327586206896554,
+      "loss": 1.4471,
+      "step": 317
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002741379310344828,
+      "loss": 1.4321,
+      "step": 318
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000275,
+      "loss": 1.5518,
+      "step": 319
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00027586206896551725,
+      "loss": 1.5189,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002767241379310345,
+      "loss": 1.5842,
+      "step": 321
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002775862068965517,
+      "loss": 1.4377,
+      "step": 322
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00027844827586206896,
+      "loss": 1.598,
+      "step": 323
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002793103448275862,
+      "loss": 1.5463,
+      "step": 324
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002801724137931035,
+      "loss": 1.4211,
+      "step": 325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002810344827586207,
+      "loss": 1.5058,
+      "step": 326
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00028189655172413796,
+      "loss": 1.5044,
+      "step": 327
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002827586206896552,
+      "loss": 1.5663,
+      "step": 328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00028362068965517243,
+      "loss": 1.5465,
+      "step": 329
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00028448275862068967,
+      "loss": 1.4737,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002853448275862069,
+      "loss": 1.551,
+      "step": 331
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002862068965517242,
+      "loss": 1.5004,
+      "step": 332
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002870689655172414,
+      "loss": 1.4783,
+      "step": 333
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00028793103448275867,
+      "loss": 1.4927,
+      "step": 334
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00028879310344827585,
+      "loss": 1.5123,
+      "step": 335
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00028965517241379314,
+      "loss": 1.5121,
+      "step": 336
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002905172413793104,
+      "loss": 1.4809,
+      "step": 337
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002913793103448276,
+      "loss": 1.4996,
+      "step": 338
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029224137931034485,
+      "loss": 1.4997,
+      "step": 339
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002931034482758621,
+      "loss": 1.5425,
+      "step": 340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002939655172413793,
+      "loss": 1.506,
+      "step": 341
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029482758620689656,
+      "loss": 1.4458,
+      "step": 342
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002956896551724138,
+      "loss": 1.5507,
+      "step": 343
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002965517241379311,
+      "loss": 1.4908,
+      "step": 344
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029741379310344827,
+      "loss": 1.4956,
+      "step": 345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029827586206896556,
+      "loss": 1.4044,
+      "step": 346
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029913793103448274,
+      "loss": 1.5293,
+      "step": 347
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00030000000000000003,
+      "loss": 1.5526,
+      "step": 348
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00030086206896551726,
+      "loss": 1.5043,
+      "step": 349
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003017241379310345,
+      "loss": 1.4724,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00030258620689655174,
+      "loss": 1.5534,
+      "step": 351
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00030344827586206897,
+      "loss": 1.5369,
+      "step": 352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003043103448275862,
+      "loss": 1.5374,
+      "step": 353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003051724137931035,
+      "loss": 1.4389,
+      "step": 354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003060344827586207,
+      "loss": 1.505,
+      "step": 355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00030689655172413797,
+      "loss": 1.4827,
+      "step": 356
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00030775862068965515,
+      "loss": 1.479,
+      "step": 357
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00030862068965517244,
+      "loss": 1.6265,
+      "step": 358
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003094827586206896,
+      "loss": 1.4617,
+      "step": 359
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003103448275862069,
+      "loss": 1.5116,
+      "step": 360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00031120689655172415,
+      "loss": 1.5238,
+      "step": 361
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003120689655172414,
+      "loss": 1.5381,
+      "step": 362
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003129310344827586,
+      "loss": 1.4137,
+      "step": 363
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00031379310344827586,
+      "loss": 1.4633,
+      "step": 364
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003146551724137931,
+      "loss": 1.5238,
+      "step": 365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003155172413793104,
+      "loss": 1.3865,
+      "step": 366
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00031637931034482757,
+      "loss": 1.4911,
+      "step": 367
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00031724137931034486,
+      "loss": 1.4433,
+      "step": 368
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003181034482758621,
+      "loss": 1.4698,
+      "step": 369
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00031896551724137933,
+      "loss": 1.4206,
+      "step": 370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00031982758620689657,
+      "loss": 1.5021,
+      "step": 371
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003206896551724138,
+      "loss": 1.5485,
+      "step": 372
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00032155172413793104,
+      "loss": 1.5107,
+      "step": 373
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003224137931034483,
+      "loss": 1.5473,
+      "step": 374
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00032327586206896557,
+      "loss": 1.4787,
+      "step": 375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00032413793103448275,
+      "loss": 1.553,
+      "step": 376
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00032500000000000004,
+      "loss": 1.4558,
+      "step": 377
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003258620689655173,
+      "loss": 1.4858,
+      "step": 378
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003267241379310345,
+      "loss": 1.5085,
+      "step": 379
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00032758620689655175,
+      "loss": 1.5783,
+      "step": 380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000328448275862069,
+      "loss": 1.4956,
+      "step": 381
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003293103448275862,
+      "loss": 1.5688,
+      "step": 382
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003301724137931035,
+      "loss": 1.4595,
+      "step": 383
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003310344827586207,
+      "loss": 1.5199,
+      "step": 384
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000331896551724138,
+      "loss": 1.4857,
+      "step": 385
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00033275862068965516,
+      "loss": 1.5615,
+      "step": 386
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00033362068965517246,
+      "loss": 1.5109,
+      "step": 387
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00033448275862068964,
+      "loss": 1.485,
+      "step": 388
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00033534482758620693,
+      "loss": 1.5382,
+      "step": 389
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00033620689655172416,
+      "loss": 1.4764,
+      "step": 390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003370689655172414,
+      "loss": 1.5354,
+      "step": 391
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00033793103448275864,
+      "loss": 1.462,
+      "step": 392
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00033879310344827587,
+      "loss": 1.5194,
+      "step": 393
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003396551724137931,
+      "loss": 1.5188,
+      "step": 394
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003405172413793104,
+      "loss": 1.5388,
+      "step": 395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003413793103448276,
+      "loss": 1.4076,
+      "step": 396
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00034224137931034487,
+      "loss": 1.439,
+      "step": 397
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00034310344827586205,
+      "loss": 1.5053,
+      "step": 398
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00034396551724137934,
+      "loss": 1.5236,
+      "step": 399
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003448275862068965,
+      "loss": 1.4828,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003456896551724138,
+      "loss": 1.5339,
+      "step": 401
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00034655172413793105,
+      "loss": 1.5017,
+      "step": 402
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003474137931034483,
+      "loss": 1.5229,
+      "step": 403
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003482758620689655,
+      "loss": 1.5199,
+      "step": 404
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00034913793103448276,
+      "loss": 1.4756,
+      "step": 405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00035,
+      "loss": 1.4971,
+      "step": 406
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003508620689655173,
+      "loss": 1.4724,
+      "step": 407
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00035172413793103447,
+      "loss": 1.5633,
+      "step": 408
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00035258620689655176,
+      "loss": 1.5673,
+      "step": 409
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00035344827586206894,
+      "loss": 1.5327,
+      "step": 410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00035431034482758623,
+      "loss": 1.457,
+      "step": 411
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00035517241379310347,
+      "loss": 1.5315,
+      "step": 412
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003560344827586207,
+      "loss": 1.458,
+      "step": 413
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00035689655172413794,
+      "loss": 1.4425,
+      "step": 414
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003577586206896552,
+      "loss": 1.4413,
+      "step": 415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003586206896551724,
+      "loss": 1.5346,
+      "step": 416
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00035948275862068965,
+      "loss": 1.3961,
+      "step": 417
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00036034482758620694,
+      "loss": 1.5267,
+      "step": 418
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003612068965517242,
+      "loss": 1.4994,
+      "step": 419
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003620689655172414,
+      "loss": 1.4213,
+      "step": 420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00036293103448275865,
+      "loss": 1.453,
+      "step": 421
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003637931034482759,
+      "loss": 1.4575,
+      "step": 422
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003646551724137931,
+      "loss": 1.4775,
+      "step": 423
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00036551724137931036,
+      "loss": 1.5033,
+      "step": 424
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003663793103448276,
+      "loss": 1.5267,
+      "step": 425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003672413793103449,
+      "loss": 1.4548,
+      "step": 426
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00036810344827586206,
+      "loss": 1.4056,
+      "step": 427
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00036896551724137935,
+      "loss": 1.4517,
+      "step": 428
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00036982758620689654,
+      "loss": 1.4961,
+      "step": 429
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003706896551724138,
+      "loss": 1.4675,
+      "step": 430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00037155172413793106,
+      "loss": 1.4724,
+      "step": 431
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003724137931034483,
+      "loss": 1.4541,
+      "step": 432
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00037327586206896554,
+      "loss": 1.5876,
+      "step": 433
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00037413793103448277,
+      "loss": 1.4827,
+      "step": 434
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000375,
+      "loss": 1.5317,
+      "step": 435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003758620689655173,
+      "loss": 1.4726,
+      "step": 436
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003767241379310345,
+      "loss": 1.512,
+      "step": 437
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00037758620689655177,
+      "loss": 1.5476,
+      "step": 438
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00037844827586206895,
+      "loss": 1.5087,
+      "step": 439
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00037931034482758624,
+      "loss": 1.5158,
+      "step": 440
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003801724137931035,
+      "loss": 1.5041,
+      "step": 441
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003810344827586207,
+      "loss": 1.4932,
+      "step": 442
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00038189655172413795,
+      "loss": 1.4891,
+      "step": 443
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003827586206896552,
+      "loss": 1.4148,
+      "step": 444
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003836206896551724,
+      "loss": 1.5505,
+      "step": 445
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00038448275862068966,
+      "loss": 1.5109,
+      "step": 446
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003853448275862069,
+      "loss": 1.4572,
+      "step": 447
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003862068965517242,
+      "loss": 1.5407,
+      "step": 448
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00038706896551724137,
+      "loss": 1.481,
+      "step": 449
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00038793103448275866,
+      "loss": 1.5444,
+      "step": 450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00038879310344827584,
+      "loss": 1.5347,
+      "step": 451
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00038965517241379313,
+      "loss": 1.5412,
+      "step": 452
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039051724137931037,
+      "loss": 1.4681,
+      "step": 453
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003913793103448276,
+      "loss": 1.4393,
+      "step": 454
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039224137931034484,
+      "loss": 1.4805,
+      "step": 455
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003931034482758621,
+      "loss": 1.5484,
+      "step": 456
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003939655172413793,
+      "loss": 1.4889,
+      "step": 457
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039482758620689655,
+      "loss": 1.4601,
+      "step": 458
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003956896551724138,
+      "loss": 1.4647,
+      "step": 459
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003965517241379311,
+      "loss": 1.4828,
+      "step": 460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003974137931034483,
+      "loss": 1.4368,
+      "step": 461
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039827586206896555,
+      "loss": 1.4625,
+      "step": 462
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003991379310344828,
+      "loss": 1.4996,
+      "step": 463
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004,
+      "loss": 1.4567,
+      "step": 464
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003999999955964988,
+      "loss": 1.5099,
+      "step": 465
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039999998238599525,
+      "loss": 1.4757,
+      "step": 466
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039999996036849003,
+      "loss": 1.5279,
+      "step": 467
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039999992954398405,
+      "loss": 1.4647,
+      "step": 468
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039999988991247867,
+      "loss": 1.4909,
+      "step": 469
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039999984147397573,
+      "loss": 1.5434,
+      "step": 470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039999978422847725,
+      "loss": 1.4473,
+      "step": 471
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039999971817598575,
+      "loss": 1.5415,
+      "step": 472
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039999964331650424,
+      "loss": 1.4876,
+      "step": 473
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003999995596500359,
+      "loss": 1.5275,
+      "step": 474
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003999994671765845,
+      "loss": 1.5134,
+      "step": 475
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039999936589615404,
+      "loss": 1.5228,
+      "step": 476
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003999992558087491,
+      "loss": 1.4936,
+      "step": 477
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039999913691437437,
+      "loss": 1.4943,
+      "step": 478
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003999990092130352,
+      "loss": 1.4479,
+      "step": 479
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003999988727047372,
+      "loss": 1.472,
+      "step": 480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039999872738948636,
+      "loss": 1.409,
+      "step": 481
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039999857326728904,
+      "loss": 1.4869,
+      "step": 482
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039999841033815206,
+      "loss": 1.4662,
+      "step": 483
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039999823860208263,
+      "loss": 1.4685,
+      "step": 484
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039999805805908823,
+      "loss": 1.5099,
+      "step": 485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003999978687091769,
+      "loss": 1.499,
+      "step": 486
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000399997670552357,
+      "loss": 1.5521,
+      "step": 487
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039999746358863707,
+      "loss": 1.4335,
+      "step": 488
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039999724781802643,
+      "loss": 1.4888,
+      "step": 489
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039999702324053447,
+      "loss": 1.5784,
+      "step": 490
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039999678985617115,
+      "loss": 1.4829,
+      "step": 491
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003999965476649466,
+      "loss": 1.4435,
+      "step": 492
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003999962966668717,
+      "loss": 1.4606,
+      "step": 493
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003999960368619574,
+      "loss": 1.4792,
+      "step": 494
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039999576825021507,
+      "loss": 1.4801,
+      "step": 495
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003999954908316566,
+      "loss": 1.5432,
+      "step": 496
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039999520460629425,
+      "loss": 1.4725,
+      "step": 497
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003999949095741406,
+      "loss": 1.4759,
+      "step": 498
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003999946057352086,
+      "loss": 1.4814,
+      "step": 499
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003999942930895117,
+      "loss": 1.4789,
+      "step": 500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003999939716370635,
+      "loss": 1.428,
+      "step": 501
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003999936413778784,
+      "loss": 1.5432,
+      "step": 502
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003999933023119708,
+      "loss": 1.4321,
+      "step": 503
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003999929544393556,
+      "loss": 1.5491,
+      "step": 504
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039999259776004817,
+      "loss": 1.4919,
+      "step": 505
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039999223227406424,
+      "loss": 1.5796,
+      "step": 506
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003999918579814199,
+      "loss": 1.5067,
+      "step": 507
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003999914748821316,
+      "loss": 1.5083,
+      "step": 508
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039999108297621616,
+      "loss": 1.4908,
+      "step": 509
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000399990682263691,
+      "loss": 1.4914,
+      "step": 510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003999902727445736,
+      "loss": 1.5459,
+      "step": 511
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039998985441888206,
+      "loss": 1.4513,
+      "step": 512
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039998942728663485,
+      "loss": 1.5659,
+      "step": 513
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003999889913478507,
+      "loss": 1.4838,
+      "step": 514
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039998854660254875,
+      "loss": 1.4163,
+      "step": 515
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003999880930507488,
+      "loss": 1.4789,
+      "step": 516
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039998763069247056,
+      "loss": 1.541,
+      "step": 517
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039998715952773464,
+      "loss": 1.5027,
+      "step": 518
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003999866795565616,
+      "loss": 1.4445,
+      "step": 519
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039998619077897266,
+      "loss": 1.4831,
+      "step": 520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039998569319498937,
+      "loss": 1.5539,
+      "step": 521
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039998518680463356,
+      "loss": 1.3426,
+      "step": 522
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039998467160792757,
+      "loss": 1.46,
+      "step": 523
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003999841476048941,
+      "loss": 1.5175,
+      "step": 524
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003999836147955562,
+      "loss": 1.4759,
+      "step": 525
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003999830731799373,
+      "loss": 1.3873,
+      "step": 526
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003999825227580613,
+      "loss": 1.5521,
+      "step": 527
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003999819635299525,
+      "loss": 1.3589,
+      "step": 528
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039998139549563536,
+      "loss": 1.5372,
+      "step": 529
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039998081865513503,
+      "loss": 1.4664,
+      "step": 530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003999802330084769,
+      "loss": 1.6005,
+      "step": 531
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003999796385556867,
+      "loss": 1.3984,
+      "step": 532
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003999790352967907,
+      "loss": 1.5039,
+      "step": 533
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039997842323181526,
+      "loss": 1.4335,
+      "step": 534
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039997780236078755,
+      "loss": 1.476,
+      "step": 535
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003999771726837348,
+      "loss": 1.4148,
+      "step": 536
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003999765342006848,
+      "loss": 1.3693,
+      "step": 537
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039997588691166564,
+      "loss": 1.5193,
+      "step": 538
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003999752308167058,
+      "loss": 1.4265,
+      "step": 539
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003999745659158342,
+      "loss": 1.5133,
+      "step": 540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003999738922090801,
+      "loss": 1.4218,
+      "step": 541
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003999732096964732,
+      "loss": 1.4019,
+      "step": 542
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003999725183780435,
+      "loss": 1.4545,
+      "step": 543
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003999718182538215,
+      "loss": 1.4006,
+      "step": 544
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000399971109323838,
+      "loss": 1.5145,
+      "step": 545
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039997039158812416,
+      "loss": 1.4644,
+      "step": 546
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003999696650467117,
+      "loss": 1.4708,
+      "step": 547
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039996892969963255,
+      "loss": 1.4969,
+      "step": 548
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003999681855469191,
+      "loss": 1.5018,
+      "step": 549
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039996743258860414,
+      "loss": 1.437,
+      "step": 550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003999666708247207,
+      "loss": 1.4601,
+      "step": 551
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003999659002553026,
+      "loss": 1.3763,
+      "step": 552
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003999651208803835,
+      "loss": 1.3681,
+      "step": 553
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039996433269999775,
+      "loss": 1.4478,
+      "step": 554
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003999635357141802,
+      "loss": 1.5215,
+      "step": 555
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039996272992296595,
+      "loss": 1.6047,
+      "step": 556
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003999619153263903,
+      "loss": 1.472,
+      "step": 557
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039996109192448924,
+      "loss": 1.409,
+      "step": 558
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003999602597172991,
+      "loss": 1.4217,
+      "step": 559
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039995941870485635,
+      "loss": 1.4386,
+      "step": 560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039995856888719815,
+      "loss": 1.4872,
+      "step": 561
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039995771026436186,
+      "loss": 1.3342,
+      "step": 562
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039995684283638535,
+      "loss": 1.4655,
+      "step": 563
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039995596660330674,
+      "loss": 1.3805,
+      "step": 564
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003999550815651647,
+      "loss": 1.4291,
+      "step": 565
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003999541877219982,
+      "loss": 1.3975,
+      "step": 566
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003999532850738465,
+      "loss": 1.5179,
+      "step": 567
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039995237362074943,
+      "loss": 1.5387,
+      "step": 568
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003999514533627471,
+      "loss": 1.3976,
+      "step": 569
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039995052429988003,
+      "loss": 1.4593,
+      "step": 570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039994958643218913,
+      "loss": 1.343,
+      "step": 571
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039994863975971576,
+      "loss": 1.4545,
+      "step": 572
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039994768428250154,
+      "loss": 1.5026,
+      "step": 573
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039994672000058854,
+      "loss": 1.5093,
+      "step": 574
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039994574691401926,
+      "loss": 1.3429,
+      "step": 575
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039994476502283653,
+      "loss": 1.4677,
+      "step": 576
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003999437743270836,
+      "loss": 1.4281,
+      "step": 577
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003999427748268041,
+      "loss": 1.4745,
+      "step": 578
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000399941766522042,
+      "loss": 1.4006,
+      "step": 579
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039994074941284173,
+      "loss": 1.5472,
+      "step": 580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039993972349924807,
+      "loss": 1.5439,
+      "step": 581
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003999386887813062,
+      "loss": 1.471,
+      "step": 582
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039993764525906173,
+      "loss": 1.414,
+      "step": 583
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003999365929325605,
+      "loss": 1.4241,
+      "step": 584
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039993553180184906,
+      "loss": 1.4489,
+      "step": 585
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003999344618669739,
+      "loss": 1.4247,
+      "step": 586
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003999333831279822,
+      "loss": 1.5111,
+      "step": 587
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039993229558492153,
+      "loss": 1.4755,
+      "step": 588
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039993119923783973,
+      "loss": 1.5041,
+      "step": 589
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003999300940867851,
+      "loss": 1.4963,
+      "step": 590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003999289801318063,
+      "loss": 1.4535,
+      "step": 591
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003999278573729524,
+      "loss": 1.4421,
+      "step": 592
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039992672581027277,
+      "loss": 1.4699,
+      "step": 593
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039992558544381726,
+      "loss": 1.393,
+      "step": 594
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003999244362736362,
+      "loss": 1.4251,
+      "step": 595
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003999232782997801,
+      "loss": 1.4127,
+      "step": 596
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003999221115222999,
+      "loss": 1.5408,
+      "step": 597
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003999209359412471,
+      "loss": 1.4633,
+      "step": 598
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039991975155667333,
+      "loss": 1.4241,
+      "step": 599
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039991855836863093,
+      "loss": 1.4109,
+      "step": 600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003999173563771722,
+      "loss": 1.4531,
+      "step": 601
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039991614558235026,
+      "loss": 1.433,
+      "step": 602
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039991492598421836,
+      "loss": 1.4252,
+      "step": 603
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039991369758283026,
+      "loss": 1.3573,
+      "step": 604
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039991246037823997,
+      "loss": 1.4792,
+      "step": 605
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000399911214370502,
+      "loss": 1.4858,
+      "step": 606
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039990995955967125,
+      "loss": 1.4199,
+      "step": 607
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039990869594580287,
+      "loss": 1.4884,
+      "step": 608
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003999074235289527,
+      "loss": 1.4663,
+      "step": 609
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003999061423091766,
+      "loss": 1.4678,
+      "step": 610
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039990485228653103,
+      "loss": 1.4878,
+      "step": 611
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003999035534610728,
+      "loss": 1.4078,
+      "step": 612
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039990224583285916,
+      "loss": 1.3087,
+      "step": 613
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039990092940194767,
+      "loss": 1.4123,
+      "step": 614
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003998996041683962,
+      "loss": 1.5195,
+      "step": 615
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003998982701322632,
+      "loss": 1.4197,
+      "step": 616
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039989692729360736,
+      "loss": 1.4207,
+      "step": 617
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003998955756524879,
+      "loss": 1.3995,
+      "step": 618
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003998942152089643,
+      "loss": 1.4504,
+      "step": 619
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003998928459630964,
+      "loss": 1.4106,
+      "step": 620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003998914679149445,
+      "loss": 1.4208,
+      "step": 621
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039989008106456947,
+      "loss": 1.4061,
+      "step": 622
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039988868541203213,
+      "loss": 1.433,
+      "step": 623
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003998872809573941,
+      "loss": 1.4815,
+      "step": 624
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039988586770071716,
+      "loss": 1.3893,
+      "step": 625
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003998844456420636,
+      "loss": 1.3919,
+      "step": 626
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003998830147814959,
+      "loss": 1.3838,
+      "step": 627
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003998815751190772,
+      "loss": 1.4536,
+      "step": 628
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003998801266548709,
+      "loss": 1.491,
+      "step": 629
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003998786693889407,
+      "loss": 1.4118,
+      "step": 630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039987720332135084,
+      "loss": 1.4605,
+      "step": 631
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039987572845216583,
+      "loss": 1.4941,
+      "step": 632
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003998742447814507,
+      "loss": 1.4899,
+      "step": 633
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003998727523092706,
+      "loss": 1.397,
+      "step": 634
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039987125103569147,
+      "loss": 1.4833,
+      "step": 635
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039986974096077924,
+      "loss": 1.4706,
+      "step": 636
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039986822208460055,
+      "loss": 1.4516,
+      "step": 637
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003998666944072222,
+      "loss": 1.3599,
+      "step": 638
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039986515792871146,
+      "loss": 1.42,
+      "step": 639
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039986361264913604,
+      "loss": 1.3729,
+      "step": 640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039986205856856397,
+      "loss": 1.446,
+      "step": 641
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039986049568706364,
+      "loss": 1.3983,
+      "step": 642
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003998589240047039,
+      "loss": 1.4498,
+      "step": 643
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039985734352155397,
+      "loss": 1.4584,
+      "step": 644
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039985575423768344,
+      "loss": 1.4697,
+      "step": 645
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003998541561531623,
+      "loss": 1.4818,
+      "step": 646
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003998525492680609,
+      "loss": 1.377,
+      "step": 647
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039985093358245005,
+      "loss": 1.4022,
+      "step": 648
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039984930909640085,
+      "loss": 1.4604,
+      "step": 649
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003998476758099848,
+      "loss": 1.3724,
+      "step": 650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003998460337232739,
+      "loss": 1.4365,
+      "step": 651
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003998443828363404,
+      "loss": 1.3793,
+      "step": 652
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003998427231492571,
+      "loss": 1.3821,
+      "step": 653
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003998410546620969,
+      "loss": 1.2778,
+      "step": 654
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003998393773749334,
+      "loss": 1.4206,
+      "step": 655
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039983769128784045,
+      "loss": 1.3388,
+      "step": 656
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003998359964008923,
+      "loss": 1.4005,
+      "step": 657
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039983429271416357,
+      "loss": 1.3815,
+      "step": 658
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003998325802277293,
+      "loss": 1.2911,
+      "step": 659
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003998308589416648,
+      "loss": 1.4652,
+      "step": 660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039982912885604604,
+      "loss": 1.4086,
+      "step": 661
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039982738997094904,
+      "loss": 1.4357,
+      "step": 662
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003998256422864505,
+      "loss": 1.4563,
+      "step": 663
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003998238858026273,
+      "loss": 1.4107,
+      "step": 664
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039982212051955677,
+      "loss": 1.375,
+      "step": 665
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003998203464373167,
+      "loss": 1.3214,
+      "step": 666
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003998185635559852,
+      "loss": 1.4244,
+      "step": 667
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003998167718756408,
+      "loss": 1.3062,
+      "step": 668
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039981497139636235,
+      "loss": 1.471,
+      "step": 669
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003998131621182292,
+      "loss": 1.3945,
+      "step": 670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003998113440413209,
+      "loss": 1.408,
+      "step": 671
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039980951716571763,
+      "loss": 1.352,
+      "step": 672
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003998076814914998,
+      "loss": 1.4132,
+      "step": 673
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003998058370187482,
+      "loss": 1.3415,
+      "step": 674
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039980398374754415,
+      "loss": 1.4037,
+      "step": 675
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003998021216779691,
+      "loss": 1.4045,
+      "step": 676
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003998002508101053,
+      "loss": 1.3478,
+      "step": 677
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003997983711440348,
+      "loss": 1.3361,
+      "step": 678
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039979648267984066,
+      "loss": 1.3804,
+      "step": 679
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039979458541760586,
+      "loss": 1.3371,
+      "step": 680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039979267935741407,
+      "loss": 1.4502,
+      "step": 681
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003997907644993491,
+      "loss": 1.3847,
+      "step": 682
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039978884084349535,
+      "loss": 1.3644,
+      "step": 683
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039978690838993755,
+      "loss": 1.3458,
+      "step": 684
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039978496713876077,
+      "loss": 1.4749,
+      "step": 685
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003997830170900504,
+      "loss": 1.4034,
+      "step": 686
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003997810582438925,
+      "loss": 1.4006,
+      "step": 687
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039977909060037313,
+      "loss": 1.4414,
+      "step": 688
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039977711415957907,
+      "loss": 1.4944,
+      "step": 689
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003997751289215973,
+      "loss": 1.3881,
+      "step": 690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039977313488651523,
+      "loss": 1.3871,
+      "step": 691
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003997711320544207,
+      "loss": 1.4016,
+      "step": 692
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003997691204254019,
+      "loss": 1.4103,
+      "step": 693
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039976709999954747,
+      "loss": 1.3944,
+      "step": 694
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003997650707769462,
+      "loss": 1.4595,
+      "step": 695
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003997630327576876,
+      "loss": 1.3358,
+      "step": 696
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003997609859418614,
+      "loss": 1.3758,
+      "step": 697
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003997589303295577,
+      "loss": 1.3392,
+      "step": 698
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000399756865920867,
+      "loss": 1.4265,
+      "step": 699
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00039975479271588024,
+      "loss": 1.4715,
+      "step": 700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00039975271071468867,
+      "loss": 1.4783,
+      "step": 701
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00039975061991738407,
+      "loss": 1.3897,
+      "step": 702
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00039974852032405837,
+      "loss": 1.4664,
+      "step": 703
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00039974641193480417,
+      "loss": 1.3797,
+      "step": 704
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00039974429474971416,
+      "loss": 1.3188,
+      "step": 705
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003997421687688817,
+      "loss": 1.2799,
+      "step": 706
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00039974003399240036,
+      "loss": 1.4444,
+      "step": 707
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003997378904203642,
+      "loss": 1.3507,
+      "step": 708
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00039973573805286743,
+      "loss": 1.3654,
+      "step": 709
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003997335768900051,
+      "loss": 1.407,
+      "step": 710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003997314069318721,
+      "loss": 1.3967,
+      "step": 711
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003997292281785642,
+      "loss": 1.4327,
+      "step": 712
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00039972704063017723,
+      "loss": 1.4095,
+      "step": 713
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003997248442868076,
+      "loss": 1.417,
+      "step": 714
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00039972263914855193,
+      "loss": 1.3755,
+      "step": 715
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003997204252155073,
+      "loss": 1.444,
+      "step": 716
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00039971820248777137,
+      "loss": 1.388,
+      "step": 717
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00039971597096544187,
+      "loss": 1.3549,
+      "step": 718
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00039971373064861713,
+      "loss": 1.4414,
+      "step": 719
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00039971148153739576,
+      "loss": 1.3321,
+      "step": 720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00039970922363187684,
+      "loss": 1.3376,
+      "step": 721
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00039970695693215976,
+      "loss": 1.405,
+      "step": 722
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00039970468143834436,
+      "loss": 1.422,
+      "step": 723
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00039970239715053077,
+      "loss": 1.462,
+      "step": 724
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003997001040688198,
+      "loss": 1.3887,
+      "step": 725
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003996978021933122,
+      "loss": 1.346,
+      "step": 726
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003996954915241093,
+      "loss": 1.3932,
+      "step": 727
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000399693172061313,
+      "loss": 1.4713,
+      "step": 728
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003996908438050255,
+      "loss": 1.3841,
+      "step": 729
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00039968850675534915,
+      "loss": 1.4638,
+      "step": 730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003996861609123869,
+      "loss": 1.4028,
+      "step": 731
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003996838062762421,
+      "loss": 1.4348,
+      "step": 732
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00039968144284701837,
+      "loss": 1.3949,
+      "step": 733
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00039967907062481986,
+      "loss": 1.4372,
+      "step": 734
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00039967668960975104,
+      "loss": 1.4011,
+      "step": 735
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003996742998019166,
+      "loss": 1.3696,
+      "step": 736
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00039967190120142196,
+      "loss": 1.3322,
+      "step": 737
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00039966949380837273,
+      "loss": 1.3429,
+      "step": 738
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003996670776228748,
+      "loss": 1.3918,
+      "step": 739
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00039966465264503463,
+      "loss": 1.3861,
+      "step": 740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00039966221887495897,
+      "loss": 1.3525,
+      "step": 741
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003996597763127551,
+      "loss": 1.3883,
+      "step": 742
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003996573249585305,
+      "loss": 1.3545,
+      "step": 743
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000399654864812393,
+      "loss": 1.3874,
+      "step": 744
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003996523958744512,
+      "loss": 1.3899,
+      "step": 745
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00039964991814481356,
+      "loss": 1.5213,
+      "step": 746
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003996474316235894,
+      "loss": 1.4545,
+      "step": 747
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00039964493631088803,
+      "loss": 1.3416,
+      "step": 748
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00039964243220681945,
+      "loss": 1.3819,
+      "step": 749
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003996399193114939,
+      "loss": 1.4639,
+      "step": 750
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000399637397625022,
+      "loss": 1.4424,
+      "step": 751
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00039963486714751477,
+      "loss": 1.4798,
+      "step": 752
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003996323278790837,
+      "loss": 1.4107,
+      "step": 753
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003996297798198406,
+      "loss": 1.3205,
+      "step": 754
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00039962722296989775,
+      "loss": 1.3578,
+      "step": 755
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003996246573293676,
+      "loss": 1.4013,
+      "step": 756
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003996220828983632,
+      "loss": 1.435,
+      "step": 757
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003996194996769979,
+      "loss": 1.4119,
+      "step": 758
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00039961690766538544,
+      "loss": 1.4468,
+      "step": 759
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00039961430686364,
+      "loss": 1.3295,
+      "step": 760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00039961169727187605,
+      "loss": 1.3154,
+      "step": 761
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003996090788902086,
+      "loss": 1.3024,
+      "step": 762
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00039960645171875284,
+      "loss": 1.3213,
+      "step": 763
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00039960381575762443,
+      "loss": 1.3607,
+      "step": 764
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003996011710069396,
+      "loss": 1.4158,
+      "step": 765
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003995985174668147,
+      "loss": 1.3844,
+      "step": 766
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003995958551373666,
+      "loss": 1.3989,
+      "step": 767
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003995931840187126,
+      "loss": 1.3995,
+      "step": 768
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003995905041109702,
+      "loss": 1.5124,
+      "step": 769
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00039958781541425746,
+      "loss": 1.4465,
+      "step": 770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00039958511792869284,
+      "loss": 1.4135,
+      "step": 771
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003995824116543951,
+      "loss": 1.3616,
+      "step": 772
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003995796965914833,
+      "loss": 1.3245,
+      "step": 773
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003995769727400772,
+      "loss": 1.4515,
+      "step": 774
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003995742401002965,
+      "loss": 1.3939,
+      "step": 775
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003995714986722618,
+      "loss": 1.4512,
+      "step": 776
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003995687484560936,
+      "loss": 1.2858,
+      "step": 777
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00039956598945191304,
+      "loss": 1.4374,
+      "step": 778
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003995632216598417,
+      "loss": 1.4711,
+      "step": 779
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003995604450800014,
+      "loss": 1.2956,
+      "step": 780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00039955765971251447,
+      "loss": 1.4461,
+      "step": 781
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003995548655575035,
+      "loss": 1.2526,
+      "step": 782
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003995520626150915,
+      "loss": 1.3433,
+      "step": 783
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000399549250885402,
+      "loss": 1.339,
+      "step": 784
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00039954643036855874,
+      "loss": 1.3896,
+      "step": 785
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00039954360106468596,
+      "loss": 1.3905,
+      "step": 786
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00039954076297390827,
+      "loss": 1.3504,
+      "step": 787
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00039953791609635057,
+      "loss": 1.3694,
+      "step": 788
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00039953506043213815,
+      "loss": 1.3872,
+      "step": 789
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00039953219598139704,
+      "loss": 1.3102,
+      "step": 790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003995293227442531,
+      "loss": 1.4115,
+      "step": 791
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000399526440720833,
+      "loss": 1.4043,
+      "step": 792
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003995235499112636,
+      "loss": 1.4051,
+      "step": 793
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003995206503156722,
+      "loss": 1.3845,
+      "step": 794
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00039951774193418645,
+      "loss": 1.4352,
+      "step": 795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003995148247669345,
+      "loss": 1.4126,
+      "step": 796
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00039951189881404475,
+      "loss": 1.3849,
+      "step": 797
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039950896407564605,
+      "loss": 1.4744,
+      "step": 798
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039950602055186766,
+      "loss": 1.4008,
+      "step": 799
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003995030682428392,
+      "loss": 1.4518,
+      "step": 800
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039950010714869064,
+      "loss": 1.4209,
+      "step": 801
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003994971372695524,
+      "loss": 1.4102,
+      "step": 802
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039949415860555514,
+      "loss": 1.3779,
+      "step": 803
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003994911711568302,
+      "loss": 1.344,
+      "step": 804
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039948817492350914,
+      "loss": 1.4564,
+      "step": 805
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039948516990572376,
+      "loss": 1.3388,
+      "step": 806
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039948215610360644,
+      "loss": 1.4181,
+      "step": 807
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003994791335172899,
+      "loss": 1.4144,
+      "step": 808
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039947610214690726,
+      "loss": 1.416,
+      "step": 809
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039947306199259196,
+      "loss": 1.2724,
+      "step": 810
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039947001305447793,
+      "loss": 1.3269,
+      "step": 811
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039946695533269936,
+      "loss": 1.3524,
+      "step": 812
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039946388882739094,
+      "loss": 1.3957,
+      "step": 813
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003994608135386877,
+      "loss": 1.2716,
+      "step": 814
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003994577294667251,
+      "loss": 1.2987,
+      "step": 815
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003994546366116389,
+      "loss": 1.3805,
+      "step": 816
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003994515349735653,
+      "loss": 1.4426,
+      "step": 817
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003994484245526408,
+      "loss": 1.4484,
+      "step": 818
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039944530534900255,
+      "loss": 1.3578,
+      "step": 819
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003994421773627878,
+      "loss": 1.5531,
+      "step": 820
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003994390405941342,
+      "loss": 1.2984,
+      "step": 821
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003994358950431801,
+      "loss": 1.3981,
+      "step": 822
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003994327407100638,
+      "loss": 1.3825,
+      "step": 823
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003994295775949243,
+      "loss": 1.3659,
+      "step": 824
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003994264056979009,
+      "loss": 1.3273,
+      "step": 825
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039942322501913325,
+      "loss": 1.363,
+      "step": 826
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003994200355587614,
+      "loss": 1.3605,
+      "step": 827
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003994168373169258,
+      "loss": 1.3934,
+      "step": 828
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003994136302937673,
+      "loss": 1.3534,
+      "step": 829
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003994104144894271,
+      "loss": 1.3222,
+      "step": 830
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039940718990404687,
+      "loss": 1.3961,
+      "step": 831
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039940395653776853,
+      "loss": 1.3731,
+      "step": 832
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039940071439073445,
+      "loss": 1.3235,
+      "step": 833
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003993974634630875,
+      "loss": 1.364,
+      "step": 834
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003993942037549708,
+      "loss": 1.3547,
+      "step": 835
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003993909352665278,
+      "loss": 1.3394,
+      "step": 836
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039938765799790255,
+      "loss": 1.3217,
+      "step": 837
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039938437194923923,
+      "loss": 1.3083,
+      "step": 838
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039938107712068265,
+      "loss": 1.3561,
+      "step": 839
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039937777351237787,
+      "loss": 1.3814,
+      "step": 840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003993744611244703,
+      "loss": 1.3398,
+      "step": 841
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039937113995710594,
+      "loss": 1.3449,
+      "step": 842
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039936781001043094,
+      "loss": 1.3681,
+      "step": 843
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003993644712845919,
+      "loss": 1.3523,
+      "step": 844
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039936112377973596,
+      "loss": 1.4023,
+      "step": 845
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003993577674960104,
+      "loss": 1.3721,
+      "step": 846
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039935440243356313,
+      "loss": 1.4036,
+      "step": 847
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003993510285925423,
+      "loss": 1.4165,
+      "step": 848
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039934764597309634,
+      "loss": 1.3613,
+      "step": 849
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003993442545753744,
+      "loss": 1.3433,
+      "step": 850
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003993408543995257,
+      "loss": 1.3889,
+      "step": 851
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00039933744544570006,
+      "loss": 1.3975,
+      "step": 852
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00039933402771404747,
+      "loss": 1.331,
+      "step": 853
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00039933060120471854,
+      "loss": 1.4241,
+      "step": 854
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003993271659178641,
+      "loss": 1.333,
+      "step": 855
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00039932372185363544,
+      "loss": 1.2999,
+      "step": 856
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00039932026901218416,
+      "loss": 1.3032,
+      "step": 857
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003993168073936624,
+      "loss": 1.2886,
+      "step": 858
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00039931333699822256,
+      "loss": 1.3628,
+      "step": 859
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00039930985782601745,
+      "loss": 1.3269,
+      "step": 860
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00039930636987720025,
+      "loss": 1.3911,
+      "step": 861
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003993028731519246,
+      "loss": 1.3767,
+      "step": 862
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00039929936765034445,
+      "loss": 1.3855,
+      "step": 863
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00039929585337261425,
+      "loss": 1.3301,
+      "step": 864
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003992923303188886,
+      "loss": 1.3851,
+      "step": 865
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003992887984893227,
+      "loss": 1.3241,
+      "step": 866
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003992852578840721,
+      "loss": 1.3637,
+      "step": 867
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00039928170850329274,
+      "loss": 1.2966,
+      "step": 868
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00039927815034714075,
+      "loss": 1.4559,
+      "step": 869
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003992745834157731,
+      "loss": 1.4433,
+      "step": 870
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003992710077093466,
+      "loss": 1.3186,
+      "step": 871
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00039926742322801884,
+      "loss": 1.3253,
+      "step": 872
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003992638299719476,
+      "loss": 1.345,
+      "step": 873
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003992602279412911,
+      "loss": 1.2812,
+      "step": 874
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00039925661713620804,
+      "loss": 1.3274,
+      "step": 875
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00039925299755685733,
+      "loss": 1.3434,
+      "step": 876
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00039924936920339844,
+      "loss": 1.4224,
+      "step": 877
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00039924573207599114,
+      "loss": 1.3574,
+      "step": 878
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00039924208617479545,
+      "loss": 1.4603,
+      "step": 879
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003992384314999721,
+      "loss": 1.3494,
+      "step": 880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000399234768051682,
+      "loss": 1.318,
+      "step": 881
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00039923109583008627,
+      "loss": 1.3055,
+      "step": 882
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00039922741483534684,
+      "loss": 1.4099,
+      "step": 883
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003992237250676257,
+      "loss": 1.418,
+      "step": 884
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00039922002652708537,
+      "loss": 1.4145,
+      "step": 885
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00039921631921388876,
+      "loss": 1.3599,
+      "step": 886
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000399212603128199,
+      "loss": 1.3498,
+      "step": 887
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00039920887827017974,
+      "loss": 1.3166,
+      "step": 888
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00039920514463999516,
+      "loss": 1.3642,
+      "step": 889
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00039920140223780954,
+      "loss": 1.3072,
+      "step": 890
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00039919765106378763,
+      "loss": 1.3903,
+      "step": 891
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00039919389111809473,
+      "loss": 1.3585,
+      "step": 892
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003991901224008964,
+      "loss": 1.3116,
+      "step": 893
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003991863449123585,
+      "loss": 1.3767,
+      "step": 894
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003991825586526474,
+      "loss": 1.3757,
+      "step": 895
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00039917876362192994,
+      "loss": 1.3872,
+      "step": 896
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003991749598203731,
+      "loss": 1.3464,
+      "step": 897
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003991711472481445,
+      "loss": 1.3146,
+      "step": 898
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003991673259054119,
+      "loss": 1.2594,
+      "step": 899
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003991634957923436,
+      "loss": 1.3403,
+      "step": 900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00039915965690910833,
+      "loss": 1.3632,
+      "step": 901
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00039915580925587514,
+      "loss": 1.3519,
+      "step": 902
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00039915195283281334,
+      "loss": 1.3492,
+      "step": 903
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00039914808764009293,
+      "loss": 1.356,
+      "step": 904
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003991442136778839,
+      "loss": 1.4518,
+      "step": 905
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000399140330946357,
+      "loss": 1.3438,
+      "step": 906
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00039913643944568315,
+      "loss": 1.3474,
+      "step": 907
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003991325391760337,
+      "loss": 1.3929,
+      "step": 908
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003991286301375804,
+      "loss": 1.4229,
+      "step": 909
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003991247123304954,
+      "loss": 1.3384,
+      "step": 910
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003991207857549514,
+      "loss": 1.3306,
+      "step": 911
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003991168504111209,
+      "loss": 1.2387,
+      "step": 912
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003991129062991775,
+      "loss": 1.3208,
+      "step": 913
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00039910895341929486,
+      "loss": 1.2946,
+      "step": 914
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00039910499177164686,
+      "loss": 1.3397,
+      "step": 915
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00039910102135640814,
+      "loss": 1.368,
+      "step": 916
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003990970421737535,
+      "loss": 1.3177,
+      "step": 917
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003990930542238581,
+      "loss": 1.2977,
+      "step": 918
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003990890575068976,
+      "loss": 1.3658,
+      "step": 919
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000399085052023048,
+      "loss": 1.309,
+      "step": 920
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003990810377724857,
+      "loss": 1.3689,
+      "step": 921
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003990770147553873,
+      "loss": 1.2778,
+      "step": 922
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003990729829719302,
+      "loss": 1.3495,
+      "step": 923
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003990689424222918,
+      "loss": 1.3176,
+      "step": 924
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00039906489310665,
+      "loss": 1.3279,
+      "step": 925
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003990608350251832,
+      "loss": 1.3972,
+      "step": 926
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003990567681780701,
+      "loss": 1.3461,
+      "step": 927
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003990526925654897,
+      "loss": 1.3749,
+      "step": 928
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003990486081876215,
+      "loss": 1.2891,
+      "step": 929
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00039904451504464537,
+      "loss": 1.4015,
+      "step": 930
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003990404131367416,
+      "loss": 1.3961,
+      "step": 931
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003990363024640907,
+      "loss": 1.3659,
+      "step": 932
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00039903218302687383,
+      "loss": 1.3133,
+      "step": 933
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003990280548252722,
+      "loss": 1.3219,
+      "step": 934
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00039902391785946777,
+      "loss": 1.3404,
+      "step": 935
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00039901977212964264,
+      "loss": 1.4313,
+      "step": 936
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003990156176359794,
+      "loss": 1.2642,
+      "step": 937
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00039901145437866095,
+      "loss": 1.3818,
+      "step": 938
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00039900728235787064,
+      "loss": 1.3585,
+      "step": 939
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00039900310157379217,
+      "loss": 1.3437,
+      "step": 940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00039899891202660963,
+      "loss": 1.3716,
+      "step": 941
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00039899471371650756,
+      "loss": 1.3471,
+      "step": 942
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003989905066436708,
+      "loss": 1.3494,
+      "step": 943
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003989862908082846,
+      "loss": 1.2878,
+      "step": 944
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00039898206621053465,
+      "loss": 1.3435,
+      "step": 945
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003989778328506069,
+      "loss": 1.3706,
+      "step": 946
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003989735907286878,
+      "loss": 1.3102,
+      "step": 947
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00039896933984496423,
+      "loss": 1.3468,
+      "step": 948
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003989650801996233,
+      "loss": 1.3938,
+      "step": 949
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003989608117928525,
+      "loss": 1.3309,
+      "step": 950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00039895653462484,
+      "loss": 1.3077,
+      "step": 951
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000398952248695774,
+      "loss": 1.3425,
+      "step": 952
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003989479540058432,
+      "loss": 1.3406,
+      "step": 953
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00039894365055523687,
+      "loss": 1.298,
+      "step": 954
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003989393383441443,
+      "loss": 1.2912,
+      "step": 955
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003989350173727556,
+      "loss": 1.3725,
+      "step": 956
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003989306876412609,
+      "loss": 1.3447,
+      "step": 957
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00039892634914985087,
+      "loss": 1.2331,
+      "step": 958
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00039892200189871666,
+      "loss": 1.3991,
+      "step": 959
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00039891764588804963,
+      "loss": 1.2684,
+      "step": 960
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00039891328111804153,
+      "loss": 1.3281,
+      "step": 961
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00039890890758888464,
+      "loss": 1.3909,
+      "step": 962
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003989045253007715,
+      "loss": 1.402,
+      "step": 963
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003989001342538952,
+      "loss": 1.327,
+      "step": 964
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000398895734448449,
+      "loss": 1.3996,
+      "step": 965
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00039889132588462664,
+      "loss": 1.3006,
+      "step": 966
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003988869085626223,
+      "loss": 1.3483,
+      "step": 967
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003988824824826305,
+      "loss": 1.4385,
+      "step": 968
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00039887804764484605,
+      "loss": 1.305,
+      "step": 969
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00039887360404946437,
+      "loss": 1.3153,
+      "step": 970
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00039886915169668103,
+      "loss": 1.4134,
+      "step": 971
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00039886469058669213,
+      "loss": 1.3409,
+      "step": 972
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003988602207196941,
+      "loss": 1.3452,
+      "step": 973
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003988557420958838,
+      "loss": 1.3222,
+      "step": 974
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003988512547154584,
+      "loss": 1.4443,
+      "step": 975
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00039884675857861557,
+      "loss": 1.3496,
+      "step": 976
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003988422536855532,
+      "loss": 1.321,
+      "step": 977
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003988377400364698,
+      "loss": 1.3681,
+      "step": 978
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00039883321763156405,
+      "loss": 1.3834,
+      "step": 979
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000398828686471035,
+      "loss": 1.3844,
+      "step": 980
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00039882414655508235,
+      "loss": 1.344,
+      "step": 981
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003988195978839059,
+      "loss": 1.3364,
+      "step": 982
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000398815040457706,
+      "loss": 1.2917,
+      "step": 983
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00039881047427668334,
+      "loss": 1.2357,
+      "step": 984
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000398805899341039,
+      "loss": 1.313,
+      "step": 985
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003988013156509744,
+      "loss": 1.2855,
+      "step": 986
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00039879672320669135,
+      "loss": 1.3493,
+      "step": 987
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00039879212200839213,
+      "loss": 1.3533,
+      "step": 988
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00039878751205627943,
+      "loss": 1.3774,
+      "step": 989
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00039878289335055616,
+      "loss": 1.3781,
+      "step": 990
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003987782658914256,
+      "loss": 1.3524,
+      "step": 991
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00039877362967909174,
+      "loss": 1.3634,
+      "step": 992
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003987689847137586,
+      "loss": 1.3606,
+      "step": 993
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003987643309956307,
+      "loss": 1.3802,
+      "step": 994
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00039875966852491316,
+      "loss": 1.2996,
+      "step": 995
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00039875499730181104,
+      "loss": 1.2838,
+      "step": 996
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00039875031732653014,
+      "loss": 1.371,
+      "step": 997
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003987456285992766,
+      "loss": 1.2935,
+      "step": 998
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003987409311202568,
+      "loss": 1.4285,
+      "step": 999
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00039873622488967764,
+      "loss": 1.3474,
+      "step": 1000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003987315099077464,
+      "loss": 1.3259,
+      "step": 1001
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00039872678617467063,
+      "loss": 1.4198,
+      "step": 1002
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003987220536906583,
+      "loss": 1.3577,
+      "step": 1003
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00039871731245591795,
+      "loss": 1.4114,
+      "step": 1004
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003987125624706582,
+      "loss": 1.2988,
+      "step": 1005
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003987078037350884,
+      "loss": 1.316,
+      "step": 1006
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003987030362494179,
+      "loss": 1.3194,
+      "step": 1007
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00039869826001385676,
+      "loss": 1.3196,
+      "step": 1008
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00039869347502861533,
+      "loss": 1.3488,
+      "step": 1009
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00039868868129390423,
+      "loss": 1.3477,
+      "step": 1010
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00039868387880993454,
+      "loss": 1.2713,
+      "step": 1011
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00039867906757691784,
+      "loss": 1.3311,
+      "step": 1012
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00039867424759506594,
+      "loss": 1.3716,
+      "step": 1013
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000398669418864591,
+      "loss": 1.3639,
+      "step": 1014
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00039866458138570576,
+      "loss": 1.3336,
+      "step": 1015
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003986597351586233,
+      "loss": 1.382,
+      "step": 1016
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00039865488018355687,
+      "loss": 1.3652,
+      "step": 1017
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00039865001646072033,
+      "loss": 1.4593,
+      "step": 1018
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00039864514399032784,
+      "loss": 1.3024,
+      "step": 1019
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000398640262772594,
+      "loss": 1.3895,
+      "step": 1020
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003986353728077337,
+      "loss": 1.3607,
+      "step": 1021
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003986304740959623,
+      "loss": 1.357,
+      "step": 1022
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003986255666374955,
+      "loss": 1.3318,
+      "step": 1023
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003986206504325494,
+      "loss": 1.4165,
+      "step": 1024
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003986157254813405,
+      "loss": 1.2692,
+      "step": 1025
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00039861079178408565,
+      "loss": 1.3249,
+      "step": 1026
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00039860584934100216,
+      "loss": 1.3525,
+      "step": 1027
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003986008981523076,
+      "loss": 1.4989,
+      "step": 1028
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00039859593821822003,
+      "loss": 1.259,
+      "step": 1029
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00039859096953895784,
+      "loss": 1.4081,
+      "step": 1030
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003985859921147399,
+      "loss": 1.3826,
+      "step": 1031
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00039858100594578527,
+      "loss": 1.3129,
+      "step": 1032
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003985760110323136,
+      "loss": 1.2719,
+      "step": 1033
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00039857100737454484,
+      "loss": 1.3204,
+      "step": 1034
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00039856599497269926,
+      "loss": 1.295,
+      "step": 1035
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003985609738269977,
+      "loss": 1.3954,
+      "step": 1036
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003985559439376611,
+      "loss": 1.3743,
+      "step": 1037
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003985509053049111,
+      "loss": 1.3887,
+      "step": 1038
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00039854585792896955,
+      "loss": 1.4199,
+      "step": 1039
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003985408018100587,
+      "loss": 1.4336,
+      "step": 1040
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00039853573694840106,
+      "loss": 1.3473,
+      "step": 1041
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003985306633442199,
+      "loss": 1.3718,
+      "step": 1042
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00039852558099773847,
+      "loss": 1.3705,
+      "step": 1043
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003985204899091806,
+      "loss": 1.3487,
+      "step": 1044
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00039851539007877045,
+      "loss": 1.3342,
+      "step": 1045
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003985102815067327,
+      "loss": 1.3554,
+      "step": 1046
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00039850516419329227,
+      "loss": 1.3413,
+      "step": 1047
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003985000381386744,
+      "loss": 1.3414,
+      "step": 1048
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003984949033431049,
+      "loss": 1.2311,
+      "step": 1049
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003984897598068099,
+      "loss": 1.3425,
+      "step": 1050
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003984846075300158,
+      "loss": 1.3898,
+      "step": 1051
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003984794465129496,
+      "loss": 1.3409,
+      "step": 1052
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003984742767558385,
+      "loss": 1.3354,
+      "step": 1053
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003984690982589102,
+      "loss": 1.2761,
+      "step": 1054
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00039846391102239267,
+      "loss": 1.3825,
+      "step": 1055
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003984587150465143,
+      "loss": 1.2523,
+      "step": 1056
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000398453510331504,
+      "loss": 1.4206,
+      "step": 1057
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00039844829687759096,
+      "loss": 1.3559,
+      "step": 1058
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00039844307468500455,
+      "loss": 1.4706,
+      "step": 1059
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00039843784375397497,
+      "loss": 1.279,
+      "step": 1060
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003984326040847325,
+      "loss": 1.3012,
+      "step": 1061
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00039842735567750787,
+      "loss": 1.337,
+      "step": 1062
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003984220985325322,
+      "loss": 1.2073,
+      "step": 1063
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003984168326500368,
+      "loss": 1.3562,
+      "step": 1064
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00039841155803025383,
+      "loss": 1.3084,
+      "step": 1065
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003984062746734154,
+      "loss": 1.2886,
+      "step": 1066
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003984009825797542,
+      "loss": 1.3171,
+      "step": 1067
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003983956817495033,
+      "loss": 1.3354,
+      "step": 1068
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003983903721828961,
+      "loss": 1.2522,
+      "step": 1069
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003983850538801664,
+      "loss": 1.271,
+      "step": 1070
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00039837972684154835,
+      "loss": 1.3178,
+      "step": 1071
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003983743910672766,
+      "loss": 1.2736,
+      "step": 1072
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00039836904655758605,
+      "loss": 1.3905,
+      "step": 1073
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003983636933127121,
+      "loss": 1.4036,
+      "step": 1074
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003983583313328904,
+      "loss": 1.392,
+      "step": 1075
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003983529606183572,
+      "loss": 1.3693,
+      "step": 1076
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003983475811693488,
+      "loss": 1.3087,
+      "step": 1077
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003983421929861023,
+      "loss": 1.3441,
+      "step": 1078
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003983367960688549,
+      "loss": 1.3002,
+      "step": 1079
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00039833139041784414,
+      "loss": 1.2612,
+      "step": 1080
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00039832597603330815,
+      "loss": 1.3888,
+      "step": 1081
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003983205529154854,
+      "loss": 1.3304,
+      "step": 1082
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003983151210646146,
+      "loss": 1.3686,
+      "step": 1083
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00039830968048093496,
+      "loss": 1.3445,
+      "step": 1084
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003983042311646862,
+      "loss": 1.3133,
+      "step": 1085
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00039829877311610797,
+      "loss": 1.3055,
+      "step": 1086
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00039829330633544093,
+      "loss": 1.3531,
+      "step": 1087
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003982878308229257,
+      "loss": 1.2905,
+      "step": 1088
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003982823465788033,
+      "loss": 1.2642,
+      "step": 1089
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00039827685360331534,
+      "loss": 1.2895,
+      "step": 1090
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003982713518967037,
+      "loss": 1.286,
+      "step": 1091
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00039826584145921055,
+      "loss": 1.263,
+      "step": 1092
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003982603222910787,
+      "loss": 1.3341,
+      "step": 1093
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00039825479439255105,
+      "loss": 1.3788,
+      "step": 1094
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003982492577638711,
+      "loss": 1.2755,
+      "step": 1095
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00039824371240528255,
+      "loss": 1.2935,
+      "step": 1096
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003982381583170297,
+      "loss": 1.3185,
+      "step": 1097
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003982325954993571,
+      "loss": 1.2648,
+      "step": 1098
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00039822702395250976,
+      "loss": 1.3853,
+      "step": 1099
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00039822144367673287,
+      "loss": 1.3822,
+      "step": 1100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00039821585467227224,
+      "loss": 1.3326,
+      "step": 1101
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00039821025693937404,
+      "loss": 1.3997,
+      "step": 1102
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003982046504782847,
+      "loss": 1.3125,
+      "step": 1103
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00039819903528925114,
+      "loss": 1.328,
+      "step": 1104
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00039819341137252055,
+      "loss": 1.3048,
+      "step": 1105
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00039818777872834064,
+      "loss": 1.3706,
+      "step": 1106
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003981821373569595,
+      "loss": 1.2998,
+      "step": 1107
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003981764872586254,
+      "loss": 1.2741,
+      "step": 1108
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00039817082843358724,
+      "loss": 1.3087,
+      "step": 1109
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00039816516088209423,
+      "loss": 1.2524,
+      "step": 1110
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00039815948460439587,
+      "loss": 1.2806,
+      "step": 1111
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003981537996007422,
+      "loss": 1.3681,
+      "step": 1112
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00039814810587138346,
+      "loss": 1.3666,
+      "step": 1113
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00039814240341657037,
+      "loss": 1.2754,
+      "step": 1114
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003981366922365542,
+      "loss": 1.3229,
+      "step": 1115
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003981309723315862,
+      "loss": 1.2467,
+      "step": 1116
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003981252437019185,
+      "loss": 1.3162,
+      "step": 1117
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00039811950634780315,
+      "loss": 1.388,
+      "step": 1118
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00039811376026949294,
+      "loss": 1.4325,
+      "step": 1119
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00039810800546724075,
+      "loss": 1.3801,
+      "step": 1120
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00039810224194130017,
+      "loss": 1.2721,
+      "step": 1121
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003980964696919249,
+      "loss": 1.3255,
+      "step": 1122
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00039809068871936906,
+      "loss": 1.3566,
+      "step": 1123
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00039808489902388734,
+      "loss": 1.2154,
+      "step": 1124
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003980791006057346,
+      "loss": 1.2674,
+      "step": 1125
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003980732934651662,
+      "loss": 1.3126,
+      "step": 1126
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003980674776024379,
+      "loss": 1.3085,
+      "step": 1127
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00039806165301780575,
+      "loss": 1.3219,
+      "step": 1128
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003980558197115263,
+      "loss": 1.3539,
+      "step": 1129
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003980499776838563,
+      "loss": 1.3891,
+      "step": 1130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00039804412693505314,
+      "loss": 1.3666,
+      "step": 1131
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003980382674653743,
+      "loss": 1.3435,
+      "step": 1132
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000398032399275078,
+      "loss": 1.3232,
+      "step": 1133
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00039802652236442244,
+      "loss": 1.3375,
+      "step": 1134
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003980206367336666,
+      "loss": 1.2456,
+      "step": 1135
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003980147423830695,
+      "loss": 1.2301,
+      "step": 1136
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00039800883931289077,
+      "loss": 1.3225,
+      "step": 1137
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003980029275233904,
+      "loss": 1.2793,
+      "step": 1138
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00039799700701482864,
+      "loss": 1.3228,
+      "step": 1139
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003979910777874662,
+      "loss": 1.2811,
+      "step": 1140
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003979851398415641,
+      "loss": 1.3549,
+      "step": 1141
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00039797919317738405,
+      "loss": 1.3869,
+      "step": 1142
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00039797323779518776,
+      "loss": 1.2989,
+      "step": 1143
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00039796727369523737,
+      "loss": 1.2919,
+      "step": 1144
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003979613008777958,
+      "loss": 1.2592,
+      "step": 1145
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003979553193431257,
+      "loss": 1.3329,
+      "step": 1146
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003979493290914908,
+      "loss": 1.24,
+      "step": 1147
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00039794333012315466,
+      "loss": 1.1599,
+      "step": 1148
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00039793732243838156,
+      "loss": 1.2922,
+      "step": 1149
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00039793130603743603,
+      "loss": 1.3441,
+      "step": 1150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00039792528092058295,
+      "loss": 1.316,
+      "step": 1151
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003979192470880876,
+      "loss": 1.2803,
+      "step": 1152
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00039791320454021583,
+      "loss": 1.3114,
+      "step": 1153
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003979071532772337,
+      "loss": 1.2639,
+      "step": 1154
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00039790109329940746,
+      "loss": 1.283,
+      "step": 1155
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00039789502460700426,
+      "loss": 1.2983,
+      "step": 1156
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003978889472002911,
+      "loss": 1.3013,
+      "step": 1157
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00039788286107953576,
+      "loss": 1.4454,
+      "step": 1158
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00039787676624500616,
+      "loss": 1.3512,
+      "step": 1159
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00039787066269697074,
+      "loss": 1.395,
+      "step": 1160
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00039786455043569816,
+      "loss": 1.2489,
+      "step": 1161
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003978584294614577,
+      "loss": 1.2918,
+      "step": 1162
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00039785229977451876,
+      "loss": 1.327,
+      "step": 1163
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00039784616137515145,
+      "loss": 1.2911,
+      "step": 1164
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003978400142636259,
+      "loss": 1.3306,
+      "step": 1165
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00039783385844021295,
+      "loss": 1.3045,
+      "step": 1166
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003978276939051835,
+      "loss": 1.2787,
+      "step": 1167
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003978215206588091,
+      "loss": 1.216,
+      "step": 1168
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00039781533870136165,
+      "loss": 1.2884,
+      "step": 1169
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003978091480331132,
+      "loss": 1.3719,
+      "step": 1170
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00039780294865433656,
+      "loss": 1.3141,
+      "step": 1171
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003977967405653046,
+      "loss": 1.3637,
+      "step": 1172
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003977905237662907,
+      "loss": 1.3342,
+      "step": 1173
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00039778429825756864,
+      "loss": 1.3023,
+      "step": 1174
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003977780640394125,
+      "loss": 1.3636,
+      "step": 1175
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00039777182111209693,
+      "loss": 1.3683,
+      "step": 1176
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00039776556947589673,
+      "loss": 1.2793,
+      "step": 1177
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00039775930913108724,
+      "loss": 1.3045,
+      "step": 1178
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003977530400779441,
+      "loss": 1.3487,
+      "step": 1179
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00039774676231674335,
+      "loss": 1.308,
+      "step": 1180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00039774047584776153,
+      "loss": 1.2691,
+      "step": 1181
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00039773418067127537,
+      "loss": 1.4265,
+      "step": 1182
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003977278767875621,
+      "loss": 1.3527,
+      "step": 1183
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00039772156419689935,
+      "loss": 1.2857,
+      "step": 1184
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003977152428995651,
+      "loss": 1.2575,
+      "step": 1185
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003977089128958376,
+      "loss": 1.3664,
+      "step": 1186
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00039770257418599573,
+      "loss": 1.302,
+      "step": 1187
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00039769622677031846,
+      "loss": 1.2441,
+      "step": 1188
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003976898706490855,
+      "loss": 1.3095,
+      "step": 1189
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00039768350582257655,
+      "loss": 1.3761,
+      "step": 1190
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000397677132291072,
+      "loss": 1.2771,
+      "step": 1191
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00039767075005485246,
+      "loss": 1.3495,
+      "step": 1192
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00039766435911419904,
+      "loss": 1.3175,
+      "step": 1193
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003976579594693931,
+      "loss": 1.2655,
+      "step": 1194
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003976515511207164,
+      "loss": 1.2557,
+      "step": 1195
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003976451340684513,
+      "loss": 1.2845,
+      "step": 1196
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00039763870831288015,
+      "loss": 1.2718,
+      "step": 1197
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003976322738542861,
+      "loss": 1.346,
+      "step": 1198
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00039762583069295244,
+      "loss": 1.2581,
+      "step": 1199
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003976193788291628,
+      "loss": 1.2524,
+      "step": 1200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003976129182632014,
+      "loss": 1.2961,
+      "step": 1201
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00039760644899535267,
+      "loss": 1.3318,
+      "step": 1202
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003975999710259015,
+      "loss": 1.3803,
+      "step": 1203
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00039759348435513316,
+      "loss": 1.303,
+      "step": 1204
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00039758698898333334,
+      "loss": 1.3318,
+      "step": 1205
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003975804849107879,
+      "loss": 1.3552,
+      "step": 1206
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003975739721377835,
+      "loss": 1.2693,
+      "step": 1207
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00039756745066460674,
+      "loss": 1.3099,
+      "step": 1208
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003975609204915448,
+      "loss": 1.2487,
+      "step": 1209
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00039755438161888525,
+      "loss": 1.2898,
+      "step": 1210
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00039754783404691605,
+      "loss": 1.304,
+      "step": 1211
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003975412777759256,
+      "loss": 1.2666,
+      "step": 1212
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00039753471280620247,
+      "loss": 1.3307,
+      "step": 1213
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00039752813913803586,
+      "loss": 1.3246,
+      "step": 1214
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003975215567717151,
+      "loss": 1.3173,
+      "step": 1215
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00039751496570753023,
+      "loss": 1.3083,
+      "step": 1216
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00039750836594577144,
+      "loss": 1.2922,
+      "step": 1217
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003975017574867292,
+      "loss": 1.3399,
+      "step": 1218
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003974951403306946,
+      "loss": 1.2101,
+      "step": 1219
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00039748851447795914,
+      "loss": 1.3053,
+      "step": 1220
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003974818799288144,
+      "loss": 1.2624,
+      "step": 1221
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003974752366835527,
+      "loss": 1.2964,
+      "step": 1222
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00039746858474246647,
+      "loss": 1.2052,
+      "step": 1223
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003974619241058487,
+      "loss": 1.2565,
+      "step": 1224
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00039745525477399255,
+      "loss": 1.3084,
+      "step": 1225
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003974485767471919,
+      "loss": 1.3028,
+      "step": 1226
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00039744189002574067,
+      "loss": 1.3036,
+      "step": 1227
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00039743519460993344,
+      "loss": 1.3052,
+      "step": 1228
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003974284905000649,
+      "loss": 1.4181,
+      "step": 1229
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00039742177769643033,
+      "loss": 1.4095,
+      "step": 1230
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00039741505619932533,
+      "loss": 1.2629,
+      "step": 1231
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003974083260090459,
+      "loss": 1.3682,
+      "step": 1232
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00039740158712588835,
+      "loss": 1.3833,
+      "step": 1233
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003973948395501495,
+      "loss": 1.3414,
+      "step": 1234
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00039738808328212646,
+      "loss": 1.3581,
+      "step": 1235
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003973813183221167,
+      "loss": 1.2936,
+      "step": 1236
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003973745446704181,
+      "loss": 1.1994,
+      "step": 1237
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00039736776232732906,
+      "loss": 1.2926,
+      "step": 1238
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003973609712931481,
+      "loss": 1.2663,
+      "step": 1239
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00039735417156817426,
+      "loss": 1.2979,
+      "step": 1240
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003973473631527071,
+      "loss": 1.2821,
+      "step": 1241
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003973405460470463,
+      "loss": 1.2704,
+      "step": 1242
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003973337202514922,
+      "loss": 1.3082,
+      "step": 1243
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00039732688576634524,
+      "loss": 1.439,
+      "step": 1244
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003973200425919064,
+      "loss": 1.2516,
+      "step": 1245
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00039731319072847706,
+      "loss": 1.2998,
+      "step": 1246
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00039730633017635885,
+      "loss": 1.3528,
+      "step": 1247
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00039729946093585404,
+      "loss": 1.2837,
+      "step": 1248
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00039729258300726496,
+      "loss": 1.2728,
+      "step": 1249
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003972856963908945,
+      "loss": 1.256,
+      "step": 1250
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000397278801087046,
+      "loss": 1.2675,
+      "step": 1251
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003972718970960231,
+      "loss": 1.2869,
+      "step": 1252
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00039726498441812973,
+      "loss": 1.2908,
+      "step": 1253
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003972580630536703,
+      "loss": 1.2546,
+      "step": 1254
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003972511330029496,
+      "loss": 1.2505,
+      "step": 1255
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003972441942662728,
+      "loss": 1.2336,
+      "step": 1256
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003972372468439455,
+      "loss": 1.2174,
+      "step": 1257
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00039723029073627353,
+      "loss": 1.3072,
+      "step": 1258
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003972233259435633,
+      "loss": 1.256,
+      "step": 1259
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00039721635246612154,
+      "loss": 1.3407,
+      "step": 1260
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003972093703042552,
+      "loss": 1.2458,
+      "step": 1261
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003972023794582717,
+      "loss": 1.2598,
+      "step": 1262
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00039719537992847907,
+      "loss": 1.3095,
+      "step": 1263
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003971883717151854,
+      "loss": 1.2126,
+      "step": 1264
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00039718135481869936,
+      "loss": 1.3139,
+      "step": 1265
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003971743292393299,
+      "loss": 1.3288,
+      "step": 1266
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003971672949773864,
+      "loss": 1.2986,
+      "step": 1267
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00039716025203317864,
+      "loss": 1.2816,
+      "step": 1268
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003971532004070167,
+      "loss": 1.402,
+      "step": 1269
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003971461400992111,
+      "loss": 1.4087,
+      "step": 1270
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003971390711100729,
+      "loss": 1.2682,
+      "step": 1271
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003971319934399132,
+      "loss": 1.4067,
+      "step": 1272
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00039712490708904364,
+      "loss": 1.3114,
+      "step": 1273
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00039711781205777644,
+      "loss": 1.3029,
+      "step": 1274
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00039711070834642394,
+      "loss": 1.3414,
+      "step": 1275
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003971035959552989,
+      "loss": 1.2535,
+      "step": 1276
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00039709647488471455,
+      "loss": 1.2772,
+      "step": 1277
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003970893451349846,
+      "loss": 1.3333,
+      "step": 1278
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003970822067064228,
+      "loss": 1.2324,
+      "step": 1279
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00039707505959934357,
+      "loss": 1.3101,
+      "step": 1280
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00039706790381406165,
+      "loss": 1.367,
+      "step": 1281
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00039706073935089216,
+      "loss": 1.3185,
+      "step": 1282
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003970535662101506,
+      "loss": 1.3131,
+      "step": 1283
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003970463843921528,
+      "loss": 1.3216,
+      "step": 1284
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00039703919389721503,
+      "loss": 1.2848,
+      "step": 1285
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00039703199472565384,
+      "loss": 1.3071,
+      "step": 1286
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00039702478687778634,
+      "loss": 1.2708,
+      "step": 1287
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00039701757035392994,
+      "loss": 1.2674,
+      "step": 1288
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00039701034515440234,
+      "loss": 1.3257,
+      "step": 1289
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003970031112795218,
+      "loss": 1.3074,
+      "step": 1290
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00039699586872960673,
+      "loss": 1.322,
+      "step": 1291
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003969886175049762,
+      "loss": 1.3014,
+      "step": 1292
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003969813576059495,
+      "loss": 1.3526,
+      "step": 1293
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00039697408903284617,
+      "loss": 1.3361,
+      "step": 1294
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003969668117859865,
+      "loss": 1.2332,
+      "step": 1295
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003969595258656907,
+      "loss": 1.4157,
+      "step": 1296
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00039695223127227976,
+      "loss": 1.315,
+      "step": 1297
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003969449280060749,
+      "loss": 1.2871,
+      "step": 1298
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003969376160673977,
+      "loss": 1.2647,
+      "step": 1299
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003969302954565701,
+      "loss": 1.3426,
+      "step": 1300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003969229661739146,
+      "loss": 1.2467,
+      "step": 1301
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00039691562821975374,
+      "loss": 1.2796,
+      "step": 1302
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00039690828159441073,
+      "loss": 1.324,
+      "step": 1303
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00039690092629820914,
+      "loss": 1.2288,
+      "step": 1304
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00039689356233147283,
+      "loss": 1.3006,
+      "step": 1305
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003968861896945261,
+      "loss": 1.2275,
+      "step": 1306
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003968788083876935,
+      "loss": 1.3638,
+      "step": 1307
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003968714184113001,
+      "loss": 1.2847,
+      "step": 1308
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003968640197656714,
+      "loss": 1.3548,
+      "step": 1309
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003968566124511331,
+      "loss": 1.3968,
+      "step": 1310
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00039684919646801146,
+      "loss": 1.2932,
+      "step": 1311
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000396841771816633,
+      "loss": 1.2777,
+      "step": 1312
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003968343384973247,
+      "loss": 1.3211,
+      "step": 1313
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00039682689651041384,
+      "loss": 1.204,
+      "step": 1314
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00039681944585622815,
+      "loss": 1.2469,
+      "step": 1315
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003968119865350957,
+      "loss": 1.2564,
+      "step": 1316
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00039680451854734504,
+      "loss": 1.3431,
+      "step": 1317
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003967970418933049,
+      "loss": 1.3861,
+      "step": 1318
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00039678955657330463,
+      "loss": 1.2892,
+      "step": 1319
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003967820625876738,
+      "loss": 1.2786,
+      "step": 1320
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00039677455993674234,
+      "loss": 1.2786,
+      "step": 1321
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003967670486208407,
+      "loss": 1.235,
+      "step": 1322
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00039675952864029967,
+      "loss": 1.2941,
+      "step": 1323
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00039675199999545036,
+      "loss": 1.248,
+      "step": 1324
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00039674446268662424,
+      "loss": 1.3445,
+      "step": 1325
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003967369167141533,
+      "loss": 1.3117,
+      "step": 1326
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003967293620783698,
+      "loss": 1.3471,
+      "step": 1327
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00039672179877960637,
+      "loss": 1.2733,
+      "step": 1328
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003967142268181961,
+      "loss": 1.2319,
+      "step": 1329
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003967066461944724,
+      "loss": 1.2943,
+      "step": 1330
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00039669905690876907,
+      "loss": 1.2469,
+      "step": 1331
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00039669145896142037,
+      "loss": 1.3042,
+      "step": 1332
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00039668385235276085,
+      "loss": 1.2508,
+      "step": 1333
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003966762370831254,
+      "loss": 1.2999,
+      "step": 1334
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003966686131528495,
+      "loss": 1.2818,
+      "step": 1335
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003966609805622687,
+      "loss": 1.2638,
+      "step": 1336
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003966533393117192,
+      "loss": 1.2824,
+      "step": 1337
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003966456894015375,
+      "loss": 1.3089,
+      "step": 1338
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00039663803083206037,
+      "loss": 1.3178,
+      "step": 1339
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00039663036360362516,
+      "loss": 1.3212,
+      "step": 1340
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00039662268771656936,
+      "loss": 1.3065,
+      "step": 1341
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00039661500317123114,
+      "loss": 1.3397,
+      "step": 1342
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00039660730996794876,
+      "loss": 1.3087,
+      "step": 1343
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00039659960810706115,
+      "loss": 1.382,
+      "step": 1344
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003965918975889073,
+      "loss": 1.2848,
+      "step": 1345
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00039658417841382675,
+      "loss": 1.2412,
+      "step": 1346
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003965764505821595,
+      "loss": 1.2897,
+      "step": 1347
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00039656871409424584,
+      "loss": 1.2946,
+      "step": 1348
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00039656096895042636,
+      "loss": 1.2594,
+      "step": 1349
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00039655321515104223,
+      "loss": 1.2966,
+      "step": 1350
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003965454526964348,
+      "loss": 1.3792,
+      "step": 1351
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00039653768158694593,
+      "loss": 1.2922,
+      "step": 1352
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00039652990182291784,
+      "loss": 1.1779,
+      "step": 1353
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00039652211340469303,
+      "loss": 1.283,
+      "step": 1354
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003965143163326145,
+      "loss": 1.2533,
+      "step": 1355
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00039650651060702566,
+      "loss": 1.2366,
+      "step": 1356
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00039649869622827025,
+      "loss": 1.3143,
+      "step": 1357
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00039649087319669225,
+      "loss": 1.3252,
+      "step": 1358
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00039648304151263615,
+      "loss": 1.2803,
+      "step": 1359
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00039647520117644695,
+      "loss": 1.3165,
+      "step": 1360
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00039646735218846975,
+      "loss": 1.3724,
+      "step": 1361
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00039645949454905036,
+      "loss": 1.352,
+      "step": 1362
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00039645162825853464,
+      "loss": 1.3798,
+      "step": 1363
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00039644375331726904,
+      "loss": 1.251,
+      "step": 1364
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00039643586972560034,
+      "loss": 1.348,
+      "step": 1365
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003964279774838756,
+      "loss": 1.2719,
+      "step": 1366
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00039642007659244254,
+      "loss": 1.297,
+      "step": 1367
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003964121670516489,
+      "loss": 1.243,
+      "step": 1368
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00039640424886184313,
+      "loss": 1.3385,
+      "step": 1369
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00039639632202337376,
+      "loss": 1.2961,
+      "step": 1370
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003963883865365899,
+      "loss": 1.3153,
+      "step": 1371
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00039638044240184107,
+      "loss": 1.2229,
+      "step": 1372
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00039637248961947694,
+      "loss": 1.3324,
+      "step": 1373
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00039636452818984783,
+      "loss": 1.2808,
+      "step": 1374
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00039635655811330427,
+      "loss": 1.2512,
+      "step": 1375
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003963485793901972,
+      "loss": 1.2867,
+      "step": 1376
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00039634059202087814,
+      "loss": 1.2908,
+      "step": 1377
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003963325960056985,
+      "loss": 1.2923,
+      "step": 1378
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00039632459134501065,
+      "loss": 1.2404,
+      "step": 1379
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00039631657803916695,
+      "loss": 1.2708,
+      "step": 1380
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003963085560885204,
+      "loss": 1.3148,
+      "step": 1381
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00039630052549342404,
+      "loss": 1.2774,
+      "step": 1382
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003962924862542316,
+      "loss": 1.2164,
+      "step": 1383
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00039628443837129716,
+      "loss": 1.3429,
+      "step": 1384
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00039627638184497495,
+      "loss": 1.2819,
+      "step": 1385
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00039626831667561994,
+      "loss": 1.3226,
+      "step": 1386
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00039626024286358716,
+      "loss": 1.2422,
+      "step": 1387
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00039625216040923205,
+      "loss": 1.265,
+      "step": 1388
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003962440693129107,
+      "loss": 1.3112,
+      "step": 1389
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00039623596957497927,
+      "loss": 1.2684,
+      "step": 1390
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00039622786119579455,
+      "loss": 1.3053,
+      "step": 1391
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00039621974417571347,
+      "loss": 1.2751,
+      "step": 1392
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003962116185150935,
+      "loss": 1.3429,
+      "step": 1393
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00039620348421429255,
+      "loss": 1.2688,
+      "step": 1394
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00039619534127366873,
+      "loss": 1.2337,
+      "step": 1395
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00039618718969358055,
+      "loss": 1.2751,
+      "step": 1396
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003961790294743871,
+      "loss": 1.2888,
+      "step": 1397
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00039617086061644757,
+      "loss": 1.3345,
+      "step": 1398
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003961626831201219,
+      "loss": 1.276,
+      "step": 1399
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00039615449698576994,
+      "loss": 1.2387,
+      "step": 1400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003961463022137523,
+      "loss": 1.1875,
+      "step": 1401
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00039613809880442984,
+      "loss": 1.2657,
+      "step": 1402
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003961298867581637,
+      "loss": 1.2345,
+      "step": 1403
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003961216660753156,
+      "loss": 1.3308,
+      "step": 1404
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00039611343675624745,
+      "loss": 1.2684,
+      "step": 1405
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00039610519880132176,
+      "loss": 1.253,
+      "step": 1406
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003960969522109012,
+      "loss": 1.3269,
+      "step": 1407
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00039608869698534884,
+      "loss": 1.3257,
+      "step": 1408
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003960804331250283,
+      "loss": 1.2671,
+      "step": 1409
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003960721606303035,
+      "loss": 1.2415,
+      "step": 1410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00039606387950153865,
+      "loss": 1.3649,
+      "step": 1411
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003960555897390984,
+      "loss": 1.3408,
+      "step": 1412
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00039604729134334784,
+      "loss": 1.3506,
+      "step": 1413
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003960389843146524,
+      "loss": 1.2272,
+      "step": 1414
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00039603066865337785,
+      "loss": 1.3237,
+      "step": 1415
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00039602234435989037,
+      "loss": 1.3033,
+      "step": 1416
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003960140114345565,
+      "loss": 1.353,
+      "step": 1417
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00039600566987774323,
+      "loss": 1.2889,
+      "step": 1418
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00039599731968981784,
+      "loss": 1.2704,
+      "step": 1419
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00039598896087114806,
+      "loss": 1.2591,
+      "step": 1420
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003959805934221019,
+      "loss": 1.2705,
+      "step": 1421
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00039597221734304793,
+      "loss": 1.2695,
+      "step": 1422
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00039596383263435495,
+      "loss": 1.3227,
+      "step": 1423
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00039595543929639215,
+      "loss": 1.2475,
+      "step": 1424
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003959470373295291,
+      "loss": 1.2258,
+      "step": 1425
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003959386267341359,
+      "loss": 1.2641,
+      "step": 1426
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00039593020751058284,
+      "loss": 1.2838,
+      "step": 1427
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00039592177965924063,
+      "loss": 1.3536,
+      "step": 1428
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003959133431804804,
+      "loss": 1.2151,
+      "step": 1429
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003959048980746737,
+      "loss": 1.2313,
+      "step": 1430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003958964443421924,
+      "loss": 1.3006,
+      "step": 1431
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003958879819834087,
+      "loss": 1.3442,
+      "step": 1432
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003958795109986953,
+      "loss": 1.3016,
+      "step": 1433
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00039587103138842523,
+      "loss": 1.1599,
+      "step": 1434
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00039586254315297184,
+      "loss": 1.2746,
+      "step": 1435
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003958540462927089,
+      "loss": 1.3369,
+      "step": 1436
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00039584554080801065,
+      "loss": 1.2688,
+      "step": 1437
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003958370266992516,
+      "loss": 1.3886,
+      "step": 1438
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003958285039668066,
+      "loss": 1.2618,
+      "step": 1439
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000395819972611051,
+      "loss": 1.2876,
+      "step": 1440
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003958114326323605,
+      "loss": 1.2775,
+      "step": 1441
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003958028840311111,
+      "loss": 1.3292,
+      "step": 1442
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003957943268076793,
+      "loss": 1.2718,
+      "step": 1443
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00039578576096244183,
+      "loss": 1.2482,
+      "step": 1444
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00039577718649577606,
+      "loss": 1.3124,
+      "step": 1445
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00039576860340805937,
+      "loss": 1.2185,
+      "step": 1446
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003957600116996698,
+      "loss": 1.1768,
+      "step": 1447
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00039575141137098574,
+      "loss": 1.3017,
+      "step": 1448
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003957428024223858,
+      "loss": 1.2816,
+      "step": 1449
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003957341848542491,
+      "loss": 1.3255,
+      "step": 1450
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00039572555866695524,
+      "loss": 1.2407,
+      "step": 1451
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003957169238608839,
+      "loss": 1.1868,
+      "step": 1452
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003957082804364154,
+      "loss": 1.2889,
+      "step": 1453
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003956996283939304,
+      "loss": 1.2113,
+      "step": 1454
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00039569096773380973,
+      "loss": 1.274,
+      "step": 1455
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003956822984564349,
+      "loss": 1.2573,
+      "step": 1456
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00039567362056218766,
+      "loss": 1.2716,
+      "step": 1457
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00039566493405145006,
+      "loss": 1.399,
+      "step": 1458
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00039565623892460475,
+      "loss": 1.2689,
+      "step": 1459
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003956475351820345,
+      "loss": 1.255,
+      "step": 1460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00039563882282412255,
+      "loss": 1.3248,
+      "step": 1461
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00039563010185125264,
+      "loss": 1.2222,
+      "step": 1462
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00039562137226380883,
+      "loss": 1.3021,
+      "step": 1463
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003956126340621754,
+      "loss": 1.2503,
+      "step": 1464
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00039560388724673725,
+      "loss": 1.2122,
+      "step": 1465
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003955951318178794,
+      "loss": 1.2888,
+      "step": 1466
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003955863677759876,
+      "loss": 1.3085,
+      "step": 1467
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00039557759512144766,
+      "loss": 1.2099,
+      "step": 1468
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00039556881385464584,
+      "loss": 1.2916,
+      "step": 1469
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003955600239759689,
+      "loss": 1.1946,
+      "step": 1470
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00039555122548580386,
+      "loss": 1.186,
+      "step": 1471
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00039554241838453814,
+      "loss": 1.2838,
+      "step": 1472
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00039553360267255964,
+      "loss": 1.2433,
+      "step": 1473
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003955247783502565,
+      "loss": 1.263,
+      "step": 1474
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003955159454180174,
+      "loss": 1.2181,
+      "step": 1475
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00039550710387623115,
+      "loss": 1.2864,
+      "step": 1476
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003954982537252872,
+      "loss": 1.3784,
+      "step": 1477
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003954893949655752,
+      "loss": 1.3158,
+      "step": 1478
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00039548052759748526,
+      "loss": 1.3378,
+      "step": 1479
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003954716516214078,
+      "loss": 1.272,
+      "step": 1480
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003954627670377338,
+      "loss": 1.2357,
+      "step": 1481
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003954538738468544,
+      "loss": 1.1943,
+      "step": 1482
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003954449720491613,
+      "loss": 1.3347,
+      "step": 1483
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00039543606164504637,
+      "loss": 1.3728,
+      "step": 1484
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00039542714263490205,
+      "loss": 1.263,
+      "step": 1485
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00039541821501912113,
+      "loss": 1.28,
+      "step": 1486
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00039540927879809664,
+      "loss": 1.2761,
+      "step": 1487
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003954003339722221,
+      "loss": 1.2974,
+      "step": 1488
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003953913805418915,
+      "loss": 1.2445,
+      "step": 1489
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00039538241850749895,
+      "loss": 1.2163,
+      "step": 1490
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00039537344786943926,
+      "loss": 1.2876,
+      "step": 1491
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00039536446862810734,
+      "loss": 1.1503,
+      "step": 1492
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00039535548078389855,
+      "loss": 1.3306,
+      "step": 1493
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003953464843372088,
+      "loss": 1.2489,
+      "step": 1494
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00039533747928843424,
+      "loss": 1.3037,
+      "step": 1495
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003953284656379713,
+      "loss": 1.1916,
+      "step": 1496
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00039531944338621694,
+      "loss": 1.2617,
+      "step": 1497
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003953104125335685,
+      "loss": 1.2688,
+      "step": 1498
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003953013730804236,
+      "loss": 1.2696,
+      "step": 1499
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00039529232502718035,
+      "loss": 1.1867,
+      "step": 1500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00039528326837423713,
+      "loss": 1.2619,
+      "step": 1501
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003952742031219927,
+      "loss": 1.2768,
+      "step": 1502
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00039526512927084636,
+      "loss": 1.288,
+      "step": 1503
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003952560468211976,
+      "loss": 1.2776,
+      "step": 1504
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003952469557734465,
+      "loss": 1.2554,
+      "step": 1505
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00039523785612799315,
+      "loss": 1.3168,
+      "step": 1506
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00039522874788523846,
+      "loss": 1.1737,
+      "step": 1507
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003952196310455834,
+      "loss": 1.296,
+      "step": 1508
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003952105056094295,
+      "loss": 1.2749,
+      "step": 1509
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003952013715771785,
+      "loss": 1.3345,
+      "step": 1510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00039519222894923267,
+      "loss": 1.278,
+      "step": 1511
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003951830777259947,
+      "loss": 1.2574,
+      "step": 1512
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00039517391790786746,
+      "loss": 1.2659,
+      "step": 1513
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00039516474949525426,
+      "loss": 1.3001,
+      "step": 1514
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000395155572488559,
+      "loss": 1.2359,
+      "step": 1515
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003951463868881856,
+      "loss": 1.2525,
+      "step": 1516
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00039513719269453865,
+      "loss": 1.1161,
+      "step": 1517
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000395127989908023,
+      "loss": 1.269,
+      "step": 1518
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039511877852904393,
+      "loss": 1.2728,
+      "step": 1519
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039510955855800696,
+      "loss": 1.2355,
+      "step": 1520
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003951003299953182,
+      "loss": 1.2098,
+      "step": 1521
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039509109284138397,
+      "loss": 1.2234,
+      "step": 1522
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039508184709661104,
+      "loss": 1.2976,
+      "step": 1523
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003950725927614066,
+      "loss": 1.3057,
+      "step": 1524
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003950633298361781,
+      "loss": 1.2693,
+      "step": 1525
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003950540583213334,
+      "loss": 1.311,
+      "step": 1526
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003950447782172808,
+      "loss": 1.2888,
+      "step": 1527
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039503548952442904,
+      "loss": 1.2799,
+      "step": 1528
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039502619224318706,
+      "loss": 1.2649,
+      "step": 1529
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039501688637396434,
+      "loss": 1.2969,
+      "step": 1530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039500757191717053,
+      "loss": 1.2877,
+      "step": 1531
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039499824887321585,
+      "loss": 1.3127,
+      "step": 1532
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039498891724251093,
+      "loss": 1.3179,
+      "step": 1533
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003949795770254666,
+      "loss": 1.1808,
+      "step": 1534
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039497022822249415,
+      "loss": 1.2924,
+      "step": 1535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003949608708340053,
+      "loss": 1.2589,
+      "step": 1536
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003949515048604121,
+      "loss": 1.2634,
+      "step": 1537
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039494213030212696,
+      "loss": 1.3217,
+      "step": 1538
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039493274715956265,
+      "loss": 1.2336,
+      "step": 1539
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039492335543313246,
+      "loss": 1.2534,
+      "step": 1540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039491395512324985,
+      "loss": 1.278,
+      "step": 1541
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039490454623032883,
+      "loss": 1.3338,
+      "step": 1542
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039489512875478367,
+      "loss": 1.2864,
+      "step": 1543
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039488570269702913,
+      "loss": 1.3266,
+      "step": 1544
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039487626805748024,
+      "loss": 1.2691,
+      "step": 1545
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003948668248365525,
+      "loss": 1.3101,
+      "step": 1546
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003948573730346617,
+      "loss": 1.2806,
+      "step": 1547
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039484791265222396,
+      "loss": 1.2869,
+      "step": 1548
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003948384436896561,
+      "loss": 1.2721,
+      "step": 1549
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003948289661473749,
+      "loss": 1.2836,
+      "step": 1550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039481948002579773,
+      "loss": 1.2785,
+      "step": 1551
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039480998532534235,
+      "loss": 1.2474,
+      "step": 1552
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003948004820464269,
+      "loss": 1.3343,
+      "step": 1553
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003947909701894698,
+      "loss": 1.3198,
+      "step": 1554
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003947814497548898,
+      "loss": 1.2573,
+      "step": 1555
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039477192074310636,
+      "loss": 1.2293,
+      "step": 1556
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003947623831545389,
+      "loss": 1.2239,
+      "step": 1557
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039475283698960756,
+      "loss": 1.3272,
+      "step": 1558
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003947432822487326,
+      "loss": 1.3272,
+      "step": 1559
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039473371893233476,
+      "loss": 1.2668,
+      "step": 1560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039472414704083523,
+      "loss": 1.2749,
+      "step": 1561
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039471456657465547,
+      "loss": 1.2667,
+      "step": 1562
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039470497753421726,
+      "loss": 1.2424,
+      "step": 1563
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039469537991994303,
+      "loss": 1.231,
+      "step": 1564
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039468577373225533,
+      "loss": 1.2666,
+      "step": 1565
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039467615897157717,
+      "loss": 1.1779,
+      "step": 1566
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039466653563833187,
+      "loss": 1.318,
+      "step": 1567
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003946569037329433,
+      "loss": 1.265,
+      "step": 1568
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003946472632558356,
+      "loss": 1.3255,
+      "step": 1569
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003946376142074332,
+      "loss": 1.3004,
+      "step": 1570
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039462795658816105,
+      "loss": 1.2466,
+      "step": 1571
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039461829039844445,
+      "loss": 1.2092,
+      "step": 1572
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039460861563870897,
+      "loss": 1.3114,
+      "step": 1573
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003945989323093807,
+      "loss": 1.2639,
+      "step": 1574
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039458924041088606,
+      "loss": 1.3118,
+      "step": 1575
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003945795399436518,
+      "loss": 1.1841,
+      "step": 1576
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000394569830908105,
+      "loss": 1.3523,
+      "step": 1577
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003945601133046734,
+      "loss": 1.249,
+      "step": 1578
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003945503871337847,
+      "loss": 1.2139,
+      "step": 1579
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039454065239586734,
+      "loss": 1.3255,
+      "step": 1580
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003945309090913499,
+      "loss": 1.2456,
+      "step": 1581
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003945211572206615,
+      "loss": 1.2729,
+      "step": 1582
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003945113967842315,
+      "loss": 1.3222,
+      "step": 1583
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039450162778248974,
+      "loss": 1.2668,
+      "step": 1584
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003944918502158664,
+      "loss": 1.21,
+      "step": 1585
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039448206408479203,
+      "loss": 1.2579,
+      "step": 1586
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039447226938969755,
+      "loss": 1.2467,
+      "step": 1587
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039446246613101426,
+      "loss": 1.2832,
+      "step": 1588
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039445265430917384,
+      "loss": 1.3381,
+      "step": 1589
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003944428339246084,
+      "loss": 1.1668,
+      "step": 1590
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039443300497775033,
+      "loss": 1.1883,
+      "step": 1591
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039442316746903254,
+      "loss": 1.3049,
+      "step": 1592
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003944133213988881,
+      "loss": 1.2791,
+      "step": 1593
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003944034667677507,
+      "loss": 1.2209,
+      "step": 1594
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039439360357605414,
+      "loss": 1.3309,
+      "step": 1595
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003943837318242329,
+      "loss": 1.2837,
+      "step": 1596
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039437385151272155,
+      "loss": 1.232,
+      "step": 1597
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003943639626419553,
+      "loss": 1.2328,
+      "step": 1598
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039435406521236955,
+      "loss": 1.2212,
+      "step": 1599
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003943441592244001,
+      "loss": 1.232,
+      "step": 1600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039433424467848313,
+      "loss": 1.3442,
+      "step": 1601
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039432432157505536,
+      "loss": 1.301,
+      "step": 1602
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039431438991455365,
+      "loss": 1.296,
+      "step": 1603
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003943044496974154,
+      "loss": 1.2382,
+      "step": 1604
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039429450092407826,
+      "loss": 1.3021,
+      "step": 1605
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003942845435949804,
+      "loss": 1.2432,
+      "step": 1606
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003942745777105602,
+      "loss": 1.2233,
+      "step": 1607
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039426460327125654,
+      "loss": 1.2694,
+      "step": 1608
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039425462027750874,
+      "loss": 1.204,
+      "step": 1609
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003942446287297563,
+      "loss": 1.2903,
+      "step": 1610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039423462862843916,
+      "loss": 1.2987,
+      "step": 1611
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039422461997399785,
+      "loss": 1.2865,
+      "step": 1612
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003942146027668729,
+      "loss": 1.2292,
+      "step": 1613
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003942045770075056,
+      "loss": 1.2797,
+      "step": 1614
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003941945426963372,
+      "loss": 1.2607,
+      "step": 1615
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003941844998338098,
+      "loss": 1.2778,
+      "step": 1616
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003941744484203656,
+      "loss": 1.2263,
+      "step": 1617
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000394164388456447,
+      "loss": 1.212,
+      "step": 1618
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003941543199424972,
+      "loss": 1.2728,
+      "step": 1619
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039414424287895954,
+      "loss": 1.2668,
+      "step": 1620
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003941341572662777,
+      "loss": 1.2974,
+      "step": 1621
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00039412406310489597,
+      "loss": 1.307,
+      "step": 1622
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00039411396039525857,
+      "loss": 1.2115,
+      "step": 1623
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00039410384913781055,
+      "loss": 1.2105,
+      "step": 1624
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003940937293329971,
+      "loss": 1.2404,
+      "step": 1625
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00039408360098126393,
+      "loss": 1.2244,
+      "step": 1626
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003940734640830569,
+      "loss": 1.2302,
+      "step": 1627
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003940633186388226,
+      "loss": 1.2415,
+      "step": 1628
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00039405316464900754,
+      "loss": 1.2174,
+      "step": 1629
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000394043002114059,
+      "loss": 1.223,
+      "step": 1630
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003940328310344245,
+      "loss": 1.2274,
+      "step": 1631
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003940226514105518,
+      "loss": 1.2572,
+      "step": 1632
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00039401246324288926,
+      "loss": 1.2441,
+      "step": 1633
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003940022665318855,
+      "loss": 1.2292,
+      "step": 1634
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00039399206127798954,
+      "loss": 1.2368,
+      "step": 1635
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00039398184748165073,
+      "loss": 1.3823,
+      "step": 1636
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003939716251433189,
+      "loss": 1.325,
+      "step": 1637
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003939613942634441,
+      "loss": 1.2262,
+      "step": 1638
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003939511548424769,
+      "loss": 1.2395,
+      "step": 1639
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00039394090688086827,
+      "loss": 1.2227,
+      "step": 1640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003939306503790693,
+      "loss": 1.2793,
+      "step": 1641
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00039392038533753176,
+      "loss": 1.3909,
+      "step": 1642
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00039391011175670765,
+      "loss": 1.3009,
+      "step": 1643
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003938998296370494,
+      "loss": 1.1847,
+      "step": 1644
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00039388953897900967,
+      "loss": 1.2098,
+      "step": 1645
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003938792397830418,
+      "loss": 1.2255,
+      "step": 1646
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003938689320495991,
+      "loss": 1.2925,
+      "step": 1647
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003938586157791356,
+      "loss": 1.2816,
+      "step": 1648
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00039384829097210554,
+      "loss": 1.2961,
+      "step": 1649
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003938379576289636,
+      "loss": 1.2722,
+      "step": 1650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00039382761575016475,
+      "loss": 1.2263,
+      "step": 1651
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00039381726533616447,
+      "loss": 1.2621,
+      "step": 1652
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00039380690638741853,
+      "loss": 1.2914,
+      "step": 1653
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00039379653890438305,
+      "loss": 1.2691,
+      "step": 1654
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003937861628875146,
+      "loss": 1.2424,
+      "step": 1655
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00039377577833727,
+      "loss": 1.2563,
+      "step": 1656
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003937653852541066,
+      "loss": 1.2668,
+      "step": 1657
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003937549836384821,
+      "loss": 1.3012,
+      "step": 1658
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00039374457349085443,
+      "loss": 1.2499,
+      "step": 1659
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00039373415481168205,
+      "loss": 1.2672,
+      "step": 1660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00039372372760142384,
+      "loss": 1.3418,
+      "step": 1661
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00039371329186053885,
+      "loss": 1.2522,
+      "step": 1662
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00039370284758948667,
+      "loss": 1.3445,
+      "step": 1663
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003936923947887271,
+      "loss": 1.2397,
+      "step": 1664
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003936819334587206,
+      "loss": 1.2075,
+      "step": 1665
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00039367146359992777,
+      "loss": 1.2506,
+      "step": 1666
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003936609852128096,
+      "loss": 1.3089,
+      "step": 1667
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003936504982978276,
+      "loss": 1.2633,
+      "step": 1668
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003936400028554434,
+      "loss": 1.3711,
+      "step": 1669
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003936294988861193,
+      "loss": 1.2684,
+      "step": 1670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00039361898639031793,
+      "loss": 1.279,
+      "step": 1671
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00039360846536850197,
+      "loss": 1.3075,
+      "step": 1672
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00039359793582113486,
+      "loss": 1.3028,
+      "step": 1673
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003935873977486803,
+      "loss": 1.2011,
+      "step": 1674
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003935768511516022,
+      "loss": 1.2539,
+      "step": 1675
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003935662960303651,
+      "loss": 1.2782,
+      "step": 1676
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00039355573238543373,
+      "loss": 1.2568,
+      "step": 1677
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003935451602172733,
+      "loss": 1.2188,
+      "step": 1678
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00039353457952634934,
+      "loss": 1.1924,
+      "step": 1679
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003935239903131278,
+      "loss": 1.2999,
+      "step": 1680
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003935133925780749,
+      "loss": 1.2408,
+      "step": 1681
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003935027863216573,
+      "loss": 1.2927,
+      "step": 1682
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003934921715443422,
+      "loss": 1.2968,
+      "step": 1683
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003934815482465969,
+      "loss": 1.299,
+      "step": 1684
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003934709164288892,
+      "loss": 1.3353,
+      "step": 1685
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00039346027609168726,
+      "loss": 1.1694,
+      "step": 1686
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00039344962723545973,
+      "loss": 1.3223,
+      "step": 1687
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003934389698606754,
+      "loss": 1.2782,
+      "step": 1688
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003934283039678037,
+      "loss": 1.3623,
+      "step": 1689
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00039341762955731423,
+      "loss": 1.261,
+      "step": 1690
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00039340694662967705,
+      "loss": 1.3961,
+      "step": 1691
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00039339625518536255,
+      "loss": 1.1817,
+      "step": 1692
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00039338555522484153,
+      "loss": 1.2254,
+      "step": 1693
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00039337484674858514,
+      "loss": 1.3389,
+      "step": 1694
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00039336412975706506,
+      "loss": 1.2667,
+      "step": 1695
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003933534042507531,
+      "loss": 1.2394,
+      "step": 1696
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00039334267023012165,
+      "loss": 1.2374,
+      "step": 1697
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003933319276956432,
+      "loss": 1.2972,
+      "step": 1698
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000393321176647791,
+      "loss": 1.2749,
+      "step": 1699
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003933104170870383,
+      "loss": 1.2544,
+      "step": 1700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00039329964901385904,
+      "loss": 1.3142,
+      "step": 1701
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003932888724287274,
+      "loss": 1.2714,
+      "step": 1702
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00039327808733211777,
+      "loss": 1.2866,
+      "step": 1703
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003932672937245052,
+      "loss": 1.2511,
+      "step": 1704
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003932564916063649,
+      "loss": 1.2568,
+      "step": 1705
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00039324568097817274,
+      "loss": 1.2897,
+      "step": 1706
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003932348618404045,
+      "loss": 1.2803,
+      "step": 1707
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00039322403419353674,
+      "loss": 1.2801,
+      "step": 1708
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00039321319803804625,
+      "loss": 1.264,
+      "step": 1709
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003932023533744102,
+      "loss": 1.2852,
+      "step": 1710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003931915002031061,
+      "loss": 1.2805,
+      "step": 1711
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00039318063852461185,
+      "loss": 1.258,
+      "step": 1712
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00039316976833940584,
+      "loss": 1.3046,
+      "step": 1713
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00039315888964796663,
+      "loss": 1.3097,
+      "step": 1714
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003931480024507734,
+      "loss": 1.3064,
+      "step": 1715
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00039313710674830545,
+      "loss": 1.2463,
+      "step": 1716
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003931262025410426,
+      "loss": 1.2424,
+      "step": 1717
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000393115289829465,
+      "loss": 1.3641,
+      "step": 1718
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00039310436861405326,
+      "loss": 1.2142,
+      "step": 1719
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00039309343889528823,
+      "loss": 1.2853,
+      "step": 1720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003930825006736512,
+      "loss": 1.273,
+      "step": 1721
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00039307155394962385,
+      "loss": 1.2888,
+      "step": 1722
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00039306059872368823,
+      "loss": 1.3749,
+      "step": 1723
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003930496349963268,
+      "loss": 1.1969,
+      "step": 1724
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003930386627680223,
+      "loss": 1.2595,
+      "step": 1725
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00039302768203925786,
+      "loss": 1.2671,
+      "step": 1726
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00039301669281051706,
+      "loss": 1.2259,
+      "step": 1727
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003930056950822838,
+      "loss": 1.3405,
+      "step": 1728
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003929946888550423,
+      "loss": 1.268,
+      "step": 1729
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00039298367412927737,
+      "loss": 1.22,
+      "step": 1730
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003929726509054739,
+      "loss": 1.2737,
+      "step": 1731
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003929616191841174,
+      "loss": 1.1924,
+      "step": 1732
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003929505789656935,
+      "loss": 1.3146,
+      "step": 1733
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003929395302506886,
+      "loss": 1.2072,
+      "step": 1734
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00039292847303958896,
+      "loss": 1.2685,
+      "step": 1735
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003929174073328817,
+      "loss": 1.2891,
+      "step": 1736
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000392906333131054,
+      "loss": 1.2664,
+      "step": 1737
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00039289525043459346,
+      "loss": 1.204,
+      "step": 1738
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00039288415924398825,
+      "loss": 1.2091,
+      "step": 1739
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003928730595597267,
+      "loss": 1.1569,
+      "step": 1740
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003928619513822976,
+      "loss": 1.4137,
+      "step": 1741
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00039285083471219003,
+      "loss": 1.2551,
+      "step": 1742
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00039283970954989353,
+      "loss": 1.1773,
+      "step": 1743
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00039282857589589804,
+      "loss": 1.1932,
+      "step": 1744
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00039281743375069384,
+      "loss": 1.2399,
+      "step": 1745
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00039280628311477153,
+      "loss": 1.2639,
+      "step": 1746
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003927951239886222,
+      "loss": 1.3453,
+      "step": 1747
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00039278395637273715,
+      "loss": 1.2661,
+      "step": 1748
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00039277278026760823,
+      "loss": 1.2266,
+      "step": 1749
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00039276159567372746,
+      "loss": 1.2624,
+      "step": 1750
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003927504025915875,
+      "loss": 1.2511,
+      "step": 1751
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003927392010216812,
+      "loss": 1.1868,
+      "step": 1752
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003927279909645017,
+      "loss": 1.3107,
+      "step": 1753
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003927167724205427,
+      "loss": 1.2421,
+      "step": 1754
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003927055453902984,
+      "loss": 1.1628,
+      "step": 1755
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003926943098742629,
+      "loss": 1.2647,
+      "step": 1756
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003926830658729311,
+      "loss": 1.2386,
+      "step": 1757
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00039267181338679806,
+      "loss": 1.2788,
+      "step": 1758
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00039266055241635937,
+      "loss": 1.267,
+      "step": 1759
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00039264928296211084,
+      "loss": 1.3396,
+      "step": 1760
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003926380050245488,
+      "loss": 1.3186,
+      "step": 1761
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00039262671860416975,
+      "loss": 1.2424,
+      "step": 1762
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003926154237014708,
+      "loss": 1.2639,
+      "step": 1763
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003926041203169493,
+      "loss": 1.2224,
+      "step": 1764
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00039259280845110287,
+      "loss": 1.2662,
+      "step": 1765
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003925814881044298,
+      "loss": 1.3011,
+      "step": 1766
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003925701592774285,
+      "loss": 1.2412,
+      "step": 1767
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00039255882197059786,
+      "loss": 1.3109,
+      "step": 1768
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00039254747618443713,
+      "loss": 1.2003,
+      "step": 1769
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003925361219194458,
+      "loss": 1.2841,
+      "step": 1770
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00039252475917612404,
+      "loss": 1.2731,
+      "step": 1771
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003925133879549721,
+      "loss": 1.1999,
+      "step": 1772
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003925020082564907,
+      "loss": 1.2767,
+      "step": 1773
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000392490620081181,
+      "loss": 1.2358,
+      "step": 1774
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00039247922342954445,
+      "loss": 1.1835,
+      "step": 1775
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00039246781830208296,
+      "loss": 1.2384,
+      "step": 1776
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00039245640469929866,
+      "loss": 1.2105,
+      "step": 1777
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003924449826216942,
+      "loss": 1.2754,
+      "step": 1778
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003924335520697725,
+      "loss": 1.2593,
+      "step": 1779
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000392422113044037,
+      "loss": 1.324,
+      "step": 1780
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00039241066554499134,
+      "loss": 1.3072,
+      "step": 1781
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00039239920957313964,
+      "loss": 1.2867,
+      "step": 1782
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00039238774512898635,
+      "loss": 1.2152,
+      "step": 1783
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00039237627221303634,
+      "loss": 1.2639,
+      "step": 1784
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00039236479082579474,
+      "loss": 1.1948,
+      "step": 1785
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00039235330096776727,
+      "loss": 1.3678,
+      "step": 1786
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00039234180263945977,
+      "loss": 1.2995,
+      "step": 1787
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00039233029584137855,
+      "loss": 1.2779,
+      "step": 1788
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00039231878057403044,
+      "loss": 1.2459,
+      "step": 1789
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003923072568379224,
+      "loss": 1.2478,
+      "step": 1790
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00039229572463356195,
+      "loss": 1.2353,
+      "step": 1791
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003922841839614568,
+      "loss": 1.1523,
+      "step": 1792
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003922726348221153,
+      "loss": 1.1974,
+      "step": 1793
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003922610772160459,
+      "loss": 1.2989,
+      "step": 1794
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00039224951114375764,
+      "loss": 1.2153,
+      "step": 1795
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003922379366057598,
+      "loss": 1.3042,
+      "step": 1796
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003922263536025619,
+      "loss": 1.2662,
+      "step": 1797
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003922147621346742,
+      "loss": 1.2512,
+      "step": 1798
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00039220316220260707,
+      "loss": 1.2878,
+      "step": 1799
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003921915538068713,
+      "loss": 1.2739,
+      "step": 1800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003921799369479781,
+      "loss": 1.1931,
+      "step": 1801
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000392168311626439,
+      "loss": 1.2176,
+      "step": 1802
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00039215667784276586,
+      "loss": 1.3518,
+      "step": 1803
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00039214503559747104,
+      "loss": 1.3,
+      "step": 1804
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003921333848910672,
+      "loss": 1.2831,
+      "step": 1805
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00039212172572406737,
+      "loss": 1.1678,
+      "step": 1806
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00039211005809698494,
+      "loss": 1.265,
+      "step": 1807
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003920983820103338,
+      "loss": 1.2,
+      "step": 1808
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000392086697464628,
+      "loss": 1.2538,
+      "step": 1809
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000392075004460382,
+      "loss": 1.2059,
+      "step": 1810
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00039206330299811087,
+      "loss": 1.2352,
+      "step": 1811
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00039205159307832975,
+      "loss": 1.1384,
+      "step": 1812
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003920398747015544,
+      "loss": 1.245,
+      "step": 1813
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003920281478683008,
+      "loss": 1.284,
+      "step": 1814
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00039201641257908534,
+      "loss": 1.194,
+      "step": 1815
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003920046688344247,
+      "loss": 1.2274,
+      "step": 1816
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003919929166348361,
+      "loss": 1.3486,
+      "step": 1817
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003919811559808371,
+      "loss": 1.2501,
+      "step": 1818
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00039196938687294545,
+      "loss": 1.1865,
+      "step": 1819
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00039195760931167947,
+      "loss": 1.3516,
+      "step": 1820
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003919458232975577,
+      "loss": 1.1538,
+      "step": 1821
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003919340288310994,
+      "loss": 1.284,
+      "step": 1822
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003919222259128236,
+      "loss": 1.2339,
+      "step": 1823
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003919104145432503,
+      "loss": 1.2329,
+      "step": 1824
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003918985947228995,
+      "loss": 1.1964,
+      "step": 1825
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00039188676645229164,
+      "loss": 1.2415,
+      "step": 1826
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00039187492973194774,
+      "loss": 1.2046,
+      "step": 1827
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003918630845623889,
+      "loss": 1.2401,
+      "step": 1828
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00039185123094413667,
+      "loss": 1.3716,
+      "step": 1829
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00039183936887771315,
+      "loss": 1.2068,
+      "step": 1830
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00039182749836364064,
+      "loss": 1.2467,
+      "step": 1831
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003918156194024419,
+      "loss": 1.243,
+      "step": 1832
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00039180373199463995,
+      "loss": 1.2603,
+      "step": 1833
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00039179183614075825,
+      "loss": 1.3096,
+      "step": 1834
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00039177993184132066,
+      "loss": 1.3057,
+      "step": 1835
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00039176801909685145,
+      "loss": 1.2145,
+      "step": 1836
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003917560979078751,
+      "loss": 1.2905,
+      "step": 1837
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003917441682749166,
+      "loss": 1.189,
+      "step": 1838
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00039173223019850126,
+      "loss": 1.2323,
+      "step": 1839
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00039172028367915484,
+      "loss": 1.3121,
+      "step": 1840
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003917083287174033,
+      "loss": 1.2529,
+      "step": 1841
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003916963653137732,
+      "loss": 1.2672,
+      "step": 1842
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00039168439346879113,
+      "loss": 1.2503,
+      "step": 1843
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00039167241318298454,
+      "loss": 1.2858,
+      "step": 1844
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003916604244568808,
+      "loss": 1.3305,
+      "step": 1845
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00039164842729100794,
+      "loss": 1.183,
+      "step": 1846
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00039163642168589416,
+      "loss": 1.1738,
+      "step": 1847
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00039162440764206813,
+      "loss": 1.2408,
+      "step": 1848
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000391612385160059,
+      "loss": 1.3491,
+      "step": 1849
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003916003542403961,
+      "loss": 1.29,
+      "step": 1850
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003915883148836092,
+      "loss": 1.2362,
+      "step": 1851
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003915762670902285,
+      "loss": 1.1712,
+      "step": 1852
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003915642108607845,
+      "loss": 1.3008,
+      "step": 1853
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003915521461958081,
+      "loss": 1.1898,
+      "step": 1854
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00039154007309583056,
+      "loss": 1.2407,
+      "step": 1855
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00039152799156138353,
+      "loss": 1.3158,
+      "step": 1856
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000391515901592999,
+      "loss": 1.2617,
+      "step": 1857
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00039150380319120937,
+      "loss": 1.1724,
+      "step": 1858
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00039149169635654745,
+      "loss": 1.3133,
+      "step": 1859
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003914795810895462,
+      "loss": 1.1838,
+      "step": 1860
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00039146745739073927,
+      "loss": 1.151,
+      "step": 1861
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003914553252606605,
+      "loss": 1.1699,
+      "step": 1862
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00039144318469984407,
+      "loss": 1.2829,
+      "step": 1863
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00039143103570882465,
+      "loss": 1.2443,
+      "step": 1864
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003914188782881372,
+      "loss": 1.2425,
+      "step": 1865
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00039140671243831707,
+      "loss": 1.1779,
+      "step": 1866
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003913945381599,
+      "loss": 1.2317,
+      "step": 1867
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000391382355453422,
+      "loss": 1.3245,
+      "step": 1868
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00039137016431941966,
+      "loss": 1.2732,
+      "step": 1869
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003913579647584297,
+      "loss": 1.2608,
+      "step": 1870
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00039134575677098944,
+      "loss": 1.1928,
+      "step": 1871
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00039133354035763635,
+      "loss": 1.183,
+      "step": 1872
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003913213155189085,
+      "loss": 1.21,
+      "step": 1873
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003913090822553441,
+      "loss": 1.2499,
+      "step": 1874
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00039129684056748195,
+      "loss": 1.2969,
+      "step": 1875
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000391284590455861,
+      "loss": 1.3082,
+      "step": 1876
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00039127233192102074,
+      "loss": 1.2193,
+      "step": 1877
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00039126006496350104,
+      "loss": 1.2182,
+      "step": 1878
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00039124778958384194,
+      "loss": 1.2027,
+      "step": 1879
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00039123550578258403,
+      "loss": 1.2923,
+      "step": 1880
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00039122321356026827,
+      "loss": 1.1736,
+      "step": 1881
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00039121091291743595,
+      "loss": 1.234,
+      "step": 1882
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00039119860385462867,
+      "loss": 1.1428,
+      "step": 1883
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003911862863723885,
+      "loss": 1.2106,
+      "step": 1884
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003911739604712579,
+      "loss": 1.2928,
+      "step": 1885
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003911616261517795,
+      "loss": 1.255,
+      "step": 1886
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00039114928341449655,
+      "loss": 1.1985,
+      "step": 1887
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00039113693225995255,
+      "loss": 1.2618,
+      "step": 1888
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00039112457268869136,
+      "loss": 1.3018,
+      "step": 1889
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00039111220470125724,
+      "loss": 1.2128,
+      "step": 1890
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003910998282981948,
+      "loss": 1.1934,
+      "step": 1891
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00039108744348004905,
+      "loss": 1.1772,
+      "step": 1892
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00039107505024736536,
+      "loss": 1.2918,
+      "step": 1893
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003910626486006895,
+      "loss": 1.3098,
+      "step": 1894
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003910502385405675,
+      "loss": 1.239,
+      "step": 1895
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00039103782006754583,
+      "loss": 1.1957,
+      "step": 1896
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003910253931821714,
+      "loss": 1.1725,
+      "step": 1897
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003910129578849915,
+      "loss": 1.231,
+      "step": 1898
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00039100051417655357,
+      "loss": 1.186,
+      "step": 1899
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003909880620574056,
+      "loss": 1.2376,
+      "step": 1900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00039097560152809595,
+      "loss": 1.3314,
+      "step": 1901
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00039096313258917334,
+      "loss": 1.2892,
+      "step": 1902
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00039095065524118687,
+      "loss": 1.2207,
+      "step": 1903
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003909381694846858,
+      "loss": 1.198,
+      "step": 1904
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00039092567532022024,
+      "loss": 1.2709,
+      "step": 1905
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00039091317274834005,
+      "loss": 1.2571,
+      "step": 1906
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00039090066176959595,
+      "loss": 1.2002,
+      "step": 1907
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00039088814238453894,
+      "loss": 1.1791,
+      "step": 1908
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00039087561459372005,
+      "loss": 1.2162,
+      "step": 1909
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00039086307839769125,
+      "loss": 1.2819,
+      "step": 1910
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003908505337970043,
+      "loss": 1.2726,
+      "step": 1911
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003908379807922118,
+      "loss": 1.2635,
+      "step": 1912
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003908254193838665,
+      "loss": 1.2643,
+      "step": 1913
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003908128495725214,
+      "loss": 1.3285,
+      "step": 1914
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003908002713587301,
+      "loss": 1.2166,
+      "step": 1915
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003907876847430465,
+      "loss": 1.2292,
+      "step": 1916
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00039077508972602484,
+      "loss": 1.191,
+      "step": 1917
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00039076248630821983,
+      "loss": 1.2372,
+      "step": 1918
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00039074987449018623,
+      "loss": 1.2789,
+      "step": 1919
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003907372542724796,
+      "loss": 1.2021,
+      "step": 1920
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003907246256556556,
+      "loss": 1.2658,
+      "step": 1921
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003907119886402703,
+      "loss": 1.2154,
+      "step": 1922
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003906993432268803,
+      "loss": 1.2598,
+      "step": 1923
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003906866894160423,
+      "loss": 1.248,
+      "step": 1924
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003906740272083136,
+      "loss": 1.2522,
+      "step": 1925
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003906613566042517,
+      "loss": 1.2473,
+      "step": 1926
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00039064867760441464,
+      "loss": 1.1572,
+      "step": 1927
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00039063599020936065,
+      "loss": 1.2902,
+      "step": 1928
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003906232944196485,
+      "loss": 1.2748,
+      "step": 1929
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00039061059023583726,
+      "loss": 1.1917,
+      "step": 1930
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003905978776584862,
+      "loss": 1.2278,
+      "step": 1931
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003905851566881553,
+      "loss": 1.1481,
+      "step": 1932
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00039057242732540466,
+      "loss": 1.2353,
+      "step": 1933
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003905596895707948,
+      "loss": 1.3572,
+      "step": 1934
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00039054694342488663,
+      "loss": 1.2515,
+      "step": 1935
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00039053418888824147,
+      "loss": 1.2015,
+      "step": 1936
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00039052142596142093,
+      "loss": 1.21,
+      "step": 1937
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000390508654644987,
+      "loss": 1.2964,
+      "step": 1938
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003904958749395022,
+      "loss": 1.2089,
+      "step": 1939
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003904830868455291,
+      "loss": 1.1978,
+      "step": 1940
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00039047029036363095,
+      "loss": 1.2361,
+      "step": 1941
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003904574854943712,
+      "loss": 1.1573,
+      "step": 1942
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003904446722383136,
+      "loss": 1.2236,
+      "step": 1943
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003904318505960226,
+      "loss": 1.237,
+      "step": 1944
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00039041902056806275,
+      "loss": 1.2281,
+      "step": 1945
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003904061821549988,
+      "loss": 1.2282,
+      "step": 1946
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00039039333535739637,
+      "loss": 1.2593,
+      "step": 1947
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00039038048017582105,
+      "loss": 1.2207,
+      "step": 1948
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003903676166108389,
+      "loss": 1.1882,
+      "step": 1949
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003903547446630164,
+      "loss": 1.2401,
+      "step": 1950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003903418643329204,
+      "loss": 1.2861,
+      "step": 1951
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003903289756211179,
+      "loss": 1.2069,
+      "step": 1952
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003903160785281768,
+      "loss": 1.25,
+      "step": 1953
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003903031730546646,
+      "loss": 1.176,
+      "step": 1954
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00039029025920114997,
+      "loss": 1.2235,
+      "step": 1955
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003902773369682014,
+      "loss": 1.3188,
+      "step": 1956
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00039026440635638786,
+      "loss": 1.2124,
+      "step": 1957
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003902514673662789,
+      "loss": 1.2628,
+      "step": 1958
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00039023851999844415,
+      "loss": 1.2746,
+      "step": 1959
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00039022556425345387,
+      "loss": 1.2203,
+      "step": 1960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00039021260013187846,
+      "loss": 1.3067,
+      "step": 1961
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00039019962763428887,
+      "loss": 1.2156,
+      "step": 1962
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003901866467612563,
+      "loss": 1.1801,
+      "step": 1963
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003901736575133524,
+      "loss": 1.2429,
+      "step": 1964
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003901606598911491,
+      "loss": 1.2496,
+      "step": 1965
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00039014765389521887,
+      "loss": 1.2211,
+      "step": 1966
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00039013463952613424,
+      "loss": 1.2057,
+      "step": 1967
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00039012161678446844,
+      "loss": 1.2161,
+      "step": 1968
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00039010858567079486,
+      "loss": 1.1626,
+      "step": 1969
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00039009554618568736,
+      "loss": 1.288,
+      "step": 1970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003900824983297202,
+      "loss": 1.1762,
+      "step": 1971
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00039006944210346776,
+      "loss": 1.2602,
+      "step": 1972
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00039005637750750514,
+      "loss": 1.2205,
+      "step": 1973
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00039004330454240753,
+      "loss": 1.2285,
+      "step": 1974
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00039003022320875067,
+      "loss": 1.2352,
+      "step": 1975
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003900171335071106,
+      "loss": 1.2363,
+      "step": 1976
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00039000403543806367,
+      "loss": 1.3168,
+      "step": 1977
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003899909290021867,
+      "loss": 1.1949,
+      "step": 1978
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003899778142000568,
+      "loss": 1.2703,
+      "step": 1979
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003899646910322515,
+      "loss": 1.2623,
+      "step": 1980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003899515594993487,
+      "loss": 1.2039,
+      "step": 1981
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00038993841960192655,
+      "loss": 1.2988,
+      "step": 1982
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00038992527134056374,
+      "loss": 1.1285,
+      "step": 1983
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00038991211471583933,
+      "loss": 1.2614,
+      "step": 1984
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003898989497283325,
+      "loss": 1.1916,
+      "step": 1985
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00038988577637862315,
+      "loss": 1.2902,
+      "step": 1986
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003898725946672912,
+      "loss": 1.246,
+      "step": 1987
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00038985940459491724,
+      "loss": 1.29,
+      "step": 1988
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000389846206162082,
+      "loss": 1.2308,
+      "step": 1989
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003898329993693667,
+      "loss": 1.2079,
+      "step": 1990
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00038981978421735296,
+      "loss": 1.2546,
+      "step": 1991
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003898065607066227,
+      "loss": 1.3812,
+      "step": 1992
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00038979332883775817,
+      "loss": 1.2369,
+      "step": 1993
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00038978008861134206,
+      "loss": 1.3055,
+      "step": 1994
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00038976684002795735,
+      "loss": 1.1768,
+      "step": 1995
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003897535830881875,
+      "loss": 1.2694,
+      "step": 1996
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00038974031779261624,
+      "loss": 1.2186,
+      "step": 1997
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00038972704414182776,
+      "loss": 1.2387,
+      "step": 1998
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00038971376213640653,
+      "loss": 1.2336,
+      "step": 1999
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00038970047177693743,
+      "loss": 1.265,
+      "step": 2000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003896871730640057,
+      "loss": 1.2591,
+      "step": 2001
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003896738659981969,
+      "loss": 1.2261,
+      "step": 2002
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003896605505800971,
+      "loss": 1.1788,
+      "step": 2003
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003896472268102926,
+      "loss": 1.1736,
+      "step": 2004
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00038963389468937013,
+      "loss": 1.1924,
+      "step": 2005
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00038962055421791676,
+      "loss": 1.264,
+      "step": 2006
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003896072053965199,
+      "loss": 1.2884,
+      "step": 2007
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00038959384822576744,
+      "loss": 1.2294,
+      "step": 2008
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003895804827062475,
+      "loss": 1.1929,
+      "step": 2009
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003895671088385487,
+      "loss": 1.2449,
+      "step": 2010
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003895537266232598,
+      "loss": 1.2325,
+      "step": 2011
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00038954033606097024,
+      "loss": 1.3529,
+      "step": 2012
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003895269371522697,
+      "loss": 1.315,
+      "step": 2013
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00038951352989774805,
+      "loss": 1.2202,
+      "step": 2014
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00038950011429799574,
+      "loss": 1.2752,
+      "step": 2015
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00038948669035360363,
+      "loss": 1.2995,
+      "step": 2016
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003894732580651627,
+      "loss": 1.2194,
+      "step": 2017
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00038945981743326456,
+      "loss": 1.2073,
+      "step": 2018
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00038944636845850096,
+      "loss": 1.2879,
+      "step": 2019
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00038943291114146414,
+      "loss": 1.2109,
+      "step": 2020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003894194454827468,
+      "loss": 1.2821,
+      "step": 2021
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00038940597148294173,
+      "loss": 1.2202,
+      "step": 2022
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003893924891426425,
+      "loss": 1.2245,
+      "step": 2023
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003893789984624425,
+      "loss": 1.2432,
+      "step": 2024
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00038936549944293603,
+      "loss": 1.2327,
+      "step": 2025
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00038935199208471746,
+      "loss": 1.1643,
+      "step": 2026
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003893384763883816,
+      "loss": 1.1891,
+      "step": 2027
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003893249523545235,
+      "loss": 1.2251,
+      "step": 2028
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00038931141998373885,
+      "loss": 1.1789,
+      "step": 2029
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00038929787927662334,
+      "loss": 1.238,
+      "step": 2030
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00038928433023377354,
+      "loss": 1.2807,
+      "step": 2031
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003892707728557858,
+      "loss": 1.1973,
+      "step": 2032
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00038925720714325725,
+      "loss": 1.1991,
+      "step": 2033
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00038924363309678523,
+      "loss": 1.2565,
+      "step": 2034
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003892300507169675,
+      "loss": 1.2087,
+      "step": 2035
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003892164600044021,
+      "loss": 1.2234,
+      "step": 2036
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00038920286095968755,
+      "loss": 1.3205,
+      "step": 2037
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00038918925358342267,
+      "loss": 1.2679,
+      "step": 2038
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00038917563787620673,
+      "loss": 1.2204,
+      "step": 2039
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003891620138386391,
+      "loss": 1.3253,
+      "step": 2040
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00038914838147131994,
+      "loss": 1.1958,
+      "step": 2041
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003891347407748494,
+      "loss": 1.2368,
+      "step": 2042
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003891210917498283,
+      "loss": 1.1943,
+      "step": 2043
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00038910743439685743,
+      "loss": 1.1806,
+      "step": 2044
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003890937687165384,
+      "loss": 1.2377,
+      "step": 2045
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000389080094709473,
+      "loss": 1.2508,
+      "step": 2046
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003890664123762632,
+      "loss": 1.1669,
+      "step": 2047
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003890527217175116,
+      "loss": 1.2903,
+      "step": 2048
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00038903902273382105,
+      "loss": 1.1952,
+      "step": 2049
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00038902531542579484,
+      "loss": 1.2161,
+      "step": 2050
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00038901159979403647,
+      "loss": 1.2151,
+      "step": 2051
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00038899787583914993,
+      "loss": 1.1881,
+      "step": 2052
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00038898414356173963,
+      "loss": 1.1957,
+      "step": 2053
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00038897040296241017,
+      "loss": 1.1489,
+      "step": 2054
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00038895665404176674,
+      "loss": 1.2941,
+      "step": 2055
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00038894289680041467,
+      "loss": 1.2614,
+      "step": 2056
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003889291312389598,
+      "loss": 1.189,
+      "step": 2057
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00038891535735800824,
+      "loss": 1.2531,
+      "step": 2058
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003889015751581666,
+      "loss": 1.1639,
+      "step": 2059
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00038888778464004183,
+      "loss": 1.2734,
+      "step": 2060
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00038887398580424106,
+      "loss": 1.2003,
+      "step": 2061
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00038886017865137195,
+      "loss": 1.2404,
+      "step": 2062
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00038884636318204264,
+      "loss": 1.2606,
+      "step": 2063
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00038883253939686125,
+      "loss": 1.218,
+      "step": 2064
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00038881870729643673,
+      "loss": 1.1555,
+      "step": 2065
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00038880486688137807,
+      "loss": 1.3205,
+      "step": 2066
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003887910181522947,
+      "loss": 1.2997,
+      "step": 2067
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00038877716110979655,
+      "loss": 1.2186,
+      "step": 2068
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003887632957544938,
+      "loss": 1.198,
+      "step": 2069
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00038874942208699695,
+      "loss": 1.1947,
+      "step": 2070
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00038873554010791697,
+      "loss": 1.2533,
+      "step": 2071
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003887216498178651,
+      "loss": 1.2655,
+      "step": 2072
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000388707751217453,
+      "loss": 1.2818,
+      "step": 2073
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003886938443072929,
+      "loss": 1.221,
+      "step": 2074
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00038867992908799687,
+      "loss": 1.3074,
+      "step": 2075
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00038866600556017783,
+      "loss": 1.2293,
+      "step": 2076
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000388652073724449,
+      "loss": 1.2692,
+      "step": 2077
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003886381335814236,
+      "loss": 1.284,
+      "step": 2078
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003886241851317158,
+      "loss": 1.2351,
+      "step": 2079
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003886102283759396,
+      "loss": 1.2738,
+      "step": 2080
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00038859626331470967,
+      "loss": 1.3043,
+      "step": 2081
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00038858228994864094,
+      "loss": 1.2763,
+      "step": 2082
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003885683082783487,
+      "loss": 1.2978,
+      "step": 2083
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003885543183044487,
+      "loss": 1.2434,
+      "step": 2084
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038854032002755696,
+      "loss": 1.2478,
+      "step": 2085
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003885263134482899,
+      "loss": 1.3128,
+      "step": 2086
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003885122985672642,
+      "loss": 1.2318,
+      "step": 2087
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003884982753850972,
+      "loss": 1.1805,
+      "step": 2088
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003884842439024063,
+      "loss": 1.2596,
+      "step": 2089
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003884702041198093,
+      "loss": 1.1998,
+      "step": 2090
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003884561560379245,
+      "loss": 1.2746,
+      "step": 2091
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038844209965737065,
+      "loss": 1.1709,
+      "step": 2092
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038842803497876653,
+      "loss": 1.2309,
+      "step": 2093
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003884139620027316,
+      "loss": 1.1984,
+      "step": 2094
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038839988072988544,
+      "loss": 1.226,
+      "step": 2095
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003883857911608482,
+      "loss": 1.2122,
+      "step": 2096
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003883716932962404,
+      "loss": 1.185,
+      "step": 2097
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003883575871366826,
+      "loss": 1.1945,
+      "step": 2098
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038834347268279626,
+      "loss": 1.2578,
+      "step": 2099
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038832934993520275,
+      "loss": 1.1856,
+      "step": 2100
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003883152188945239,
+      "loss": 1.138,
+      "step": 2101
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003883010795613821,
+      "loss": 1.2182,
+      "step": 2102
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003882869319363999,
+      "loss": 1.1814,
+      "step": 2103
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038827277602020035,
+      "loss": 1.2972,
+      "step": 2104
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003882586118134067,
+      "loss": 1.1853,
+      "step": 2105
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003882444393166428,
+      "loss": 1.1897,
+      "step": 2106
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038823025853053274,
+      "loss": 1.088,
+      "step": 2107
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003882160694557008,
+      "loss": 1.2134,
+      "step": 2108
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038820187209277193,
+      "loss": 1.2206,
+      "step": 2109
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003881876664423713,
+      "loss": 1.1143,
+      "step": 2110
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003881734525051244,
+      "loss": 1.1932,
+      "step": 2111
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003881592302816573,
+      "loss": 1.2672,
+      "step": 2112
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003881449997725961,
+      "loss": 1.2639,
+      "step": 2113
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003881307609785674,
+      "loss": 1.2695,
+      "step": 2114
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003881165139001985,
+      "loss": 1.2267,
+      "step": 2115
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003881022585381164,
+      "loss": 1.2359,
+      "step": 2116
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038808799489294907,
+      "loss": 1.1691,
+      "step": 2117
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003880737229653246,
+      "loss": 1.2432,
+      "step": 2118
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038805944275587136,
+      "loss": 1.2814,
+      "step": 2119
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038804515426521817,
+      "loss": 1.2789,
+      "step": 2120
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038803085749399436,
+      "loss": 1.2366,
+      "step": 2121
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003880165524428294,
+      "loss": 1.1668,
+      "step": 2122
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003880022391123532,
+      "loss": 1.2169,
+      "step": 2123
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038798791750319604,
+      "loss": 1.2764,
+      "step": 2124
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003879735876159887,
+      "loss": 1.2654,
+      "step": 2125
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000387959249451362,
+      "loss": 1.1986,
+      "step": 2126
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038794490300994745,
+      "loss": 1.2062,
+      "step": 2127
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038793054829237676,
+      "loss": 1.2634,
+      "step": 2128
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000387916185299282,
+      "loss": 1.2791,
+      "step": 2129
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003879018140312958,
+      "loss": 1.2468,
+      "step": 2130
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003878874344890508,
+      "loss": 1.1722,
+      "step": 2131
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003878730466731803,
+      "loss": 1.2553,
+      "step": 2132
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038785865058431785,
+      "loss": 1.1864,
+      "step": 2133
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038784424622309745,
+      "loss": 1.2657,
+      "step": 2134
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038782983359015325,
+      "loss": 1.2382,
+      "step": 2135
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003878154126861201,
+      "loss": 1.2385,
+      "step": 2136
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003878009835116328,
+      "loss": 1.2773,
+      "step": 2137
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003877865460673269,
+      "loss": 1.2292,
+      "step": 2138
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003877721003538382,
+      "loss": 1.2617,
+      "step": 2139
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003877576463718027,
+      "loss": 1.2752,
+      "step": 2140
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00038774318412185683,
+      "loss": 1.2784,
+      "step": 2141
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003877287136046376,
+      "loss": 1.2543,
+      "step": 2142
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003877142348207821,
+      "loss": 1.1272,
+      "step": 2143
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00038769974777092794,
+      "loss": 1.1859,
+      "step": 2144
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003876852524557131,
+      "loss": 1.2112,
+      "step": 2145
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00038767074887577574,
+      "loss": 1.1835,
+      "step": 2146
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00038765623703175474,
+      "loss": 1.2294,
+      "step": 2147
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003876417169242889,
+      "loss": 1.1904,
+      "step": 2148
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003876271885540178,
+      "loss": 1.2288,
+      "step": 2149
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00038761265192158113,
+      "loss": 1.2978,
+      "step": 2150
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000387598107027619,
+      "loss": 1.2226,
+      "step": 2151
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00038758355387277186,
+      "loss": 1.2607,
+      "step": 2152
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00038756899245768057,
+      "loss": 1.2088,
+      "step": 2153
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003875544227829865,
+      "loss": 1.2948,
+      "step": 2154
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000387539844849331,
+      "loss": 1.2923,
+      "step": 2155
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003875252586573561,
+      "loss": 1.2576,
+      "step": 2156
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003875106642077041,
+      "loss": 1.2198,
+      "step": 2157
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003874960615010177,
+      "loss": 1.1544,
+      "step": 2158
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00038748145053794,
+      "loss": 1.3002,
+      "step": 2159
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003874668313191141,
+      "loss": 1.1918,
+      "step": 2160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003874522038451841,
+      "loss": 1.2379,
+      "step": 2161
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000387437568116794,
+      "loss": 1.269,
+      "step": 2162
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00038742292413458814,
+      "loss": 1.2504,
+      "step": 2163
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00038740827189921153,
+      "loss": 1.193,
+      "step": 2164
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00038739361141130937,
+      "loss": 1.2756,
+      "step": 2165
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003873789426715272,
+      "loss": 1.1024,
+      "step": 2166
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00038736426568051097,
+      "loss": 1.2371,
+      "step": 2167
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00038734958043890695,
+      "loss": 1.2429,
+      "step": 2168
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003873348869473619,
+      "loss": 1.1944,
+      "step": 2169
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003873201852065227,
+      "loss": 1.1793,
+      "step": 2170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00038730547521703684,
+      "loss": 1.1663,
+      "step": 2171
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003872907569795521,
+      "loss": 1.2475,
+      "step": 2172
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003872760304947165,
+      "loss": 1.2297,
+      "step": 2173
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00038726129576317856,
+      "loss": 1.1971,
+      "step": 2174
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00038724655278558716,
+      "loss": 1.2097,
+      "step": 2175
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003872318015625915,
+      "loss": 1.1955,
+      "step": 2176
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003872170420948411,
+      "loss": 1.2197,
+      "step": 2177
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00038720227438298596,
+      "loss": 1.2234,
+      "step": 2178
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00038718749842767635,
+      "loss": 1.1964,
+      "step": 2179
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00038717271422956294,
+      "loss": 1.2583,
+      "step": 2180
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00038715792178929665,
+      "loss": 1.1873,
+      "step": 2181
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000387143121107529,
+      "loss": 1.3357,
+      "step": 2182
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00038712831218491175,
+      "loss": 1.1971,
+      "step": 2183
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00038711349502209685,
+      "loss": 1.2276,
+      "step": 2184
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00038709866961973694,
+      "loss": 1.2578,
+      "step": 2185
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00038708383597848477,
+      "loss": 1.2361,
+      "step": 2186
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00038706899409899357,
+      "loss": 1.2448,
+      "step": 2187
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003870541439819169,
+      "loss": 1.2156,
+      "step": 2188
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003870392856279087,
+      "loss": 1.2649,
+      "step": 2189
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00038702441903762315,
+      "loss": 1.253,
+      "step": 2190
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00038700954421171503,
+      "loss": 1.2027,
+      "step": 2191
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003869946611508393,
+      "loss": 1.1786,
+      "step": 2192
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00038697976985565136,
+      "loss": 1.2683,
+      "step": 2193
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00038696487032680694,
+      "loss": 1.3262,
+      "step": 2194
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00038694996256496213,
+      "loss": 1.2617,
+      "step": 2195
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003869350465707735,
+      "loss": 1.2284,
+      "step": 2196
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00038692012234489763,
+      "loss": 1.2368,
+      "step": 2197
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003869051898879919,
+      "loss": 1.3094,
+      "step": 2198
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00038689024920071375,
+      "loss": 1.1903,
+      "step": 2199
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00038687530028372124,
+      "loss": 1.2426,
+      "step": 2200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00038686034313767257,
+      "loss": 1.3662,
+      "step": 2201
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003868453777632263,
+      "loss": 1.1622,
+      "step": 2202
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003868304041610416,
+      "loss": 1.1763,
+      "step": 2203
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00038681542233177767,
+      "loss": 1.1985,
+      "step": 2204
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003868004322760943,
+      "loss": 1.2513,
+      "step": 2205
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003867854339946516,
+      "loss": 1.2285,
+      "step": 2206
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00038677042748810995,
+      "loss": 1.2117,
+      "step": 2207
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00038675541275713025,
+      "loss": 1.3088,
+      "step": 2208
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003867403898023736,
+      "loss": 1.182,
+      "step": 2209
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003867253586245016,
+      "loss": 1.1868,
+      "step": 2210
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00038671031922417613,
+      "loss": 1.2341,
+      "step": 2211
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00038669527160205936,
+      "loss": 1.2201,
+      "step": 2212
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00038668021575881406,
+      "loss": 1.2394,
+      "step": 2213
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00038666515169510316,
+      "loss": 1.1876,
+      "step": 2214
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003866500794115899,
+      "loss": 1.2802,
+      "step": 2215
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00038663499890893816,
+      "loss": 1.2689,
+      "step": 2216
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00038661991018781183,
+      "loss": 1.2062,
+      "step": 2217
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00038660481324887555,
+      "loss": 1.201,
+      "step": 2218
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00038658970809279393,
+      "loss": 1.3033,
+      "step": 2219
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00038657459472023225,
+      "loss": 1.2162,
+      "step": 2220
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00038655947313185593,
+      "loss": 1.1986,
+      "step": 2221
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00038654434332833094,
+      "loss": 1.1824,
+      "step": 2222
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003865292053103234,
+      "loss": 1.2109,
+      "step": 2223
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003865140590785001,
+      "loss": 1.2017,
+      "step": 2224
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00038649890463352785,
+      "loss": 1.2044,
+      "step": 2225
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00038648374197607397,
+      "loss": 1.1959,
+      "step": 2226
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003864685711068062,
+      "loss": 1.1893,
+      "step": 2227
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003864533920263926,
+      "loss": 1.2692,
+      "step": 2228
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003864382047355016,
+      "loss": 1.1525,
+      "step": 2229
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000386423009234802,
+      "loss": 1.1844,
+      "step": 2230
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003864078055249628,
+      "loss": 1.193,
+      "step": 2231
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003863925936066536,
+      "loss": 1.1501,
+      "step": 2232
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00038637737348054416,
+      "loss": 1.2024,
+      "step": 2233
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003863621451473048,
+      "loss": 1.3053,
+      "step": 2234
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00038634690860760607,
+      "loss": 1.2623,
+      "step": 2235
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00038633166386211893,
+      "loss": 1.219,
+      "step": 2236
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003863164109115146,
+      "loss": 1.2414,
+      "step": 2237
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00038630114975646485,
+      "loss": 1.1761,
+      "step": 2238
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00038628588039764166,
+      "loss": 1.1895,
+      "step": 2239
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003862706028357174,
+      "loss": 1.2787,
+      "step": 2240
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00038625531707136483,
+      "loss": 1.2716,
+      "step": 2241
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003862400231052571,
+      "loss": 1.1795,
+      "step": 2242
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00038622472093806755,
+      "loss": 1.1461,
+      "step": 2243
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003862094105704702,
+      "loss": 1.261,
+      "step": 2244
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003861940920031391,
+      "loss": 1.1555,
+      "step": 2245
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00038617876523674885,
+      "loss": 1.2089,
+      "step": 2246
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00038616343027197443,
+      "loss": 1.2305,
+      "step": 2247
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00038614808710949096,
+      "loss": 1.1929,
+      "step": 2248
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00038613273574997415,
+      "loss": 1.2483,
+      "step": 2249
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003861173761941001,
+      "loss": 1.2073,
+      "step": 2250
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000386102008442545,
+      "loss": 1.2546,
+      "step": 2251
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003860866324959857,
+      "loss": 1.1969,
+      "step": 2252
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00038607124835509917,
+      "loss": 1.1851,
+      "step": 2253
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000386055856020563,
+      "loss": 1.2201,
+      "step": 2254
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003860404554930548,
+      "loss": 1.2115,
+      "step": 2255
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00038602504677325293,
+      "loss": 1.2036,
+      "step": 2256
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00038600962986183574,
+      "loss": 1.2922,
+      "step": 2257
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00038599420475948223,
+      "loss": 1.2097,
+      "step": 2258
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00038597877146687165,
+      "loss": 1.2154,
+      "step": 2259
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003859633299846834,
+      "loss": 1.1909,
+      "step": 2260
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00038594788031359777,
+      "loss": 1.1904,
+      "step": 2261
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00038593242245429487,
+      "loss": 1.2004,
+      "step": 2262
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003859169564074554,
+      "loss": 1.2457,
+      "step": 2263
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003859014821737605,
+      "loss": 1.219,
+      "step": 2264
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003858859997538915,
+      "loss": 1.2377,
+      "step": 2265
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003858705091485302,
+      "loss": 1.2262,
+      "step": 2266
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00038585501035835875,
+      "loss": 1.2049,
+      "step": 2267
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003858395033840596,
+      "loss": 1.2376,
+      "step": 2268
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003858239882263156,
+      "loss": 1.2361,
+      "step": 2269
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00038580846488581,
+      "loss": 1.2698,
+      "step": 2270
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00038579293336322626,
+      "loss": 1.2865,
+      "step": 2271
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00038577739365924844,
+      "loss": 1.215,
+      "step": 2272
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00038576184577456083,
+      "loss": 1.213,
+      "step": 2273
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00038574628970984803,
+      "loss": 1.238,
+      "step": 2274
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00038573072546579495,
+      "loss": 1.1559,
+      "step": 2275
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00038571515304308717,
+      "loss": 1.2787,
+      "step": 2276
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003856995724424103,
+      "loss": 1.1847,
+      "step": 2277
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00038568398366445046,
+      "loss": 1.2672,
+      "step": 2278
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00038566838670989407,
+      "loss": 1.1553,
+      "step": 2279
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000385652781579428,
+      "loss": 1.1943,
+      "step": 2280
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003856371682737394,
+      "loss": 1.2727,
+      "step": 2281
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00038562154679351574,
+      "loss": 1.2138,
+      "step": 2282
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00038560591713944503,
+      "loss": 1.2608,
+      "step": 2283
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00038559027931221534,
+      "loss": 1.1992,
+      "step": 2284
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003855746333125155,
+      "loss": 1.1574,
+      "step": 2285
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00038555897914103435,
+      "loss": 1.2539,
+      "step": 2286
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00038554331679846123,
+      "loss": 1.2275,
+      "step": 2287
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00038552764628548583,
+      "loss": 1.2935,
+      "step": 2288
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00038551196760279834,
+      "loss": 1.2178,
+      "step": 2289
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003854962807510889,
+      "loss": 1.1928,
+      "step": 2290
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00038548058573104857,
+      "loss": 1.1752,
+      "step": 2291
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00038546488254336827,
+      "loss": 1.2048,
+      "step": 2292
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00038544917118873956,
+      "loss": 1.2536,
+      "step": 2293
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00038543345166785434,
+      "loss": 1.1743,
+      "step": 2294
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003854177239814048,
+      "loss": 1.2561,
+      "step": 2295
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00038540198813008346,
+      "loss": 1.2084,
+      "step": 2296
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00038538624411458324,
+      "loss": 1.3013,
+      "step": 2297
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003853704919355974,
+      "loss": 1.311,
+      "step": 2298
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00038535473159381973,
+      "loss": 1.2326,
+      "step": 2299
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00038533896308994416,
+      "loss": 1.2106,
+      "step": 2300
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00038532318642466506,
+      "loss": 1.2074,
+      "step": 2301
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00038530740159867705,
+      "loss": 1.2347,
+      "step": 2302
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00038529160861267546,
+      "loss": 1.2193,
+      "step": 2303
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003852758074673555,
+      "loss": 1.2192,
+      "step": 2304
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00038525999816341303,
+      "loss": 1.2195,
+      "step": 2305
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003852441807015443,
+      "loss": 1.2157,
+      "step": 2306
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003852283550824457,
+      "loss": 1.2149,
+      "step": 2307
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003852125213068143,
+      "loss": 1.144,
+      "step": 2308
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00038519667937534713,
+      "loss": 1.2068,
+      "step": 2309
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003851808292887419,
+      "loss": 1.226,
+      "step": 2310
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00038516497104769653,
+      "loss": 1.336,
+      "step": 2311
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003851491046529095,
+      "loss": 1.2482,
+      "step": 2312
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00038513323010507916,
+      "loss": 1.1661,
+      "step": 2313
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00038511734740490484,
+      "loss": 1.3134,
+      "step": 2314
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00038510145655308586,
+      "loss": 1.2281,
+      "step": 2315
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003850855575503219,
+      "loss": 1.2472,
+      "step": 2316
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003850696503973131,
+      "loss": 1.2038,
+      "step": 2317
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003850537350947599,
+      "loss": 1.2726,
+      "step": 2318
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00038503781164336327,
+      "loss": 1.1542,
+      "step": 2319
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003850218800438243,
+      "loss": 1.2402,
+      "step": 2320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003850059402968445,
+      "loss": 1.1933,
+      "step": 2321
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00038498999240312585,
+      "loss": 1.2405,
+      "step": 2322
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003849740363633706,
+      "loss": 1.2104,
+      "step": 2323
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003849580721782814,
+      "loss": 1.2026,
+      "step": 2324
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003849420998485611,
+      "loss": 1.1892,
+      "step": 2325
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003849261193749132,
+      "loss": 1.1788,
+      "step": 2326
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003849101307580413,
+      "loss": 1.206,
+      "step": 2327
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00038489413399864957,
+      "loss": 1.1738,
+      "step": 2328
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003848781290974423,
+      "loss": 1.2184,
+      "step": 2329
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003848621160551243,
+      "loss": 1.1326,
+      "step": 2330
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003848460948724008,
+      "loss": 1.2611,
+      "step": 2331
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00038483006554997714,
+      "loss": 1.2568,
+      "step": 2332
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003848140280885593,
+      "loss": 1.3598,
+      "step": 2333
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00038479798248885344,
+      "loss": 1.1757,
+      "step": 2334
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00038478192875156605,
+      "loss": 1.1878,
+      "step": 2335
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003847658668774042,
+      "loss": 1.2582,
+      "step": 2336
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00038474979686707513,
+      "loss": 1.2312,
+      "step": 2337
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00038473371872128643,
+      "loss": 1.2328,
+      "step": 2338
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003847176324407461,
+      "loss": 1.173,
+      "step": 2339
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00038470153802616256,
+      "loss": 1.213,
+      "step": 2340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00038468543547824456,
+      "loss": 1.1819,
+      "step": 2341
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000384669324797701,
+      "loss": 1.1637,
+      "step": 2342
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00038465320598524155,
+      "loss": 1.2763,
+      "step": 2343
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003846370790415758,
+      "loss": 1.229,
+      "step": 2344
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00038462094396741407,
+      "loss": 1.2509,
+      "step": 2345
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003846048007634667,
+      "loss": 1.2914,
+      "step": 2346
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00038458864943044464,
+      "loss": 1.204,
+      "step": 2347
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00038457248996905916,
+      "loss": 1.2022,
+      "step": 2348
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003845563223800218,
+      "loss": 1.27,
+      "step": 2349
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00038454014666404446,
+      "loss": 1.1429,
+      "step": 2350
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003845239628218394,
+      "loss": 1.2109,
+      "step": 2351
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00038450777085411944,
+      "loss": 1.1887,
+      "step": 2352
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00038449157076159753,
+      "loss": 1.2499,
+      "step": 2353
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000384475362544987,
+      "loss": 1.2214,
+      "step": 2354
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003844591462050015,
+      "loss": 1.2471,
+      "step": 2355
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00038444292174235527,
+      "loss": 1.2006,
+      "step": 2356
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003844266891577627,
+      "loss": 1.1745,
+      "step": 2357
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003844104484519385,
+      "loss": 1.2458,
+      "step": 2358
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00038439419962559804,
+      "loss": 1.1919,
+      "step": 2359
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00038437794267945667,
+      "loss": 1.2607,
+      "step": 2360
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00038436167761423024,
+      "loss": 1.1977,
+      "step": 2361
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003843454044306351,
+      "loss": 1.1996,
+      "step": 2362
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003843291231293878,
+      "loss": 1.2557,
+      "step": 2363
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00038431283371120526,
+      "loss": 1.2707,
+      "step": 2364
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003842965361768048,
+      "loss": 1.2467,
+      "step": 2365
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003842802305269041,
+      "loss": 1.1393,
+      "step": 2366
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003842639167622212,
+      "loss": 1.2086,
+      "step": 2367
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00038424759488347434,
+      "loss": 1.3064,
+      "step": 2368
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003842312648913824,
+      "loss": 1.224,
+      "step": 2369
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003842149267866644,
+      "loss": 1.2604,
+      "step": 2370
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00038419858057003983,
+      "loss": 1.252,
+      "step": 2371
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003841822262422285,
+      "loss": 1.2372,
+      "step": 2372
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003841658638039506,
+      "loss": 1.2756,
+      "step": 2373
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00038414949325592647,
+      "loss": 1.2512,
+      "step": 2374
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003841331145988772,
+      "loss": 1.1748,
+      "step": 2375
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003841167278335239,
+      "loss": 1.1925,
+      "step": 2376
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003841003329605883,
+      "loss": 1.2608,
+      "step": 2377
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00038408392998079215,
+      "loss": 1.2107,
+      "step": 2378
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003840675188948579,
+      "loss": 1.1566,
+      "step": 2379
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00038405109970350814,
+      "loss": 1.1869,
+      "step": 2380
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000384034672407466,
+      "loss": 1.1992,
+      "step": 2381
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003840182370074547,
+      "loss": 1.1642,
+      "step": 2382
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00038400179350419803,
+      "loss": 1.1751,
+      "step": 2383
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003839853418984202,
+      "loss": 1.2409,
+      "step": 2384
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003839688821908455,
+      "loss": 1.2305,
+      "step": 2385
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00038395241438219876,
+      "loss": 1.2074,
+      "step": 2386
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00038393593847320525,
+      "loss": 1.2887,
+      "step": 2387
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00038391945446459034,
+      "loss": 1.2296,
+      "step": 2388
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00038390296235707997,
+      "loss": 1.1547,
+      "step": 2389
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00038388646215140043,
+      "loss": 1.2344,
+      "step": 2390
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00038386995384827815,
+      "loss": 1.279,
+      "step": 2391
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00038385343744844024,
+      "loss": 1.2279,
+      "step": 2392
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00038383691295261393,
+      "loss": 1.2122,
+      "step": 2393
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003838203803615269,
+      "loss": 1.2235,
+      "step": 2394
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003838038396759071,
+      "loss": 1.2366,
+      "step": 2395
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00038378729089648286,
+      "loss": 1.2507,
+      "step": 2396
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003837707340239831,
+      "loss": 1.1205,
+      "step": 2397
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003837541690591368,
+      "loss": 1.1734,
+      "step": 2398
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00038373759600267326,
+      "loss": 1.3716,
+      "step": 2399
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00038372101485532255,
+      "loss": 1.2239,
+      "step": 2400
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00038370442561781455,
+      "loss": 1.2283,
+      "step": 2401
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003836878282908799,
+      "loss": 1.1752,
+      "step": 2402
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003836712228752496,
+      "loss": 1.2086,
+      "step": 2403
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003836546093716546,
+      "loss": 1.2364,
+      "step": 2404
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003836379877808266,
+      "loss": 1.1513,
+      "step": 2405
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003836213581034975,
+      "loss": 1.1635,
+      "step": 2406
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003836047203403997,
+      "loss": 1.1835,
+      "step": 2407
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003835880744922657,
+      "loss": 1.1881,
+      "step": 2408
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00038357142055982853,
+      "loss": 1.2688,
+      "step": 2409
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00038355475854382165,
+      "loss": 1.2455,
+      "step": 2410
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00038353808844497866,
+      "loss": 1.1675,
+      "step": 2411
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003835214102640337,
+      "loss": 1.1698,
+      "step": 2412
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003835047240017211,
+      "loss": 1.1383,
+      "step": 2413
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003834880296587758,
+      "loss": 1.1902,
+      "step": 2414
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00038347132723593277,
+      "loss": 1.1483,
+      "step": 2415
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003834546167339276,
+      "loss": 1.1634,
+      "step": 2416
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000383437898153496,
+      "loss": 1.196,
+      "step": 2417
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00038342117149537443,
+      "loss": 1.2746,
+      "step": 2418
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003834044367602992,
+      "loss": 1.1978,
+      "step": 2419
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00038338769394900737,
+      "loss": 1.278,
+      "step": 2420
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003833709430622361,
+      "loss": 1.2251,
+      "step": 2421
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00038335418410072314,
+      "loss": 1.2634,
+      "step": 2422
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003833374170652064,
+      "loss": 1.2278,
+      "step": 2423
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003833206419564242,
+      "loss": 1.188,
+      "step": 2424
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003833038587751153,
+      "loss": 1.1017,
+      "step": 2425
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00038328706752201863,
+      "loss": 1.1762,
+      "step": 2426
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003832702681978737,
+      "loss": 1.282,
+      "step": 2427
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003832534608034203,
+      "loss": 1.2255,
+      "step": 2428
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003832366453393984,
+      "loss": 1.1854,
+      "step": 2429
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00038321982180654856,
+      "loss": 1.271,
+      "step": 2430
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003832029902056116,
+      "loss": 1.1824,
+      "step": 2431
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00038318615053732874,
+      "loss": 1.1393,
+      "step": 2432
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003831693028024414,
+      "loss": 1.1687,
+      "step": 2433
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00038315244700169157,
+      "loss": 1.2884,
+      "step": 2434
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003831355831358215,
+      "loss": 1.2325,
+      "step": 2435
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00038311871120557365,
+      "loss": 1.1188,
+      "step": 2436
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00038310183121169117,
+      "loss": 1.1988,
+      "step": 2437
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00038308494315491724,
+      "loss": 1.1458,
+      "step": 2438
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00038306804703599557,
+      "loss": 1.1966,
+      "step": 2439
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003830511428556702,
+      "loss": 1.2659,
+      "step": 2440
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003830342306146855,
+      "loss": 1.1909,
+      "step": 2441
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00038301731031378615,
+      "loss": 1.2377,
+      "step": 2442
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00038300038195371723,
+      "loss": 1.2588,
+      "step": 2443
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003829834455352242,
+      "loss": 1.2376,
+      "step": 2444
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00038296650105905296,
+      "loss": 1.1446,
+      "step": 2445
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00038294954852594954,
+      "loss": 1.1814,
+      "step": 2446
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003829325879366605,
+      "loss": 1.2102,
+      "step": 2447
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00038291561929193273,
+      "loss": 1.2319,
+      "step": 2448
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00038289864259251333,
+      "loss": 1.2513,
+      "step": 2449
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003828816578391499,
+      "loss": 1.2267,
+      "step": 2450
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003828646650325904,
+      "loss": 1.1701,
+      "step": 2451
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00038284766417358317,
+      "loss": 1.2358,
+      "step": 2452
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00038283065526287675,
+      "loss": 1.2655,
+      "step": 2453
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00038281363830122017,
+      "loss": 1.2012,
+      "step": 2454
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00038279661328936273,
+      "loss": 1.2085,
+      "step": 2455
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00038277958022805416,
+      "loss": 1.143,
+      "step": 2456
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00038276253911804456,
+      "loss": 1.2771,
+      "step": 2457
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00038274548996008426,
+      "loss": 1.2394,
+      "step": 2458
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000382728432754924,
+      "loss": 1.2529,
+      "step": 2459
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00038271136750331493,
+      "loss": 1.1312,
+      "step": 2460
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003826942942060086,
+      "loss": 1.293,
+      "step": 2461
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003826772128637567,
+      "loss": 1.1887,
+      "step": 2462
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00038266012347731144,
+      "loss": 1.2064,
+      "step": 2463
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003826430260474254,
+      "loss": 1.2925,
+      "step": 2464
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00038262592057485147,
+      "loss": 1.2112,
+      "step": 2465
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00038260880706034287,
+      "loss": 1.1171,
+      "step": 2466
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003825916855046531,
+      "loss": 1.2191,
+      "step": 2467
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00038257455590853627,
+      "loss": 1.3451,
+      "step": 2468
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003825574182727466,
+      "loss": 1.1039,
+      "step": 2469
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00038254027259803877,
+      "loss": 1.2089,
+      "step": 2470
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003825231188851677,
+      "loss": 1.2927,
+      "step": 2471
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00038250595713488887,
+      "loss": 1.2139,
+      "step": 2472
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003824887873479579,
+      "loss": 1.1352,
+      "step": 2473
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00038247160952513095,
+      "loss": 1.2169,
+      "step": 2474
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00038245442366716446,
+      "loss": 1.2594,
+      "step": 2475
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003824372297748151,
+      "loss": 1.3038,
+      "step": 2476
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00038242002784884006,
+      "loss": 1.2111,
+      "step": 2477
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003824028178899969,
+      "loss": 1.1874,
+      "step": 2478
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003823855998990433,
+      "loss": 1.2278,
+      "step": 2479
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00038236837387673763,
+      "loss": 1.1863,
+      "step": 2480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003823511398238383,
+      "loss": 1.2035,
+      "step": 2481
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003823338977411043,
+      "loss": 1.1641,
+      "step": 2482
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003823166476292948,
+      "loss": 1.1333,
+      "step": 2483
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00038229938948916944,
+      "loss": 1.2936,
+      "step": 2484
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00038228212332148823,
+      "loss": 1.2822,
+      "step": 2485
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00038226484912701154,
+      "loss": 1.1615,
+      "step": 2486
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003822475669064999,
+      "loss": 1.1458,
+      "step": 2487
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00038223027666071436,
+      "loss": 1.1577,
+      "step": 2488
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00038221297839041637,
+      "loss": 1.2581,
+      "step": 2489
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003821956720963676,
+      "loss": 1.1369,
+      "step": 2490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003821783577793301,
+      "loss": 1.1761,
+      "step": 2491
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00038216103544006636,
+      "loss": 1.1805,
+      "step": 2492
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00038214370507933925,
+      "loss": 1.1801,
+      "step": 2493
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003821263666979118,
+      "loss": 1.0652,
+      "step": 2494
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003821090202965475,
+      "loss": 1.2068,
+      "step": 2495
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00038209166587601024,
+      "loss": 1.2288,
+      "step": 2496
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003820743034370642,
+      "loss": 1.2826,
+      "step": 2497
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00038205693298047394,
+      "loss": 1.214,
+      "step": 2498
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003820395545070044,
+      "loss": 1.1959,
+      "step": 2499
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003820221680174208,
+      "loss": 1.1377,
+      "step": 2500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003820047735124888,
+      "loss": 1.1199,
+      "step": 2501
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003819873709929743,
+      "loss": 1.2006,
+      "step": 2502
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003819699604596437,
+      "loss": 1.2271,
+      "step": 2503
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003819525419132636,
+      "loss": 1.1771,
+      "step": 2504
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003819351153546011,
+      "loss": 1.2032,
+      "step": 2505
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003819176807844235,
+      "loss": 1.2803,
+      "step": 2506
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003819002382034986,
+      "loss": 1.1773,
+      "step": 2507
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003818827876125944,
+      "loss": 1.1455,
+      "step": 2508
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00038186532901247943,
+      "loss": 1.2094,
+      "step": 2509
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00038184786240392246,
+      "loss": 1.1879,
+      "step": 2510
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003818303877876926,
+      "loss": 1.1952,
+      "step": 2511
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00038181290516455934,
+      "loss": 1.2417,
+      "step": 2512
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003817954145352926,
+      "loss": 1.2178,
+      "step": 2513
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003817779159006624,
+      "loss": 1.1563,
+      "step": 2514
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00038176040926143955,
+      "loss": 1.2723,
+      "step": 2515
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003817428946183947,
+      "loss": 1.1912,
+      "step": 2516
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00038172537197229934,
+      "loss": 1.3139,
+      "step": 2517
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003817078413239249,
+      "loss": 1.2517,
+      "step": 2518
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00038169030267404345,
+      "loss": 1.1896,
+      "step": 2519
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003816727560234273,
+      "loss": 1.2853,
+      "step": 2520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000381655201372849,
+      "loss": 1.1445,
+      "step": 2521
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003816376387230817,
+      "loss": 1.2663,
+      "step": 2522
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00038162006807489874,
+      "loss": 1.1895,
+      "step": 2523
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00038160248942907375,
+      "loss": 1.2221,
+      "step": 2524
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000381584902786381,
+      "loss": 1.26,
+      "step": 2525
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00038156730814759476,
+      "loss": 1.2226,
+      "step": 2526
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00038154970551348977,
+      "loss": 1.1537,
+      "step": 2527
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003815320948848413,
+      "loss": 1.1543,
+      "step": 2528
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00038151447626242485,
+      "loss": 1.2859,
+      "step": 2529
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003814968496470161,
+      "loss": 1.2136,
+      "step": 2530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00038147921503939135,
+      "loss": 1.3204,
+      "step": 2531
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003814615724403271,
+      "loss": 1.2539,
+      "step": 2532
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003814439218506003,
+      "loss": 1.2502,
+      "step": 2533
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003814262632709881,
+      "loss": 1.2601,
+      "step": 2534
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003814085967022682,
+      "loss": 1.1803,
+      "step": 2535
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003813909221452184,
+      "loss": 1.2429,
+      "step": 2536
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003813732396006172,
+      "loss": 1.2393,
+      "step": 2537
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00038135554906924303,
+      "loss": 1.1804,
+      "step": 2538
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00038133785055187506,
+      "loss": 1.1816,
+      "step": 2539
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003813201440492926,
+      "loss": 1.1755,
+      "step": 2540
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003813024295622754,
+      "loss": 1.216,
+      "step": 2541
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003812847070916034,
+      "loss": 1.253,
+      "step": 2542
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00038126697663805705,
+      "loss": 1.2578,
+      "step": 2543
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00038124923820241716,
+      "loss": 1.2326,
+      "step": 2544
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00038123149178546483,
+      "loss": 1.219,
+      "step": 2545
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003812137373879815,
+      "loss": 1.1874,
+      "step": 2546
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000381195975010749,
+      "loss": 1.208,
+      "step": 2547
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00038117820465454947,
+      "loss": 1.2334,
+      "step": 2548
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00038116042632016555,
+      "loss": 1.1244,
+      "step": 2549
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003811426400083799,
+      "loss": 1.1778,
+      "step": 2550
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00038112484571997587,
+      "loss": 1.1296,
+      "step": 2551
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000381107043455737,
+      "loss": 1.2114,
+      "step": 2552
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003810892332164473,
+      "loss": 1.1885,
+      "step": 2553
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00038107141500289086,
+      "loss": 1.2261,
+      "step": 2554
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003810535888158524,
+      "loss": 1.3116,
+      "step": 2555
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00038103575465611706,
+      "loss": 1.1681,
+      "step": 2556
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003810179125244699,
+      "loss": 1.1907,
+      "step": 2557
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00038100006242169676,
+      "loss": 1.219,
+      "step": 2558
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003809822043485836,
+      "loss": 1.2238,
+      "step": 2559
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00038096433830591683,
+      "loss": 1.2245,
+      "step": 2560
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003809464642944832,
+      "loss": 1.1724,
+      "step": 2561
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00038092858231506975,
+      "loss": 1.3,
+      "step": 2562
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000380910692368464,
+      "loss": 1.2597,
+      "step": 2563
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003808927944554536,
+      "loss": 1.2642,
+      "step": 2564
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003808748885768267,
+      "loss": 1.2153,
+      "step": 2565
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00038085697473337187,
+      "loss": 1.2112,
+      "step": 2566
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00038083905292587793,
+      "loss": 1.2213,
+      "step": 2567
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003808211231551341,
+      "loss": 1.2658,
+      "step": 2568
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003808031854219298,
+      "loss": 1.2524,
+      "step": 2569
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00038078523972705507,
+      "loss": 1.1712,
+      "step": 2570
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00038076728607129996,
+      "loss": 1.1797,
+      "step": 2571
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003807493244554552,
+      "loss": 1.2246,
+      "step": 2572
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00038073135488031164,
+      "loss": 1.2088,
+      "step": 2573
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003807133773466608,
+      "loss": 1.2306,
+      "step": 2574
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000380695391855294,
+      "loss": 1.1585,
+      "step": 2575
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003806773984070034,
+      "loss": 1.2155,
+      "step": 2576
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003806593970025813,
+      "loss": 1.2694,
+      "step": 2577
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003806413876428205,
+      "loss": 1.2191,
+      "step": 2578
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00038062337032851386,
+      "loss": 1.2162,
+      "step": 2579
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003806053450604549,
+      "loss": 1.1968,
+      "step": 2580
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003805873118394373,
+      "loss": 1.199,
+      "step": 2581
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00038056927066625523,
+      "loss": 1.2593,
+      "step": 2582
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00038055122154170303,
+      "loss": 1.1981,
+      "step": 2583
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003805331644665756,
+      "loss": 1.1401,
+      "step": 2584
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000380515099441668,
+      "loss": 1.3024,
+      "step": 2585
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003804970264677758,
+      "loss": 1.2295,
+      "step": 2586
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003804789455456948,
+      "loss": 1.174,
+      "step": 2587
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00038046085667622117,
+      "loss": 1.1955,
+      "step": 2588
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00038044275986015146,
+      "loss": 1.2875,
+      "step": 2589
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00038042465509828256,
+      "loss": 1.1876,
+      "step": 2590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003804065423914118,
+      "loss": 1.205,
+      "step": 2591
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00038038842174033667,
+      "loss": 1.2992,
+      "step": 2592
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003803702931458552,
+      "loss": 1.15,
+      "step": 2593
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003803521566087655,
+      "loss": 1.1854,
+      "step": 2594
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003803340121298665,
+      "loss": 1.2217,
+      "step": 2595
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003803158597099569,
+      "loss": 1.2101,
+      "step": 2596
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003802976993498363,
+      "loss": 1.1658,
+      "step": 2597
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003802795310503042,
+      "loss": 1.1239,
+      "step": 2598
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00038026135481216073,
+      "loss": 1.1589,
+      "step": 2599
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003802431706362063,
+      "loss": 1.2684,
+      "step": 2600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00038022497852324157,
+      "loss": 1.1389,
+      "step": 2601
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003802067784740677,
+      "loss": 1.1713,
+      "step": 2602
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00038018857048948605,
+      "loss": 1.1743,
+      "step": 2603
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003801703545702985,
+      "loss": 1.2018,
+      "step": 2604
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003801521307173072,
+      "loss": 1.2158,
+      "step": 2605
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00038013389893131453,
+      "loss": 1.267,
+      "step": 2606
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00038011565921312337,
+      "loss": 1.1441,
+      "step": 2607
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000380097411563537,
+      "loss": 1.2403,
+      "step": 2608
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003800791559833588,
+      "loss": 1.1866,
+      "step": 2609
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00038006089247339276,
+      "loss": 1.2435,
+      "step": 2610
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003800426210344431,
+      "loss": 1.1812,
+      "step": 2611
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003800243416673143,
+      "loss": 1.2454,
+      "step": 2612
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003800060543728115,
+      "loss": 1.2376,
+      "step": 2613
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003799877591517398,
+      "loss": 1.1723,
+      "step": 2614
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00037996945600490486,
+      "loss": 1.2737,
+      "step": 2615
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003799511449331128,
+      "loss": 1.2901,
+      "step": 2616
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00037993282593716977,
+      "loss": 1.2059,
+      "step": 2617
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00037991449901788254,
+      "loss": 1.2387,
+      "step": 2618
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00037989616417605807,
+      "loss": 1.1332,
+      "step": 2619
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003798778214125038,
+      "loss": 1.1677,
+      "step": 2620
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00037985947072802744,
+      "loss": 1.2367,
+      "step": 2621
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00037984111212343707,
+      "loss": 1.1788,
+      "step": 2622
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000379822745599541,
+      "loss": 1.1972,
+      "step": 2623
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003798043711571483,
+      "loss": 1.2966,
+      "step": 2624
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00037978598879706766,
+      "loss": 1.2349,
+      "step": 2625
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003797675985201089,
+      "loss": 1.1626,
+      "step": 2626
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00037974920032708166,
+      "loss": 1.1537,
+      "step": 2627
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00037973079421879614,
+      "loss": 1.1681,
+      "step": 2628
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000379712380196063,
+      "loss": 1.1267,
+      "step": 2629
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00037969395825969285,
+      "loss": 1.2192,
+      "step": 2630
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000379675528410497,
+      "loss": 1.1517,
+      "step": 2631
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00037965709064928705,
+      "loss": 1.1303,
+      "step": 2632
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003796386449768749,
+      "loss": 1.1312,
+      "step": 2633
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003796201913940728,
+      "loss": 1.2103,
+      "step": 2634
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003796017299016933,
+      "loss": 1.1946,
+      "step": 2635
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003795832605005494,
+      "loss": 1.2405,
+      "step": 2636
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00037956478319145444,
+      "loss": 1.2312,
+      "step": 2637
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00037954629797522194,
+      "loss": 1.1823,
+      "step": 2638
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00037952780485266607,
+      "loss": 1.1279,
+      "step": 2639
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00037950930382460105,
+      "loss": 1.242,
+      "step": 2640
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00037949079489184165,
+      "loss": 1.2184,
+      "step": 2641
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00037947227805520283,
+      "loss": 1.1917,
+      "step": 2642
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003794537533155,
+      "loss": 1.2261,
+      "step": 2643
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00037943522067354895,
+      "loss": 1.2412,
+      "step": 2644
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003794166801301657,
+      "loss": 1.1541,
+      "step": 2645
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003793981316861667,
+      "loss": 1.2635,
+      "step": 2646
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00037937957534236885,
+      "loss": 1.1965,
+      "step": 2647
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003793610110995891,
+      "loss": 1.2362,
+      "step": 2648
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000379342438958645,
+      "loss": 1.2226,
+      "step": 2649
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037932385892035437,
+      "loss": 1.1858,
+      "step": 2650
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037930527098553536,
+      "loss": 1.159,
+      "step": 2651
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003792866751550066,
+      "loss": 1.2467,
+      "step": 2652
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003792680714295868,
+      "loss": 1.2732,
+      "step": 2653
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037924945981009523,
+      "loss": 1.2174,
+      "step": 2654
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037923084029735154,
+      "loss": 1.1718,
+      "step": 2655
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037921221289217554,
+      "loss": 1.2833,
+      "step": 2656
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003791935775953876,
+      "loss": 1.1708,
+      "step": 2657
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003791749344078081,
+      "loss": 1.1079,
+      "step": 2658
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037915628333025825,
+      "loss": 1.1953,
+      "step": 2659
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003791376243635592,
+      "loss": 1.1801,
+      "step": 2660
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037911895750853265,
+      "loss": 1.271,
+      "step": 2661
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037910028276600055,
+      "loss": 1.1937,
+      "step": 2662
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003790816001367853,
+      "loss": 1.2674,
+      "step": 2663
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003790629096217096,
+      "loss": 1.277,
+      "step": 2664
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037904421122159646,
+      "loss": 1.2152,
+      "step": 2665
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003790255049372692,
+      "loss": 1.1586,
+      "step": 2666
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037900679076955166,
+      "loss": 1.2084,
+      "step": 2667
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003789880687192678,
+      "loss": 1.2674,
+      "step": 2668
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003789693387872421,
+      "loss": 1.2403,
+      "step": 2669
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037895060097429943,
+      "loss": 1.2002,
+      "step": 2670
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003789318552812648,
+      "loss": 1.2186,
+      "step": 2671
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037891310170896366,
+      "loss": 1.1231,
+      "step": 2672
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037889434025822194,
+      "loss": 1.1684,
+      "step": 2673
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037887557092986565,
+      "loss": 1.2724,
+      "step": 2674
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003788567937247215,
+      "loss": 1.1963,
+      "step": 2675
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003788380086436161,
+      "loss": 1.2584,
+      "step": 2676
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003788192156873768,
+      "loss": 1.2083,
+      "step": 2677
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037880041485683106,
+      "loss": 1.2573,
+      "step": 2678
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037878160615280695,
+      "loss": 1.1987,
+      "step": 2679
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003787627895761325,
+      "loss": 1.2047,
+      "step": 2680
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003787439651276364,
+      "loss": 1.1635,
+      "step": 2681
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003787251328081476,
+      "loss": 1.2692,
+      "step": 2682
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037870629261849533,
+      "loss": 1.2005,
+      "step": 2683
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003786874445595094,
+      "loss": 1.2745,
+      "step": 2684
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037866858863201946,
+      "loss": 1.0994,
+      "step": 2685
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000378649724836856,
+      "loss": 1.1609,
+      "step": 2686
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003786308531748498,
+      "loss": 1.1936,
+      "step": 2687
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003786119736468317,
+      "loss": 1.2309,
+      "step": 2688
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037859308625363313,
+      "loss": 1.2462,
+      "step": 2689
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037857419099608577,
+      "loss": 1.2308,
+      "step": 2690
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003785552878750217,
+      "loss": 1.2908,
+      "step": 2691
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037853637689127335,
+      "loss": 1.1735,
+      "step": 2692
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003785174580456734,
+      "loss": 1.2161,
+      "step": 2693
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000378498531339055,
+      "loss": 1.2651,
+      "step": 2694
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037847959677225144,
+      "loss": 1.2672,
+      "step": 2695
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003784606543460968,
+      "loss": 1.162,
+      "step": 2696
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003784417040614249,
+      "loss": 1.2206,
+      "step": 2697
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003784227459190704,
+      "loss": 1.2847,
+      "step": 2698
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037840377991986803,
+      "loss": 1.2234,
+      "step": 2699
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037838480606465307,
+      "loss": 1.2507,
+      "step": 2700
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00037836582435426093,
+      "loss": 1.1398,
+      "step": 2701
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037834683478952743,
+      "loss": 1.183,
+      "step": 2702
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003783278373712889,
+      "loss": 1.205,
+      "step": 2703
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037830883210038187,
+      "loss": 1.179,
+      "step": 2704
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003782898189776432,
+      "loss": 1.2018,
+      "step": 2705
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037827079800391007,
+      "loss": 1.0836,
+      "step": 2706
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037825176918002026,
+      "loss": 1.1312,
+      "step": 2707
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003782327325068115,
+      "loss": 1.198,
+      "step": 2708
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003782136879851221,
+      "loss": 1.1712,
+      "step": 2709
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037819463561579084,
+      "loss": 1.2248,
+      "step": 2710
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037817557539965654,
+      "loss": 1.1882,
+      "step": 2711
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037815650733755856,
+      "loss": 1.2312,
+      "step": 2712
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003781374314303366,
+      "loss": 1.1862,
+      "step": 2713
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037811834767883065,
+      "loss": 1.1701,
+      "step": 2714
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037809925608388095,
+      "loss": 1.1797,
+      "step": 2715
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003780801566463284,
+      "loss": 1.2139,
+      "step": 2716
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003780610493670139,
+      "loss": 1.1711,
+      "step": 2717
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003780419342467789,
+      "loss": 1.1858,
+      "step": 2718
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037802281128646514,
+      "loss": 1.1822,
+      "step": 2719
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003780036804869146,
+      "loss": 1.214,
+      "step": 2720
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037798454184896987,
+      "loss": 1.153,
+      "step": 2721
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037796539537347356,
+      "loss": 1.1843,
+      "step": 2722
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037794624106126886,
+      "loss": 1.2019,
+      "step": 2723
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037792707891319923,
+      "loss": 1.2013,
+      "step": 2724
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003779079089301085,
+      "loss": 1.1859,
+      "step": 2725
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003778887311128408,
+      "loss": 1.1994,
+      "step": 2726
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003778695454622406,
+      "loss": 1.2031,
+      "step": 2727
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037785035197915274,
+      "loss": 1.1817,
+      "step": 2728
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003778311506644225,
+      "loss": 1.1966,
+      "step": 2729
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003778119415188952,
+      "loss": 1.2222,
+      "step": 2730
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037779272454341697,
+      "loss": 1.1933,
+      "step": 2731
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037777349973883384,
+      "loss": 1.1594,
+      "step": 2732
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003777542671059925,
+      "loss": 1.1977,
+      "step": 2733
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037773502664573973,
+      "loss": 1.1997,
+      "step": 2734
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003777157783589229,
+      "loss": 1.3153,
+      "step": 2735
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037769652224638954,
+      "loss": 1.1887,
+      "step": 2736
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003776772583089876,
+      "loss": 1.2522,
+      "step": 2737
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003776579865475655,
+      "loss": 1.1573,
+      "step": 2738
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037763870696297164,
+      "loss": 1.1635,
+      "step": 2739
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037761941955605514,
+      "loss": 1.1738,
+      "step": 2740
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003776001243276653,
+      "loss": 1.1698,
+      "step": 2741
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003775808212786518,
+      "loss": 1.2291,
+      "step": 2742
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037756151040986465,
+      "loss": 1.1991,
+      "step": 2743
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037754219172215413,
+      "loss": 1.2685,
+      "step": 2744
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000377522865216371,
+      "loss": 1.2108,
+      "step": 2745
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003775035308933664,
+      "loss": 1.2109,
+      "step": 2746
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003774841887539915,
+      "loss": 1.2359,
+      "step": 2747
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037746483879909815,
+      "loss": 1.2699,
+      "step": 2748
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037744548102953843,
+      "loss": 1.2229,
+      "step": 2749
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003774261154461648,
+      "loss": 1.1997,
+      "step": 2750
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003774067420498299,
+      "loss": 1.1356,
+      "step": 2751
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00037738736084138696,
+      "loss": 1.2114,
+      "step": 2752
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003773679718216894,
+      "loss": 1.1762,
+      "step": 2753
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000377348574991591,
+      "loss": 1.199,
+      "step": 2754
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00037732917035194585,
+      "loss": 1.1854,
+      "step": 2755
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003773097579036085,
+      "loss": 1.1232,
+      "step": 2756
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003772903376474338,
+      "loss": 1.1335,
+      "step": 2757
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00037727090958427685,
+      "loss": 1.1884,
+      "step": 2758
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00037725147371499323,
+      "loss": 1.2349,
+      "step": 2759
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003772320300404387,
+      "loss": 1.1807,
+      "step": 2760
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00037721257856146955,
+      "loss": 1.2364,
+      "step": 2761
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003771931192789423,
+      "loss": 1.1621,
+      "step": 2762
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00037717365219371387,
+      "loss": 1.1142,
+      "step": 2763
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00037715417730664147,
+      "loss": 1.1948,
+      "step": 2764
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003771346946185827,
+      "loss": 1.1938,
+      "step": 2765
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003771152041303954,
+      "loss": 1.1629,
+      "step": 2766
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003770957058429379,
+      "loss": 1.242,
+      "step": 2767
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003770761997570688,
+      "loss": 1.1962,
+      "step": 2768
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000377056685873647,
+      "loss": 1.1738,
+      "step": 2769
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003770371641935319,
+      "loss": 1.2033,
+      "step": 2770
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00037701763471758307,
+      "loss": 1.1681,
+      "step": 2771
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00037699809744666047,
+      "loss": 1.2435,
+      "step": 2772
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003769785523816245,
+      "loss": 1.1864,
+      "step": 2773
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003769589995233358,
+      "loss": 1.2327,
+      "step": 2774
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00037693943887265525,
+      "loss": 1.1704,
+      "step": 2775
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003769198704304444,
+      "loss": 1.1953,
+      "step": 2776
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00037690029419756486,
+      "loss": 1.1991,
+      "step": 2777
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003768807101748786,
+      "loss": 1.1649,
+      "step": 2778
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00037686111836324817,
+      "loss": 1.2329,
+      "step": 2779
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00037684151876353614,
+      "loss": 1.1797,
+      "step": 2780
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00037682191137660564,
+      "loss": 1.2385,
+      "step": 2781
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00037680229620332004,
+      "loss": 1.1274,
+      "step": 2782
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003767826732445432,
+      "loss": 1.1854,
+      "step": 2783
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003767630425011391,
+      "loss": 1.2893,
+      "step": 2784
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00037674340397397224,
+      "loss": 1.157,
+      "step": 2785
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00037672375766390745,
+      "loss": 1.1876,
+      "step": 2786
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00037670410357180977,
+      "loss": 1.1382,
+      "step": 2787
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00037668444169854467,
+      "loss": 1.2365,
+      "step": 2788
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000376664772044978,
+      "loss": 1.2062,
+      "step": 2789
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003766450946119759,
+      "loss": 1.2566,
+      "step": 2790
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003766254094004048,
+      "loss": 1.1177,
+      "step": 2791
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00037660571641113177,
+      "loss": 1.2249,
+      "step": 2792
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003765860156450237,
+      "loss": 1.1388,
+      "step": 2793
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00037656630710294827,
+      "loss": 1.2115,
+      "step": 2794
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003765465907857733,
+      "loss": 1.2301,
+      "step": 2795
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000376526866694367,
+      "loss": 1.1679,
+      "step": 2796
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00037650713482959794,
+      "loss": 1.276,
+      "step": 2797
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003764873951923351,
+      "loss": 1.2238,
+      "step": 2798
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003764676477834475,
+      "loss": 1.2746,
+      "step": 2799
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003764478926038049,
+      "loss": 1.1376,
+      "step": 2800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00037642812965427716,
+      "loss": 1.2029,
+      "step": 2801
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00037640835893573455,
+      "loss": 1.1906,
+      "step": 2802
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003763885804490476,
+      "loss": 1.0914,
+      "step": 2803
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003763687941950874,
+      "loss": 1.1826,
+      "step": 2804
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00037634900017472514,
+      "loss": 1.2758,
+      "step": 2805
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00037632919838883244,
+      "loss": 1.1561,
+      "step": 2806
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003763093888382813,
+      "loss": 1.2016,
+      "step": 2807
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003762895715239441,
+      "loss": 1.2109,
+      "step": 2808
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003762697464466934,
+      "loss": 1.2943,
+      "step": 2809
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003762499136074022,
+      "loss": 1.2349,
+      "step": 2810
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003762300730069439,
+      "loss": 1.2069,
+      "step": 2811
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00037621022464619203,
+      "loss": 1.2506,
+      "step": 2812
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00037619036852602083,
+      "loss": 1.1291,
+      "step": 2813
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00037617050464730463,
+      "loss": 1.1617,
+      "step": 2814
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00037615063301091794,
+      "loss": 1.2746,
+      "step": 2815
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000376130753617736,
+      "loss": 1.1789,
+      "step": 2816
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003761108664686341,
+      "loss": 1.3353,
+      "step": 2817
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00037609097156448814,
+      "loss": 1.2079,
+      "step": 2818
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003760710689061739,
+      "loss": 1.2071,
+      "step": 2819
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003760511584945681,
+      "loss": 1.2115,
+      "step": 2820
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00037603124033054727,
+      "loss": 1.238,
+      "step": 2821
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00037601131441498855,
+      "loss": 1.1685,
+      "step": 2822
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003759913807487695,
+      "loss": 1.2307,
+      "step": 2823
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00037597143933276777,
+      "loss": 1.1734,
+      "step": 2824
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003759514901678615,
+      "loss": 1.1309,
+      "step": 2825
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00037593153325492923,
+      "loss": 1.1523,
+      "step": 2826
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003759115685948497,
+      "loss": 1.1839,
+      "step": 2827
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00037589159618850205,
+      "loss": 1.1674,
+      "step": 2828
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00037587161603676577,
+      "loss": 1.1716,
+      "step": 2829
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003758516281405207,
+      "loss": 1.1879,
+      "step": 2830
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000375831632500647,
+      "loss": 1.2078,
+      "step": 2831
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003758116291180252,
+      "loss": 1.1467,
+      "step": 2832
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00037579161799353606,
+      "loss": 1.2685,
+      "step": 2833
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003757715991280609,
+      "loss": 1.1687,
+      "step": 2834
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00037575157252248126,
+      "loss": 1.1862,
+      "step": 2835
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003757315381776788,
+      "loss": 1.1722,
+      "step": 2836
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00037571149609453593,
+      "loss": 1.2001,
+      "step": 2837
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003756914462739352,
+      "loss": 1.1393,
+      "step": 2838
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003756713887167594,
+      "loss": 1.2001,
+      "step": 2839
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003756513234238919,
+      "loss": 1.1875,
+      "step": 2840
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00037563125039621615,
+      "loss": 1.2033,
+      "step": 2841
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000375611169634616,
+      "loss": 1.1503,
+      "step": 2842
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00037559108113997597,
+      "loss": 1.1594,
+      "step": 2843
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003755709849131804,
+      "loss": 1.2121,
+      "step": 2844
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003755508809551144,
+      "loss": 1.1735,
+      "step": 2845
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003755307692666632,
+      "loss": 1.1769,
+      "step": 2846
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003755106498487123,
+      "loss": 1.1632,
+      "step": 2847
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003754905227021478,
+      "loss": 1.1471,
+      "step": 2848
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00037547038782785597,
+      "loss": 1.2239,
+      "step": 2849
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003754502452267234,
+      "loss": 1.1328,
+      "step": 2850
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00037543009489963714,
+      "loss": 1.2098,
+      "step": 2851
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00037540993684748446,
+      "loss": 1.2531,
+      "step": 2852
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000375389771071153,
+      "loss": 1.1795,
+      "step": 2853
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00037536959757153085,
+      "loss": 1.1343,
+      "step": 2854
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003753494163495063,
+      "loss": 1.1439,
+      "step": 2855
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00037532922740596796,
+      "loss": 1.1944,
+      "step": 2856
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000375309030741805,
+      "loss": 1.2422,
+      "step": 2857
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00037528882635790657,
+      "loss": 1.1765,
+      "step": 2858
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003752686142551626,
+      "loss": 1.2191,
+      "step": 2859
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000375248394434463,
+      "loss": 1.1939,
+      "step": 2860
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00037522816689669814,
+      "loss": 1.2375,
+      "step": 2861
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00037520793164275883,
+      "loss": 1.196,
+      "step": 2862
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003751876886735361,
+      "loss": 1.2093,
+      "step": 2863
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003751674379899213,
+      "loss": 1.154,
+      "step": 2864
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00037514717959280617,
+      "loss": 1.1764,
+      "step": 2865
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00037512691348308284,
+      "loss": 1.2029,
+      "step": 2866
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00037510663966164364,
+      "loss": 1.1857,
+      "step": 2867
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003750863581293814,
+      "loss": 1.1856,
+      "step": 2868
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00037506606888718926,
+      "loss": 1.152,
+      "step": 2869
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003750457719359606,
+      "loss": 1.2139,
+      "step": 2870
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00037502546727658913,
+      "loss": 1.0936,
+      "step": 2871
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003750051549099691,
+      "loss": 1.1889,
+      "step": 2872
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003749848348369949,
+      "loss": 1.138,
+      "step": 2873
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003749645070585613,
+      "loss": 1.1644,
+      "step": 2874
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003749441715755635,
+      "loss": 1.2217,
+      "step": 2875
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00037492382838889696,
+      "loss": 1.0889,
+      "step": 2876
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00037490347749945734,
+      "loss": 1.2016,
+      "step": 2877
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00037488311890814105,
+      "loss": 1.2529,
+      "step": 2878
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003748627526158444,
+      "loss": 1.178,
+      "step": 2879
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003748423786234643,
+      "loss": 1.1505,
+      "step": 2880
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003748219969318978,
+      "loss": 1.1934,
+      "step": 2881
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003748016075420425,
+      "loss": 1.1906,
+      "step": 2882
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003747812104547963,
+      "loss": 1.1857,
+      "step": 2883
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003747608056710574,
+      "loss": 1.1877,
+      "step": 2884
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00037474039319172417,
+      "loss": 1.0886,
+      "step": 2885
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00037471997301769547,
+      "loss": 1.2534,
+      "step": 2886
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003746995451498707,
+      "loss": 1.2134,
+      "step": 2887
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00037467910958914933,
+      "loss": 1.2306,
+      "step": 2888
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003746586663364311,
+      "loss": 1.1626,
+      "step": 2889
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00037463821539261634,
+      "loss": 1.1447,
+      "step": 2890
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00037461775675860564,
+      "loss": 1.2157,
+      "step": 2891
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003745972904352999,
+      "loss": 1.1555,
+      "step": 2892
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00037457681642360017,
+      "loss": 1.2229,
+      "step": 2893
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00037455633472440825,
+      "loss": 1.2113,
+      "step": 2894
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000374535845338626,
+      "loss": 1.2136,
+      "step": 2895
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003745153482671555,
+      "loss": 1.2043,
+      "step": 2896
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00037449484351089965,
+      "loss": 1.1543,
+      "step": 2897
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00037447433107076116,
+      "loss": 1.1995,
+      "step": 2898
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00037445381094764325,
+      "loss": 1.192,
+      "step": 2899
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00037443328314244967,
+      "loss": 1.2116,
+      "step": 2900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00037441274765608425,
+      "loss": 1.2382,
+      "step": 2901
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00037439220448945134,
+      "loss": 1.1756,
+      "step": 2902
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003743716536434556,
+      "loss": 1.1912,
+      "step": 2903
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003743510951190019,
+      "loss": 1.1699,
+      "step": 2904
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003743305289169955,
+      "loss": 1.2077,
+      "step": 2905
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00037430995503834216,
+      "loss": 1.2027,
+      "step": 2906
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003742893734839478,
+      "loss": 1.2635,
+      "step": 2907
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003742687842547187,
+      "loss": 1.2672,
+      "step": 2908
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003742481873515615,
+      "loss": 1.195,
+      "step": 2909
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003742275827753832,
+      "loss": 1.1796,
+      "step": 2910
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003742069705270912,
+      "loss": 1.2674,
+      "step": 2911
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000374186350607593,
+      "loss": 1.1611,
+      "step": 2912
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00037416572301779676,
+      "loss": 1.179,
+      "step": 2913
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00037414508775861064,
+      "loss": 1.2313,
+      "step": 2914
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003741244448309435,
+      "loss": 1.1342,
+      "step": 2915
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00037410379423570426,
+      "loss": 1.1599,
+      "step": 2916
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003740831359738023,
+      "loss": 1.2241,
+      "step": 2917
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003740624700461473,
+      "loss": 1.1379,
+      "step": 2918
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00037404179645364927,
+      "loss": 1.1938,
+      "step": 2919
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003740211151972185,
+      "loss": 1.2387,
+      "step": 2920
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00037400042627776584,
+      "loss": 1.213,
+      "step": 2921
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00037397972969620215,
+      "loss": 1.14,
+      "step": 2922
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00037395902545343904,
+      "loss": 1.2342,
+      "step": 2923
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000373938313550388,
+      "loss": 1.1587,
+      "step": 2924
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003739175939879612,
+      "loss": 1.2003,
+      "step": 2925
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000373896866767071,
+      "loss": 1.1364,
+      "step": 2926
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00037387613188863005,
+      "loss": 1.2788,
+      "step": 2927
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003738553893535515,
+      "loss": 1.2662,
+      "step": 2928
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00037383463916274874,
+      "loss": 1.1375,
+      "step": 2929
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00037381388131713545,
+      "loss": 1.2846,
+      "step": 2930
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00037379311581762576,
+      "loss": 1.2087,
+      "step": 2931
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003737723426651341,
+      "loss": 1.1844,
+      "step": 2932
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003737515618605751,
+      "loss": 1.1428,
+      "step": 2933
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00037373077340486404,
+      "loss": 1.1864,
+      "step": 2934
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003737099772989161,
+      "loss": 1.2323,
+      "step": 2935
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003736891735436472,
+      "loss": 1.2172,
+      "step": 2936
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003736683621399734,
+      "loss": 1.1891,
+      "step": 2937
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003736475430888111,
+      "loss": 1.2721,
+      "step": 2938
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00037362671639107705,
+      "loss": 1.1707,
+      "step": 2939
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003736058820476884,
+      "loss": 1.145,
+      "step": 2940
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003735850400595626,
+      "loss": 1.1501,
+      "step": 2941
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003735641904276174,
+      "loss": 1.1644,
+      "step": 2942
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00037354333315277094,
+      "loss": 1.2155,
+      "step": 2943
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00037352246823594166,
+      "loss": 1.2681,
+      "step": 2944
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003735015956780483,
+      "loss": 1.2305,
+      "step": 2945
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00037348071548001007,
+      "loss": 1.0791,
+      "step": 2946
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00037345982764274636,
+      "loss": 1.1881,
+      "step": 2947
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00037343893216717697,
+      "loss": 1.2079,
+      "step": 2948
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000373418029054222,
+      "loss": 1.1725,
+      "step": 2949
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000373397118304802,
+      "loss": 1.1804,
+      "step": 2950
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00037337619991983776,
+      "loss": 1.1595,
+      "step": 2951
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003733552739002505,
+      "loss": 1.1382,
+      "step": 2952
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003733343402469615,
+      "loss": 1.2649,
+      "step": 2953
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003733133989608927,
+      "loss": 1.1524,
+      "step": 2954
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003732924500429662,
+      "loss": 1.136,
+      "step": 2955
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003732714934941045,
+      "loss": 1.1935,
+      "step": 2956
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00037325052931523053,
+      "loss": 1.2097,
+      "step": 2957
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003732295575072672,
+      "loss": 1.1984,
+      "step": 2958
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00037320857807113825,
+      "loss": 1.1837,
+      "step": 2959
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00037318759100776744,
+      "loss": 1.1895,
+      "step": 2960
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00037316659631807887,
+      "loss": 1.1451,
+      "step": 2961
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003731455940029971,
+      "loss": 1.1513,
+      "step": 2962
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003731245840634469,
+      "loss": 1.2046,
+      "step": 2963
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003731035665003535,
+      "loss": 1.2167,
+      "step": 2964
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003730825413146425,
+      "loss": 1.2188,
+      "step": 2965
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00037306150850723957,
+      "loss": 1.1708,
+      "step": 2966
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00037304046807907094,
+      "loss": 1.1245,
+      "step": 2967
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003730194200310632,
+      "loss": 1.2241,
+      "step": 2968
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003729983643641431,
+      "loss": 1.2582,
+      "step": 2969
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00037297730107923796,
+      "loss": 1.211,
+      "step": 2970
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00037295623017727516,
+      "loss": 1.1821,
+      "step": 2971
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003729351516591826,
+      "loss": 1.2648,
+      "step": 2972
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00037291406552588855,
+      "loss": 1.1713,
+      "step": 2973
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00037289297177832143,
+      "loss": 1.1836,
+      "step": 2974
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003728718704174102,
+      "loss": 1.2485,
+      "step": 2975
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00037285076144408393,
+      "loss": 1.1977,
+      "step": 2976
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003728296448592723,
+      "loss": 1.209,
+      "step": 2977
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003728085206639051,
+      "loss": 1.2427,
+      "step": 2978
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003727873888589126,
+      "loss": 1.1767,
+      "step": 2979
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00037276624944522523,
+      "loss": 1.2336,
+      "step": 2980
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00037274510242377385,
+      "loss": 1.1697,
+      "step": 2981
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00037272394779548987,
+      "loss": 1.1033,
+      "step": 2982
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00037270278556130463,
+      "loss": 1.1625,
+      "step": 2983
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003726816157221501,
+      "loss": 1.1068,
+      "step": 2984
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003726604382789585,
+      "loss": 1.1614,
+      "step": 2985
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00037263925323266235,
+      "loss": 1.2991,
+      "step": 2986
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003726180605841945,
+      "loss": 1.2439,
+      "step": 2987
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00037259686033448827,
+      "loss": 1.1708,
+      "step": 2988
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00037257565248447714,
+      "loss": 1.1819,
+      "step": 2989
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000372554437035095,
+      "loss": 1.1638,
+      "step": 2990
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00037253321398727607,
+      "loss": 1.1109,
+      "step": 2991
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003725119833419549,
+      "loss": 1.2318,
+      "step": 2992
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003724907451000665,
+      "loss": 1.1927,
+      "step": 2993
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003724694992625459,
+      "loss": 1.1921,
+      "step": 2994
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003724482458303288,
+      "loss": 1.1915,
+      "step": 2995
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00037242698480435105,
+      "loss": 1.1591,
+      "step": 2996
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00037240571618554893,
+      "loss": 1.1514,
+      "step": 2997
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00037238443997485886,
+      "loss": 1.1861,
+      "step": 2998
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003723631561732179,
+      "loss": 1.1484,
+      "step": 2999
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003723418647815632,
+      "loss": 1.218,
+      "step": 3000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003723205658008324,
+      "loss": 1.1807,
+      "step": 3001
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003722992592319633,
+      "loss": 1.2368,
+      "step": 3002
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00037227794507589413,
+      "loss": 1.202,
+      "step": 3003
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003722566233335636,
+      "loss": 1.1962,
+      "step": 3004
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00037223529400591055,
+      "loss": 1.1628,
+      "step": 3005
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000372213957093874,
+      "loss": 1.1662,
+      "step": 3006
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00037219261259839386,
+      "loss": 1.1287,
+      "step": 3007
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003721712605204099,
+      "loss": 1.2222,
+      "step": 3008
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003721499008608623,
+      "loss": 1.1828,
+      "step": 3009
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00037212853362069174,
+      "loss": 1.1957,
+      "step": 3010
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000372107158800839,
+      "loss": 1.2387,
+      "step": 3011
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003720857764022454,
+      "loss": 1.1379,
+      "step": 3012
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037206438642585255,
+      "loss": 1.1993,
+      "step": 3013
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003720429888726022,
+      "loss": 1.1899,
+      "step": 3014
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003720215837434368,
+      "loss": 1.2136,
+      "step": 3015
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003720001710392988,
+      "loss": 1.2021,
+      "step": 3016
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037197875076113113,
+      "loss": 1.1934,
+      "step": 3017
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037195732290987697,
+      "loss": 1.1227,
+      "step": 3018
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037193588748648,
+      "loss": 1.1911,
+      "step": 3019
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037191444449188415,
+      "loss": 1.1958,
+      "step": 3020
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003718929939270336,
+      "loss": 1.2008,
+      "step": 3021
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003718715357928729,
+      "loss": 1.1932,
+      "step": 3022
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037185007009034695,
+      "loss": 1.1946,
+      "step": 3023
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003718285968204011,
+      "loss": 1.1678,
+      "step": 3024
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003718071159839808,
+      "loss": 1.1851,
+      "step": 3025
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037178562758203204,
+      "loss": 1.2344,
+      "step": 3026
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000371764131615501,
+      "loss": 1.1999,
+      "step": 3027
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037174262808533433,
+      "loss": 1.1587,
+      "step": 3028
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037172111699247895,
+      "loss": 1.1482,
+      "step": 3029
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000371699598337882,
+      "loss": 1.2077,
+      "step": 3030
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003716780721224911,
+      "loss": 1.1561,
+      "step": 3031
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037165653834725414,
+      "loss": 1.1999,
+      "step": 3032
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037163499701311944,
+      "loss": 1.2132,
+      "step": 3033
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037161344812103547,
+      "loss": 1.1654,
+      "step": 3034
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037159189167195123,
+      "loss": 1.1954,
+      "step": 3035
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003715703276668158,
+      "loss": 1.2588,
+      "step": 3036
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000371548756106579,
+      "loss": 1.1378,
+      "step": 3037
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037152717699219057,
+      "loss": 1.1943,
+      "step": 3038
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003715055903246007,
+      "loss": 1.2024,
+      "step": 3039
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003714839961047601,
+      "loss": 1.2637,
+      "step": 3040
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037146239433361955,
+      "loss": 1.1866,
+      "step": 3041
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003714407850121304,
+      "loss": 1.1501,
+      "step": 3042
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003714191681412441,
+      "loss": 1.2096,
+      "step": 3043
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037139754372191264,
+      "loss": 1.1695,
+      "step": 3044
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037137591175508816,
+      "loss": 1.1634,
+      "step": 3045
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003713542722417233,
+      "loss": 1.3078,
+      "step": 3046
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037133262518277105,
+      "loss": 1.2015,
+      "step": 3047
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003713109705791844,
+      "loss": 1.325,
+      "step": 3048
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037128930843191707,
+      "loss": 1.057,
+      "step": 3049
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037126763874192285,
+      "loss": 1.1358,
+      "step": 3050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003712459615101561,
+      "loss": 1.1523,
+      "step": 3051
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003712242767375713,
+      "loss": 1.2198,
+      "step": 3052
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037120258442512337,
+      "loss": 1.222,
+      "step": 3053
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003711808845737675,
+      "loss": 1.3094,
+      "step": 3054
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037115917718445925,
+      "loss": 1.2319,
+      "step": 3055
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003711374622581545,
+      "loss": 1.1732,
+      "step": 3056
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003711157397958095,
+      "loss": 1.185,
+      "step": 3057
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037109400979838073,
+      "loss": 1.2168,
+      "step": 3058
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037107227226682515,
+      "loss": 1.1945,
+      "step": 3059
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003710505272020999,
+      "loss": 1.1816,
+      "step": 3060
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003710287746051626,
+      "loss": 1.1962,
+      "step": 3061
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037100701447697113,
+      "loss": 1.1643,
+      "step": 3062
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037098524681848357,
+      "loss": 1.0995,
+      "step": 3063
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037096347163065855,
+      "loss": 1.1883,
+      "step": 3064
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037094168891445495,
+      "loss": 1.1129,
+      "step": 3065
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037091989867083197,
+      "loss": 1.219,
+      "step": 3066
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037089810090074917,
+      "loss": 1.1751,
+      "step": 3067
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003708762956051663,
+      "loss": 1.1481,
+      "step": 3068
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003708544827850437,
+      "loss": 1.1669,
+      "step": 3069
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003708326624413418,
+      "loss": 1.1925,
+      "step": 3070
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037081083457502146,
+      "loss": 1.1595,
+      "step": 3071
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037078899918704396,
+      "loss": 1.1766,
+      "step": 3072
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003707671562783707,
+      "loss": 1.2576,
+      "step": 3073
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037074530584996364,
+      "loss": 1.2304,
+      "step": 3074
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037072344790278495,
+      "loss": 1.2704,
+      "step": 3075
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037070158243779697,
+      "loss": 1.1609,
+      "step": 3076
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037067970945596284,
+      "loss": 1.1833,
+      "step": 3077
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003706578289582455,
+      "loss": 1.207,
+      "step": 3078
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003706359409456086,
+      "loss": 1.2798,
+      "step": 3079
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003706140454190159,
+      "loss": 1.1316,
+      "step": 3080
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037059214237943153,
+      "loss": 1.204,
+      "step": 3081
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003705702318278202,
+      "loss": 1.216,
+      "step": 3082
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037054831376514646,
+      "loss": 1.2011,
+      "step": 3083
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037052638819237563,
+      "loss": 1.0763,
+      "step": 3084
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037050445511047325,
+      "loss": 1.2352,
+      "step": 3085
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000370482514520405,
+      "loss": 1.2621,
+      "step": 3086
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037046056642313715,
+      "loss": 1.2409,
+      "step": 3087
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003704386108196361,
+      "loss": 1.1527,
+      "step": 3088
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037041664771086876,
+      "loss": 1.1358,
+      "step": 3089
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037039467709780216,
+      "loss": 1.1668,
+      "step": 3090
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003703726989814039,
+      "loss": 1.2593,
+      "step": 3091
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037035071336264167,
+      "loss": 1.2323,
+      "step": 3092
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037032872024248377,
+      "loss": 1.1738,
+      "step": 3093
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003703067196218984,
+      "loss": 1.211,
+      "step": 3094
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037028471150185463,
+      "loss": 1.1812,
+      "step": 3095
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037026269588332145,
+      "loss": 1.2226,
+      "step": 3096
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037024067276726836,
+      "loss": 1.1116,
+      "step": 3097
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003702186421546651,
+      "loss": 1.1327,
+      "step": 3098
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003701966040464818,
+      "loss": 1.2152,
+      "step": 3099
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000370174558443689,
+      "loss": 1.1445,
+      "step": 3100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037015250534725737,
+      "loss": 1.1633,
+      "step": 3101
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000370130444758158,
+      "loss": 1.1653,
+      "step": 3102
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003701083766773625,
+      "loss": 1.2125,
+      "step": 3103
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037008630110584246,
+      "loss": 1.192,
+      "step": 3104
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037006421804457006,
+      "loss": 1.0832,
+      "step": 3105
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037004212749451766,
+      "loss": 1.2646,
+      "step": 3106
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003700200294566581,
+      "loss": 1.1258,
+      "step": 3107
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003699979239319644,
+      "loss": 1.2061,
+      "step": 3108
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003699758109214101,
+      "loss": 1.1763,
+      "step": 3109
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003699536904259688,
+      "loss": 1.2244,
+      "step": 3110
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003699315624466146,
+      "loss": 1.2052,
+      "step": 3111
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036990942698432195,
+      "loss": 1.1925,
+      "step": 3112
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036988728404006567,
+      "loss": 1.1536,
+      "step": 3113
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00036986513361482063,
+      "loss": 1.2592,
+      "step": 3114
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003698429757095623,
+      "loss": 1.2013,
+      "step": 3115
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00036982081032526645,
+      "loss": 1.1019,
+      "step": 3116
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003697986374629092,
+      "loss": 1.1669,
+      "step": 3117
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00036977645712346673,
+      "loss": 1.1736,
+      "step": 3118
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003697542693079159,
+      "loss": 1.1667,
+      "step": 3119
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00036973207401723375,
+      "loss": 1.218,
+      "step": 3120
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00036970987125239756,
+      "loss": 1.228,
+      "step": 3121
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003696876610143851,
+      "loss": 1.2098,
+      "step": 3122
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00036966544330417443,
+      "loss": 1.245,
+      "step": 3123
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003696432181227438,
+      "loss": 1.2052,
+      "step": 3124
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00036962098547107196,
+      "loss": 1.1925,
+      "step": 3125
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00036959874535013795,
+      "loss": 1.1361,
+      "step": 3126
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000369576497760921,
+      "loss": 1.2342,
+      "step": 3127
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000369554242704401,
+      "loss": 1.181,
+      "step": 3128
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003695319801815577,
+      "loss": 1.2531,
+      "step": 3129
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00036950971019337163,
+      "loss": 1.1765,
+      "step": 3130
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00036948743274082337,
+      "loss": 1.1898,
+      "step": 3131
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003694651478248939,
+      "loss": 1.1713,
+      "step": 3132
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00036944285544656443,
+      "loss": 1.2561,
+      "step": 3133
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00036942055560681687,
+      "loss": 1.2323,
+      "step": 3134
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00036939824830663296,
+      "loss": 1.1349,
+      "step": 3135
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00036937593354699504,
+      "loss": 1.1563,
+      "step": 3136
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003693536113288859,
+      "loss": 1.2309,
+      "step": 3137
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00036933128165328834,
+      "loss": 1.1731,
+      "step": 3138
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00036930894452118565,
+      "loss": 1.2227,
+      "step": 3139
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003692865999335615,
+      "loss": 1.1133,
+      "step": 3140
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00036926424789139985,
+      "loss": 1.1258,
+      "step": 3141
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003692418883956849,
+      "loss": 1.1694,
+      "step": 3142
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003692195214474014,
+      "loss": 1.1722,
+      "step": 3143
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000369197147047534,
+      "loss": 1.0955,
+      "step": 3144
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003691747651970683,
+      "loss": 1.1719,
+      "step": 3145
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003691523758969896,
+      "loss": 1.1725,
+      "step": 3146
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00036912997914828404,
+      "loss": 1.1429,
+      "step": 3147
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003691075749519376,
+      "loss": 1.2139,
+      "step": 3148
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00036908516330893713,
+      "loss": 1.2281,
+      "step": 3149
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00036906274422026937,
+      "loss": 1.2217,
+      "step": 3150
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00036904031768692153,
+      "loss": 1.2048,
+      "step": 3151
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003690178837098812,
+      "loss": 1.1751,
+      "step": 3152
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00036899544229013626,
+      "loss": 1.1678,
+      "step": 3153
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00036897299342867495,
+      "loss": 1.2795,
+      "step": 3154
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00036895053712648575,
+      "loss": 1.0869,
+      "step": 3155
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003689280733845576,
+      "loss": 1.1657,
+      "step": 3156
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00036890560220387964,
+      "loss": 1.2228,
+      "step": 3157
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003688831235854413,
+      "loss": 1.1965,
+      "step": 3158
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00036886063753023264,
+      "loss": 1.1552,
+      "step": 3159
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003688381440392436,
+      "loss": 1.0928,
+      "step": 3160
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003688156431134648,
+      "loss": 1.1925,
+      "step": 3161
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00036879313475388716,
+      "loss": 1.1307,
+      "step": 3162
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00036877061896150166,
+      "loss": 1.286,
+      "step": 3163
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00036874809573729994,
+      "loss": 1.1283,
+      "step": 3164
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00036872556508227365,
+      "loss": 1.1945,
+      "step": 3165
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000368703026997415,
+      "loss": 1.16,
+      "step": 3166
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00036868048148371653,
+      "loss": 1.1693,
+      "step": 3167
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003686579285421709,
+      "loss": 1.2145,
+      "step": 3168
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00036863536817377133,
+      "loss": 1.1601,
+      "step": 3169
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003686128003795112,
+      "loss": 1.218,
+      "step": 3170
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00036859022516038434,
+      "loss": 1.1883,
+      "step": 3171
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003685676425173848,
+      "loss": 1.174,
+      "step": 3172
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003685450524515071,
+      "loss": 1.1274,
+      "step": 3173
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003685224549637458,
+      "loss": 1.1775,
+      "step": 3174
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003684998500550962,
+      "loss": 1.1146,
+      "step": 3175
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003684772377265536,
+      "loss": 1.2282,
+      "step": 3176
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003684546179791137,
+      "loss": 1.1937,
+      "step": 3177
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003684319908137726,
+      "loss": 1.2534,
+      "step": 3178
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003684093562315268,
+      "loss": 1.154,
+      "step": 3179
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003683867142333728,
+      "loss": 1.2205,
+      "step": 3180
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003683640648203078,
+      "loss": 1.1169,
+      "step": 3181
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003683414079933291,
+      "loss": 1.17,
+      "step": 3182
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003683187437534344,
+      "loss": 1.1221,
+      "step": 3183
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00036829607210162176,
+      "loss": 1.1992,
+      "step": 3184
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003682733930388895,
+      "loss": 1.2078,
+      "step": 3185
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00036825070656623627,
+      "loss": 1.2041,
+      "step": 3186
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00036822801268466106,
+      "loss": 1.2565,
+      "step": 3187
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00036820531139516327,
+      "loss": 1.155,
+      "step": 3188
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003681826026987425,
+      "loss": 1.1676,
+      "step": 3189
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003681598865963987,
+      "loss": 1.1706,
+      "step": 3190
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003681371630891322,
+      "loss": 1.1742,
+      "step": 3191
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00036811443217794367,
+      "loss": 1.2562,
+      "step": 3192
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00036809169386383403,
+      "loss": 1.207,
+      "step": 3193
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00036806894814780456,
+      "loss": 1.1484,
+      "step": 3194
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00036804619503085694,
+      "loss": 1.2035,
+      "step": 3195
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003680234345139929,
+      "loss": 1.1121,
+      "step": 3196
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00036800066659821494,
+      "loss": 1.1878,
+      "step": 3197
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00036797789128452553,
+      "loss": 1.1866,
+      "step": 3198
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00036795510857392757,
+      "loss": 1.233,
+      "step": 3199
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00036793231846742427,
+      "loss": 1.1965,
+      "step": 3200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003679095209660193,
+      "loss": 1.2195,
+      "step": 3201
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003678867160707165,
+      "loss": 1.2683,
+      "step": 3202
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00036786390378252,
+      "loss": 1.138,
+      "step": 3203
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003678410841024345,
+      "loss": 1.2331,
+      "step": 3204
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00036781825703146475,
+      "loss": 1.1062,
+      "step": 3205
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000367795422570616,
+      "loss": 1.1644,
+      "step": 3206
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00036777258072089367,
+      "loss": 1.1168,
+      "step": 3207
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003677497314833037,
+      "loss": 1.2211,
+      "step": 3208
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00036772687485885227,
+      "loss": 1.1933,
+      "step": 3209
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00036770401084854575,
+      "loss": 1.1457,
+      "step": 3210
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000367681139453391,
+      "loss": 1.1261,
+      "step": 3211
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003676582606743953,
+      "loss": 1.1948,
+      "step": 3212
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000367635374512566,
+      "loss": 1.1723,
+      "step": 3213
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003676124809689109,
+      "loss": 1.1903,
+      "step": 3214
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00036758958004443797,
+      "loss": 1.1526,
+      "step": 3215
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00036756667174015596,
+      "loss": 1.2701,
+      "step": 3216
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00036754375605707346,
+      "loss": 1.1791,
+      "step": 3217
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003675208329961995,
+      "loss": 1.2385,
+      "step": 3218
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003674979025585437,
+      "loss": 1.1432,
+      "step": 3219
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003674749647451156,
+      "loss": 1.1323,
+      "step": 3220
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00036745201955692535,
+      "loss": 1.1466,
+      "step": 3221
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00036742906699498335,
+      "loss": 1.1968,
+      "step": 3222
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003674061070603004,
+      "loss": 1.1978,
+      "step": 3223
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00036738313975388727,
+      "loss": 1.2206,
+      "step": 3224
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00036736016507675566,
+      "loss": 1.1749,
+      "step": 3225
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00036733718302991704,
+      "loss": 1.2392,
+      "step": 3226
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00036731419361438346,
+      "loss": 1.2161,
+      "step": 3227
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00036729119683116736,
+      "loss": 1.1661,
+      "step": 3228
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003672681926812813,
+      "loss": 1.1453,
+      "step": 3229
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003672451811657383,
+      "loss": 1.2071,
+      "step": 3230
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00036722216228555164,
+      "loss": 1.085,
+      "step": 3231
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003671991360417351,
+      "loss": 1.1847,
+      "step": 3232
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00036717610243530236,
+      "loss": 1.1301,
+      "step": 3233
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00036715306146726796,
+      "loss": 1.1551,
+      "step": 3234
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00036713001313864644,
+      "loss": 1.1586,
+      "step": 3235
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003671069574504527,
+      "loss": 1.2014,
+      "step": 3236
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000367083894403702,
+      "loss": 1.1216,
+      "step": 3237
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00036706082399940994,
+      "loss": 1.1691,
+      "step": 3238
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003670377462385925,
+      "loss": 1.1901,
+      "step": 3239
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003670146611222657,
+      "loss": 1.2411,
+      "step": 3240
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003669915686514463,
+      "loss": 1.1949,
+      "step": 3241
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003669684688271511,
+      "loss": 1.1529,
+      "step": 3242
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00036694536165039727,
+      "loss": 1.1376,
+      "step": 3243
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00036692224712220236,
+      "loss": 1.2045,
+      "step": 3244
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00036689912524358434,
+      "loss": 1.1642,
+      "step": 3245
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00036687599601556117,
+      "loss": 1.0981,
+      "step": 3246
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00036685285943915144,
+      "loss": 1.091,
+      "step": 3247
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000366829715515374,
+      "loss": 1.2169,
+      "step": 3248
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003668065642452479,
+      "loss": 1.1617,
+      "step": 3249
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003667834056297928,
+      "loss": 1.2004,
+      "step": 3250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003667602396700283,
+      "loss": 1.2691,
+      "step": 3251
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00036673706636697453,
+      "loss": 1.1431,
+      "step": 3252
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00036671388572165207,
+      "loss": 1.1458,
+      "step": 3253
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00036669069773508155,
+      "loss": 1.1872,
+      "step": 3254
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000366667502408284,
+      "loss": 1.2679,
+      "step": 3255
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000366644299742281,
+      "loss": 1.1984,
+      "step": 3256
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00036662108973809415,
+      "loss": 1.1714,
+      "step": 3257
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003665978723967456,
+      "loss": 1.2846,
+      "step": 3258
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00036657464771925767,
+      "loss": 1.1935,
+      "step": 3259
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00036655141570665303,
+      "loss": 1.1325,
+      "step": 3260
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00036652817635995477,
+      "loss": 1.2534,
+      "step": 3261
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003665049296801861,
+      "loss": 1.2029,
+      "step": 3262
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003664816756683709,
+      "loss": 1.1727,
+      "step": 3263
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000366458414325533,
+      "loss": 1.24,
+      "step": 3264
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00036643514565269674,
+      "loss": 1.1473,
+      "step": 3265
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003664118696508868,
+      "loss": 1.2248,
+      "step": 3266
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003663885863211281,
+      "loss": 1.1041,
+      "step": 3267
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003663652956644459,
+      "loss": 1.237,
+      "step": 3268
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003663419976818659,
+      "loss": 1.2093,
+      "step": 3269
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00036631869237441396,
+      "loss": 1.108,
+      "step": 3270
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003662953797431163,
+      "loss": 1.203,
+      "step": 3271
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00036627205978899955,
+      "loss": 1.2483,
+      "step": 3272
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003662487325130906,
+      "loss": 1.1299,
+      "step": 3273
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00036622539791641664,
+      "loss": 1.2234,
+      "step": 3274
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003662020560000052,
+      "loss": 1.169,
+      "step": 3275
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003661787067648842,
+      "loss": 1.1458,
+      "step": 3276
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003661553502120818,
+      "loss": 1.16,
+      "step": 3277
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003661319863426264,
+      "loss": 1.1588,
+      "step": 3278
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000366108615157547,
+      "loss": 1.0715,
+      "step": 3279
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00036608523665787265,
+      "loss": 1.1465,
+      "step": 3280
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003660618508446328,
+      "loss": 1.1748,
+      "step": 3281
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00036603845771885734,
+      "loss": 1.1701,
+      "step": 3282
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00036601505728157637,
+      "loss": 1.1023,
+      "step": 3283
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003659916495338202,
+      "loss": 1.1521,
+      "step": 3284
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003659682344766197,
+      "loss": 1.1913,
+      "step": 3285
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00036594481211100604,
+      "loss": 1.1782,
+      "step": 3286
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003659213824380104,
+      "loss": 1.1615,
+      "step": 3287
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00036589794545866466,
+      "loss": 1.1635,
+      "step": 3288
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003658745011740009,
+      "loss": 1.1964,
+      "step": 3289
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00036585104958505133,
+      "loss": 1.2154,
+      "step": 3290
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003658275906928488,
+      "loss": 1.1204,
+      "step": 3291
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00036580412449842617,
+      "loss": 1.2073,
+      "step": 3292
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00036578065100281693,
+      "loss": 1.1755,
+      "step": 3293
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00036575717020705464,
+      "loss": 1.1804,
+      "step": 3294
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00036573368211217336,
+      "loss": 1.1971,
+      "step": 3295
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003657101867192072,
+      "loss": 1.1006,
+      "step": 3296
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000365686684029191,
+      "loss": 1.2019,
+      "step": 3297
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003656631740431596,
+      "loss": 1.1393,
+      "step": 3298
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003656396567621483,
+      "loss": 1.2352,
+      "step": 3299
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00036561613218719265,
+      "loss": 1.184,
+      "step": 3300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00036559260031932857,
+      "loss": 1.1332,
+      "step": 3301
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00036556906115959225,
+      "loss": 1.2201,
+      "step": 3302
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003655455147090203,
+      "loss": 1.2059,
+      "step": 3303
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003655219609686495,
+      "loss": 1.1489,
+      "step": 3304
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003654983999395171,
+      "loss": 1.1274,
+      "step": 3305
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003654748316226607,
+      "loss": 1.2374,
+      "step": 3306
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00036545125601911794,
+      "loss": 1.2113,
+      "step": 3307
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00036542767312992716,
+      "loss": 1.2329,
+      "step": 3308
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003654040829561266,
+      "loss": 1.2161,
+      "step": 3309
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003653804854987553,
+      "loss": 1.1891,
+      "step": 3310
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00036535688075885225,
+      "loss": 1.2037,
+      "step": 3311
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003653332687374569,
+      "loss": 1.1512,
+      "step": 3312
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000365309649435609,
+      "loss": 1.1807,
+      "step": 3313
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003652860228543486,
+      "loss": 1.1905,
+      "step": 3314
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003652623889947162,
+      "loss": 1.1207,
+      "step": 3315
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00036523874785775244,
+      "loss": 1.1813,
+      "step": 3316
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003652150994444983,
+      "loss": 1.1268,
+      "step": 3317
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003651914437559953,
+      "loss": 1.1461,
+      "step": 3318
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000365167780793285,
+      "loss": 1.2223,
+      "step": 3319
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0003651441105574094,
+      "loss": 1.1807,
+      "step": 3320
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00036512043304941087,
+      "loss": 1.1772,
+      "step": 3321
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00036509674827033196,
+      "loss": 1.0883,
+      "step": 3322
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0003650730562212157,
+      "loss": 1.1221,
+      "step": 3323
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0003650493569031054,
+      "loss": 1.1973,
+      "step": 3324
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0003650256503170447,
+      "loss": 1.1763,
+      "step": 3325
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0003650019364640773,
+      "loss": 1.2327,
+      "step": 3326
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00036497821534524756,
+      "loss": 1.2236,
+      "step": 3327
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00036495448696160024,
+      "loss": 1.2098,
+      "step": 3328
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0003649307513141799,
+      "loss": 1.2512,
+      "step": 3329
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0003649070084040319,
+      "loss": 1.18,
+      "step": 3330
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0003648832582322017,
+      "loss": 1.1667,
+      "step": 3331
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00036485950079973527,
+      "loss": 1.1862,
+      "step": 3332
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00036483573610767865,
+      "loss": 1.173,
+      "step": 3333
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0003648119641570783,
+      "loss": 1.1837,
+      "step": 3334
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0003647881849489811,
+      "loss": 1.1367,
+      "step": 3335
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0003647643984844341,
+      "loss": 1.2023,
+      "step": 3336
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0003647406047644848,
+      "loss": 1.2294,
+      "step": 3337
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0003647168037901809,
+      "loss": 1.1373,
+      "step": 3338
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0003646929955625705,
+      "loss": 1.1646,
+      "step": 3339
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00036466918008270203,
+      "loss": 1.1769,
+      "step": 3340
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0003646453573516242,
+      "loss": 1.2484,
+      "step": 3341
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00036462152737038594,
+      "loss": 1.1764,
+      "step": 3342
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00036459769014003673,
+      "loss": 1.2015,
+      "step": 3343
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0003645738456616261,
+      "loss": 1.2011,
+      "step": 3344
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00036454999393620426,
+      "loss": 1.1311,
+      "step": 3345
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00036452613496482133,
+      "loss": 1.179,
+      "step": 3346
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000364502268748528,
+      "loss": 1.1866,
+      "step": 3347
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0003644783952883752,
+      "loss": 1.1698,
+      "step": 3348
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0003644545145854143,
+      "loss": 1.2505,
+      "step": 3349
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0003644306266406968,
+      "loss": 1.1981,
+      "step": 3350
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00036440673145527456,
+      "loss": 1.2121,
+      "step": 3351
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00036438282903019986,
+      "loss": 1.1419,
+      "step": 3352
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0003643589193665253,
+      "loss": 1.1578,
+      "step": 3353
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00036433500246530365,
+      "loss": 1.1221,
+      "step": 3354
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00036431107832758815,
+      "loss": 1.2048,
+      "step": 3355
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0003642871469544322,
+      "loss": 1.2082,
+      "step": 3356
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0003642632083468898,
+      "loss": 1.2224,
+      "step": 3357
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0003642392625060149,
+      "loss": 1.176,
+      "step": 3358
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00036421530943286213,
+      "loss": 1.0941,
+      "step": 3359
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0003641913491284861,
+      "loss": 1.1836,
+      "step": 3360
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000364167381593942,
+      "loss": 1.1548,
+      "step": 3361
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00036414340683028516,
+      "loss": 1.3106,
+      "step": 3362
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00036411942483857143,
+      "loss": 1.1838,
+      "step": 3363
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00036409543561985675,
+      "loss": 1.1506,
+      "step": 3364
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00036407143917519756,
+      "loss": 1.2717,
+      "step": 3365
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00036404743550565047,
+      "loss": 1.1599,
+      "step": 3366
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00036402342461227253,
+      "loss": 1.1831,
+      "step": 3367
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0003639994064961211,
+      "loss": 1.0989,
+      "step": 3368
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0003639753811582537,
+      "loss": 1.2123,
+      "step": 3369
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00036395134859972846,
+      "loss": 1.2439,
+      "step": 3370
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00036392730882160344,
+      "loss": 1.2022,
+      "step": 3371
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0003639032618249374,
+      "loss": 1.1774,
+      "step": 3372
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00036387920761078913,
+      "loss": 1.1379,
+      "step": 3373
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000363855146180218,
+      "loss": 1.0979,
+      "step": 3374
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0003638310775342834,
+      "loss": 1.2014,
+      "step": 3375
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00036380700167404535,
+      "loss": 1.1305,
+      "step": 3376
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0003637829186005639,
+      "loss": 1.1564,
+      "step": 3377
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0003637588283148996,
+      "loss": 1.1869,
+      "step": 3378
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00036373473081811327,
+      "loss": 1.1213,
+      "step": 3379
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00036371062611126597,
+      "loss": 1.1888,
+      "step": 3380
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00036368651419541926,
+      "loss": 1.1966,
+      "step": 3381
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00036366239507163497,
+      "loss": 1.1817,
+      "step": 3382
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00036363826874097495,
+      "loss": 1.1946,
+      "step": 3383
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00036361413520450173,
+      "loss": 1.1686,
+      "step": 3384
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0003635899944632781,
+      "loss": 1.0965,
+      "step": 3385
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00036356584651836706,
+      "loss": 1.1519,
+      "step": 3386
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0003635416913708319,
+      "loss": 1.1703,
+      "step": 3387
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00036351752902173635,
+      "loss": 1.0744,
+      "step": 3388
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0003634933594721444,
+      "loss": 1.2017,
+      "step": 3389
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0003634691827231202,
+      "loss": 1.217,
+      "step": 3390
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00036344499877572867,
+      "loss": 1.2413,
+      "step": 3391
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0003634208076310345,
+      "loss": 1.1664,
+      "step": 3392
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00036339660929010306,
+      "loss": 1.1448,
+      "step": 3393
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00036337240375399995,
+      "loss": 1.2064,
+      "step": 3394
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00036334819102379096,
+      "loss": 1.2381,
+      "step": 3395
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00036332397110054235,
+      "loss": 1.2767,
+      "step": 3396
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00036329974398532065,
+      "loss": 1.1714,
+      "step": 3397
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0003632755096791927,
+      "loss": 1.1177,
+      "step": 3398
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00036325126818322566,
+      "loss": 1.2383,
+      "step": 3399
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000363227019498487,
+      "loss": 1.2458,
+      "step": 3400
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00036320276362604453,
+      "loss": 1.1942,
+      "step": 3401
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0003631785005669663,
+      "loss": 1.1502,
+      "step": 3402
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0003631542303223208,
+      "loss": 1.1331,
+      "step": 3403
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0003631299528931767,
+      "loss": 1.2082,
+      "step": 3404
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0003631056682806031,
+      "loss": 1.1814,
+      "step": 3405
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0003630813764856694,
+      "loss": 1.1966,
+      "step": 3406
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00036305707750944533,
+      "loss": 1.1982,
+      "step": 3407
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0003630327713530008,
+      "loss": 1.2255,
+      "step": 3408
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0003630084580174061,
+      "loss": 1.2003,
+      "step": 3409
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00036298413750373193,
+      "loss": 1.1976,
+      "step": 3410
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0003629598098130493,
+      "loss": 1.1617,
+      "step": 3411
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00036293547494642935,
+      "loss": 1.1236,
+      "step": 3412
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00036291113290494383,
+      "loss": 1.146,
+      "step": 3413
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0003628867836896645,
+      "loss": 1.167,
+      "step": 3414
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00036286242730166365,
+      "loss": 1.1891,
+      "step": 3415
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0003628380637420137,
+      "loss": 1.2357,
+      "step": 3416
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00036281369301178774,
+      "loss": 1.1852,
+      "step": 3417
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00036278931511205865,
+      "loss": 1.1918,
+      "step": 3418
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0003627649300439002,
+      "loss": 1.2115,
+      "step": 3419
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0003627405378083859,
+      "loss": 1.1519,
+      "step": 3420
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00036271613840659,
+      "loss": 1.2318,
+      "step": 3421
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000362691731839587,
+      "loss": 1.137,
+      "step": 3422
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0003626673181084516,
+      "loss": 1.1006,
+      "step": 3423
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0003626428972142587,
+      "loss": 1.1619,
+      "step": 3424
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0003626184691580839,
+      "loss": 1.2268,
+      "step": 3425
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0003625940339410027,
+      "loss": 1.2378,
+      "step": 3426
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0003625695915640913,
+      "loss": 1.206,
+      "step": 3427
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0003625451420284259,
+      "loss": 1.2076,
+      "step": 3428
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00036252068533508314,
+      "loss": 1.1697,
+      "step": 3429
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00036249622148514,
+      "loss": 1.1426,
+      "step": 3430
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00036247175047967364,
+      "loss": 1.1698,
+      "step": 3431
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00036244727231976185,
+      "loss": 1.2274,
+      "step": 3432
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00036242278700648233,
+      "loss": 1.0841,
+      "step": 3433
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0003623982945409134,
+      "loss": 1.1379,
+      "step": 3434
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00036237379492413355,
+      "loss": 1.1841,
+      "step": 3435
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00036234928815722165,
+      "loss": 1.2144,
+      "step": 3436
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0003623247742412568,
+      "loss": 1.1897,
+      "step": 3437
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00036230025317731854,
+      "loss": 1.2087,
+      "step": 3438
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0003622757249664866,
+      "loss": 1.2401,
+      "step": 3439
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000362251189609841,
+      "loss": 1.1965,
+      "step": 3440
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00036222664710846235,
+      "loss": 1.1507,
+      "step": 3441
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0003622020974634313,
+      "loss": 1.1185,
+      "step": 3442
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0003621775406758288,
+      "loss": 1.22,
+      "step": 3443
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00036215297674673636,
+      "loss": 1.1476,
+      "step": 3444
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00036212840567723554,
+      "loss": 1.1522,
+      "step": 3445
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00036210382746840837,
+      "loss": 1.1947,
+      "step": 3446
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0003620792421213371,
+      "loss": 1.1649,
+      "step": 3447
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0003620546496371045,
+      "loss": 1.2574,
+      "step": 3448
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00036203005001679333,
+      "loss": 1.1469,
+      "step": 3449
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00036200544326148686,
+      "loss": 1.1729,
+      "step": 3450
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0003619808293722687,
+      "loss": 1.1561,
+      "step": 3451
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0003619562083502228,
+      "loss": 1.1436,
+      "step": 3452
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0003619315801964331,
+      "loss": 1.2707,
+      "step": 3453
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0003619069449119844,
+      "loss": 1.1827,
+      "step": 3454
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0003618823024979613,
+      "loss": 1.1847,
+      "step": 3455
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00036185765295544903,
+      "loss": 1.1626,
+      "step": 3456
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00036183299628553294,
+      "loss": 1.1446,
+      "step": 3457
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0003618083324892989,
+      "loss": 1.1423,
+      "step": 3458
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0003617836615678329,
+      "loss": 1.19,
+      "step": 3459
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0003617589835222214,
+      "loss": 1.1106,
+      "step": 3460
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000361734298353551,
+      "loss": 1.2466,
+      "step": 3461
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0003617096060629089,
+      "loss": 1.129,
+      "step": 3462
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00036168490665138215,
+      "loss": 1.1231,
+      "step": 3463
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0003616602001200585,
+      "loss": 1.204,
+      "step": 3464
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000361635486470026,
+      "loss": 1.1739,
+      "step": 3465
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0003616107657023729,
+      "loss": 1.2615,
+      "step": 3466
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00036158603781818766,
+      "loss": 1.1759,
+      "step": 3467
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0003615613028185592,
+      "loss": 1.1552,
+      "step": 3468
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00036153656070457693,
+      "loss": 1.1501,
+      "step": 3469
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00036151181147733006,
+      "loss": 1.1876,
+      "step": 3470
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0003614870551379086,
+      "loss": 1.0954,
+      "step": 3471
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00036146229168740264,
+      "loss": 1.1277,
+      "step": 3472
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00036143752112690274,
+      "loss": 1.1402,
+      "step": 3473
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0003614127434574995,
+      "loss": 1.2164,
+      "step": 3474
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0003613879586802841,
+      "loss": 1.2741,
+      "step": 3475
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000361363166796348,
+      "loss": 1.1437,
+      "step": 3476
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00036133836780678286,
+      "loss": 1.1319,
+      "step": 3477
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00036131356171268064,
+      "loss": 1.1623,
+      "step": 3478
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00036128874851513375,
+      "loss": 1.1135,
+      "step": 3479
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0003612639282152348,
+      "loss": 1.1471,
+      "step": 3480
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0003612391008140768,
+      "loss": 1.1112,
+      "step": 3481
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00036121426631275294,
+      "loss": 1.1893,
+      "step": 3482
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0003611894247123569,
+      "loss": 1.1641,
+      "step": 3483
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0003611645760139825,
+      "loss": 1.1523,
+      "step": 3484
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0003611397202187241,
+      "loss": 1.2009,
+      "step": 3485
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000361114857327676,
+      "loss": 1.2419,
+      "step": 3486
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00036108998734193324,
+      "loss": 1.0977,
+      "step": 3487
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0003610651102625909,
+      "loss": 1.2083,
+      "step": 3488
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00036104022609074435,
+      "loss": 1.1345,
+      "step": 3489
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0003610153348274895,
+      "loss": 1.1653,
+      "step": 3490
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0003609904364739223,
+      "loss": 1.1762,
+      "step": 3491
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00036096553103113935,
+      "loss": 1.1652,
+      "step": 3492
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00036094061850023716,
+      "loss": 1.1609,
+      "step": 3493
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00036091569888231287,
+      "loss": 1.1221,
+      "step": 3494
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00036089077217846385,
+      "loss": 1.1733,
+      "step": 3495
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00036086583838978753,
+      "loss": 1.0689,
+      "step": 3496
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00036084089751738216,
+      "loss": 1.1784,
+      "step": 3497
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00036081594956234586,
+      "loss": 1.1671,
+      "step": 3498
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0003607909945257771,
+      "loss": 1.1376,
+      "step": 3499
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00036076603240877504,
+      "loss": 1.1524,
+      "step": 3500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0003607410632124387,
+      "loss": 1.1178,
+      "step": 3501
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0003607160869378676,
+      "loss": 1.1883,
+      "step": 3502
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0003606911035861617,
+      "loss": 1.1039,
+      "step": 3503
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00036066611315842105,
+      "loss": 1.1134,
+      "step": 3504
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000360641115655746,
+      "loss": 1.19,
+      "step": 3505
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0003606161110792375,
+      "loss": 1.1708,
+      "step": 3506
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0003605910994299966,
+      "loss": 1.1493,
+      "step": 3507
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0003605660807091246,
+      "loss": 1.1775,
+      "step": 3508
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0003605410549177233,
+      "loss": 1.1224,
+      "step": 3509
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0003605160220568945,
+      "loss": 1.2532,
+      "step": 3510
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00036049098212774083,
+      "loss": 1.1862,
+      "step": 3511
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00036046593513136473,
+      "loss": 1.2535,
+      "step": 3512
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0003604408810688692,
+      "loss": 1.1474,
+      "step": 3513
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00036041581994135743,
+      "loss": 1.1222,
+      "step": 3514
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000360390751749933,
+      "loss": 1.1731,
+      "step": 3515
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0003603656764956999,
+      "loss": 1.1561,
+      "step": 3516
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00036034059417976226,
+      "loss": 1.1921,
+      "step": 3517
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00036031550480322455,
+      "loss": 1.2218,
+      "step": 3518
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0003602904083671916,
+      "loss": 1.1558,
+      "step": 3519
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00036026530487276854,
+      "loss": 1.1351,
+      "step": 3520
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0003602401943210607,
+      "loss": 1.1633,
+      "step": 3521
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00036021507671317405,
+      "loss": 1.1627,
+      "step": 3522
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0003601899520502144,
+      "loss": 1.2005,
+      "step": 3523
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0003601648203332883,
+      "loss": 1.0967,
+      "step": 3524
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0003601396815635023,
+      "loss": 1.2033,
+      "step": 3525
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0003601145357419634,
+      "loss": 1.142,
+      "step": 3526
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000360089382869779,
+      "loss": 1.116,
+      "step": 3527
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00036006422294805666,
+      "loss": 1.2288,
+      "step": 3528
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0003600390559779042,
+      "loss": 1.1811,
+      "step": 3529
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00036001388196042994,
+      "loss": 1.2024,
+      "step": 3530
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00035998870089674244,
+      "loss": 1.1182,
+      "step": 3531
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0003599635127879505,
+      "loss": 1.0754,
+      "step": 3532
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0003599383176351633,
+      "loss": 1.1048,
+      "step": 3533
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0003599131154394903,
+      "loss": 1.1331,
+      "step": 3534
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00035988790620204123,
+      "loss": 1.1776,
+      "step": 3535
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0003598626899239262,
+      "loss": 1.1509,
+      "step": 3536
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00035983746660625575,
+      "loss": 1.0838,
+      "step": 3537
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0003598122362501404,
+      "loss": 1.2127,
+      "step": 3538
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0003597869988566912,
+      "loss": 1.225,
+      "step": 3539
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0003597617544270196,
+      "loss": 1.1789,
+      "step": 3540
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00035973650296223717,
+      "loss": 1.1909,
+      "step": 3541
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0003597112444634558,
+      "loss": 1.1246,
+      "step": 3542
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00035968597893178784,
+      "loss": 1.1924,
+      "step": 3543
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0003596607063683458,
+      "loss": 1.1343,
+      "step": 3544
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00035963542677424254,
+      "loss": 1.1701,
+      "step": 3545
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00035961014015059134,
+      "loss": 1.2871,
+      "step": 3546
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0003595848464985056,
+      "loss": 1.2089,
+      "step": 3547
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00035955954581909914,
+      "loss": 1.2002,
+      "step": 3548
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0003595342381134861,
+      "loss": 1.2006,
+      "step": 3549
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00035950892338278096,
+      "loss": 1.1475,
+      "step": 3550
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0003594836016280984,
+      "loss": 1.2114,
+      "step": 3551
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0003594582728505534,
+      "loss": 1.1346,
+      "step": 3552
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0003594329370512615,
+      "loss": 1.1898,
+      "step": 3553
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0003594075942313381,
+      "loss": 1.1334,
+      "step": 3554
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00035938224439189935,
+      "loss": 1.2141,
+      "step": 3555
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00035935688753406146,
+      "loss": 1.1693,
+      "step": 3556
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00035933152365894114,
+      "loss": 1.1377,
+      "step": 3557
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0003593061527676552,
+      "loss": 1.2499,
+      "step": 3558
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00035928077486132074,
+      "loss": 1.0683,
+      "step": 3559
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0003592553899410554,
+      "loss": 1.2339,
+      "step": 3560
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00035922999800797707,
+      "loss": 1.1928,
+      "step": 3561
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0003592045990632038,
+      "loss": 1.1288,
+      "step": 3562
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0003591791931078539,
+      "loss": 1.2464,
+      "step": 3563
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00035915378014304643,
+      "loss": 1.202,
+      "step": 3564
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00035912836016990015,
+      "loss": 1.1863,
+      "step": 3565
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0003591029331895346,
+      "loss": 1.1443,
+      "step": 3566
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0003590774992030694,
+      "loss": 1.2099,
+      "step": 3567
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0003590520582116246,
+      "loss": 1.1292,
+      "step": 3568
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00035902661021632043,
+      "loss": 1.2343,
+      "step": 3569
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0003590011552182774,
+      "loss": 1.1912,
+      "step": 3570
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0003589756932186166,
+      "loss": 1.1794,
+      "step": 3571
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0003589502242184592,
+      "loss": 1.2735,
+      "step": 3572
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0003589247482189267,
+      "loss": 1.1403,
+      "step": 3573
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00035889926522114093,
+      "loss": 1.1606,
+      "step": 3574
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00035887377522622404,
+      "loss": 1.1908,
+      "step": 3575
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00035884827823529845,
+      "loss": 1.06,
+      "step": 3576
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000358822774249487,
+      "loss": 1.214,
+      "step": 3577
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0003587972632699127,
+      "loss": 1.16,
+      "step": 3578
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000358771745297699,
+      "loss": 1.2112,
+      "step": 3579
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0003587462203339694,
+      "loss": 1.0974,
+      "step": 3580
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0003587206883798481,
+      "loss": 1.1371,
+      "step": 3581
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0003586951494364593,
+      "loss": 1.2374,
+      "step": 3582
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00035866960350492766,
+      "loss": 1.2051,
+      "step": 3583
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00035864405058637805,
+      "loss": 1.2201,
+      "step": 3584
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0003586184906819357,
+      "loss": 1.171,
+      "step": 3585
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00035859292379272615,
+      "loss": 1.094,
+      "step": 3586
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00035856734991987515,
+      "loss": 1.1174,
+      "step": 3587
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00035854176906450905,
+      "loss": 1.2106,
+      "step": 3588
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0003585161812277541,
+      "loss": 1.1867,
+      "step": 3589
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0003584905864107372,
+      "loss": 1.1987,
+      "step": 3590
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0003584649846145853,
+      "loss": 1.1468,
+      "step": 3591
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0003584393758404259,
+      "loss": 1.1601,
+      "step": 3592
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0003584137600893866,
+      "loss": 1.1183,
+      "step": 3593
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0003583881373625953,
+      "loss": 1.1837,
+      "step": 3594
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00035836250766118054,
+      "loss": 1.151,
+      "step": 3595
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00035833687098627074,
+      "loss": 1.1361,
+      "step": 3596
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0003583112273389949,
+      "loss": 1.1079,
+      "step": 3597
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00035828557672048215,
+      "loss": 1.1397,
+      "step": 3598
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0003582599191318621,
+      "loss": 1.2499,
+      "step": 3599
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00035823425457426453,
+      "loss": 1.1181,
+      "step": 3600
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0003582085830488196,
+      "loss": 1.1702,
+      "step": 3601
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00035818290455665773,
+      "loss": 1.1538,
+      "step": 3602
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0003581572190989097,
+      "loss": 1.1371,
+      "step": 3603
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0003581315266767066,
+      "loss": 1.1814,
+      "step": 3604
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0003581058272911797,
+      "loss": 1.1775,
+      "step": 3605
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0003580801209434608,
+      "loss": 1.1092,
+      "step": 3606
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00035805440763468177,
+      "loss": 1.1842,
+      "step": 3607
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000358028687365975,
+      "loss": 1.193,
+      "step": 3608
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00035800296013847296,
+      "loss": 1.1336,
+      "step": 3609
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0003579772259533086,
+      "loss": 1.195,
+      "step": 3610
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00035795148481161513,
+      "loss": 1.2044,
+      "step": 3611
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0003579257367145261,
+      "loss": 1.1672,
+      "step": 3612
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0003578999816631753,
+      "loss": 1.1871,
+      "step": 3613
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0003578742196586968,
+      "loss": 1.2921,
+      "step": 3614
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00035784845070222513,
+      "loss": 1.2586,
+      "step": 3615
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000357822674794895,
+      "loss": 1.1591,
+      "step": 3616
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00035779689193784136,
+      "loss": 1.1448,
+      "step": 3617
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00035777110213219967,
+      "loss": 1.1203,
+      "step": 3618
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0003577453053791055,
+      "loss": 1.1862,
+      "step": 3619
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00035771950167969485,
+      "loss": 1.2181,
+      "step": 3620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000357693691035104,
+      "loss": 1.286,
+      "step": 3621
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00035766787344646955,
+      "loss": 1.1268,
+      "step": 3622
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0003576420489149283,
+      "loss": 1.2037,
+      "step": 3623
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00035761621744161745,
+      "loss": 1.1361,
+      "step": 3624
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00035759037902767456,
+      "loss": 1.2141,
+      "step": 3625
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00035756453367423734,
+      "loss": 1.1056,
+      "step": 3626
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0003575386813824439,
+      "loss": 1.1082,
+      "step": 3627
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00035751282215343267,
+      "loss": 1.1466,
+      "step": 3628
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00035748695598834243,
+      "loss": 1.2289,
+      "step": 3629
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003574610828883121,
+      "loss": 1.1341,
+      "step": 3630
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000357435202854481,
+      "loss": 1.2118,
+      "step": 3631
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00035740931588798883,
+      "loss": 1.1498,
+      "step": 3632
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003573834219899754,
+      "loss": 1.1174,
+      "step": 3633
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003573575211615811,
+      "loss": 1.1544,
+      "step": 3634
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003573316134039464,
+      "loss": 1.1267,
+      "step": 3635
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003573056987182121,
+      "loss": 1.1784,
+      "step": 3636
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003572797771055195,
+      "loss": 1.1991,
+      "step": 3637
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00035725384856700993,
+      "loss": 1.225,
+      "step": 3638
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00035722791310382514,
+      "loss": 1.155,
+      "step": 3639
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003572019707171073,
+      "loss": 1.1661,
+      "step": 3640
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003571760214079987,
+      "loss": 1.2272,
+      "step": 3641
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000357150065177642,
+      "loss": 1.1641,
+      "step": 3642
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00035712410202718033,
+      "loss": 1.1689,
+      "step": 3643
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00035709813195775676,
+      "loss": 1.1751,
+      "step": 3644
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003570721549705151,
+      "loss": 1.1307,
+      "step": 3645
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003570461710665991,
+      "loss": 1.1985,
+      "step": 3646
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000357020180247153,
+      "loss": 1.1227,
+      "step": 3647
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00035699418251332137,
+      "loss": 1.1831,
+      "step": 3648
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00035696817786624895,
+      "loss": 1.1392,
+      "step": 3649
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00035694216630708086,
+      "loss": 1.2166,
+      "step": 3650
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003569161478369625,
+      "loss": 1.1913,
+      "step": 3651
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00035689012245703963,
+      "loss": 1.1341,
+      "step": 3652
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00035686409016845823,
+      "loss": 1.1146,
+      "step": 3653
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003568380509723648,
+      "loss": 1.1774,
+      "step": 3654
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003568120048699057,
+      "loss": 1.2468,
+      "step": 3655
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003567859518622281,
+      "loss": 1.2387,
+      "step": 3656
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003567598919504792,
+      "loss": 1.1227,
+      "step": 3657
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00035673382513580644,
+      "loss": 1.0647,
+      "step": 3658
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003567077514193578,
+      "loss": 1.1836,
+      "step": 3659
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003566816708022813,
+      "loss": 1.1736,
+      "step": 3660
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00035665558328572554,
+      "loss": 1.1167,
+      "step": 3661
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003566294888708392,
+      "loss": 1.1639,
+      "step": 3662
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00035660338755877147,
+      "loss": 1.0933,
+      "step": 3663
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00035657727935067157,
+      "loss": 1.2167,
+      "step": 3664
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00035655116424768916,
+      "loss": 1.1635,
+      "step": 3665
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00035652504225097437,
+      "loss": 1.1059,
+      "step": 3666
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003564989133616774,
+      "loss": 1.1638,
+      "step": 3667
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003564727775809488,
+      "loss": 1.1755,
+      "step": 3668
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00035644663490993946,
+      "loss": 1.2345,
+      "step": 3669
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003564204853498007,
+      "loss": 1.1689,
+      "step": 3670
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003563943289016839,
+      "loss": 1.082,
+      "step": 3671
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00035636816556674085,
+      "loss": 1.1994,
+      "step": 3672
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003563419953461237,
+      "loss": 1.145,
+      "step": 3673
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003563158182409849,
+      "loss": 1.1717,
+      "step": 3674
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000356289634252477,
+      "loss": 1.1228,
+      "step": 3675
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003562634433817532,
+      "loss": 1.2067,
+      "step": 3676
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003562372456299667,
+      "loss": 1.1492,
+      "step": 3677
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003562110409982712,
+      "loss": 1.1286,
+      "step": 3678
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003561848294878205,
+      "loss": 1.1993,
+      "step": 3679
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003561586110997689,
+      "loss": 1.1316,
+      "step": 3680
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00035613238583527096,
+      "loss": 1.1527,
+      "step": 3681
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00035610615369548144,
+      "loss": 1.2107,
+      "step": 3682
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003560799146815555,
+      "loss": 1.1711,
+      "step": 3683
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003560536687946487,
+      "loss": 1.1536,
+      "step": 3684
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003560274160359165,
+      "loss": 1.1916,
+      "step": 3685
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003560011564065151,
+      "loss": 1.2103,
+      "step": 3686
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003559748899076009,
+      "loss": 1.2061,
+      "step": 3687
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00035594861654033054,
+      "loss": 1.0838,
+      "step": 3688
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00035592233630586087,
+      "loss": 1.1844,
+      "step": 3689
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003558960492053492,
+      "loss": 1.1577,
+      "step": 3690
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000355869755239953,
+      "loss": 1.109,
+      "step": 3691
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00035584345441083025,
+      "loss": 1.1589,
+      "step": 3692
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00035581714671913903,
+      "loss": 1.1235,
+      "step": 3693
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003557908321660379,
+      "loss": 1.1499,
+      "step": 3694
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00035576451075268543,
+      "loss": 1.1312,
+      "step": 3695
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00035573818248024085,
+      "loss": 1.189,
+      "step": 3696
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003557118473498635,
+      "loss": 1.1406,
+      "step": 3697
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00035568550536271295,
+      "loss": 1.1078,
+      "step": 3698
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00035565915651994927,
+      "loss": 1.2055,
+      "step": 3699
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003556328008227327,
+      "loss": 1.0779,
+      "step": 3700
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003556064382722238,
+      "loss": 1.1674,
+      "step": 3701
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003555800688695835,
+      "loss": 1.2454,
+      "step": 3702
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003555536926159729,
+      "loss": 1.1643,
+      "step": 3703
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00035552730951255357,
+      "loss": 1.1488,
+      "step": 3704
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003555009195604872,
+      "loss": 1.0987,
+      "step": 3705
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003554745227609359,
+      "loss": 1.1394,
+      "step": 3706
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000355448119115062,
+      "loss": 1.2452,
+      "step": 3707
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00035542170862402835,
+      "loss": 1.1282,
+      "step": 3708
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003553952912889978,
+      "loss": 1.2085,
+      "step": 3709
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00035536886711113366,
+      "loss": 1.1733,
+      "step": 3710
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003553424360915995,
+      "loss": 1.1036,
+      "step": 3711
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00035531599823155924,
+      "loss": 1.1965,
+      "step": 3712
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00035528955353217713,
+      "loss": 1.1095,
+      "step": 3713
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00035526310199461754,
+      "loss": 1.1102,
+      "step": 3714
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003552366436200454,
+      "loss": 1.0992,
+      "step": 3715
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003552101784096257,
+      "loss": 1.1962,
+      "step": 3716
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00035518370636452386,
+      "loss": 1.1887,
+      "step": 3717
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003551572274859056,
+      "loss": 1.1229,
+      "step": 3718
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003551307417749368,
+      "loss": 1.131,
+      "step": 3719
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00035510424923278395,
+      "loss": 1.1526,
+      "step": 3720
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00035507774986061363,
+      "loss": 1.1686,
+      "step": 3721
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003550512436595925,
+      "loss": 1.1671,
+      "step": 3722
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003550247306308881,
+      "loss": 1.2349,
+      "step": 3723
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003549982107756677,
+      "loss": 1.1224,
+      "step": 3724
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00035497168409509913,
+      "loss": 1.1494,
+      "step": 3725
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00035494515059035053,
+      "loss": 1.1818,
+      "step": 3726
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003549186102625903,
+      "loss": 1.1765,
+      "step": 3727
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003548920631129872,
+      "loss": 1.1293,
+      "step": 3728
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003548655091427102,
+      "loss": 1.2451,
+      "step": 3729
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00035483894835292853,
+      "loss": 1.2103,
+      "step": 3730
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003548123807448118,
+      "loss": 1.1227,
+      "step": 3731
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035478580631953006,
+      "loss": 1.1328,
+      "step": 3732
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035475922507825335,
+      "loss": 1.1784,
+      "step": 3733
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003547326370221522,
+      "loss": 1.1232,
+      "step": 3734
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035470604215239755,
+      "loss": 1.1336,
+      "step": 3735
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003546794404701604,
+      "loss": 1.1387,
+      "step": 3736
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003546528319766122,
+      "loss": 1.2859,
+      "step": 3737
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035462621667292453,
+      "loss": 1.213,
+      "step": 3738
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035459959456026955,
+      "loss": 1.165,
+      "step": 3739
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035457296563981945,
+      "loss": 1.1535,
+      "step": 3740
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035454632991274693,
+      "loss": 1.1755,
+      "step": 3741
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035451968738022487,
+      "loss": 1.1441,
+      "step": 3742
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003544930380434265,
+      "loss": 1.0576,
+      "step": 3743
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003544663819035252,
+      "loss": 1.173,
+      "step": 3744
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003544397189616949,
+      "loss": 1.0754,
+      "step": 3745
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003544130492191096,
+      "loss": 1.159,
+      "step": 3746
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035438637267694377,
+      "loss": 1.152,
+      "step": 3747
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035435968933637216,
+      "loss": 1.2014,
+      "step": 3748
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003543329991985697,
+      "loss": 1.1377,
+      "step": 3749
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035430630226471167,
+      "loss": 1.1879,
+      "step": 3750
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003542795985359738,
+      "loss": 1.1635,
+      "step": 3751
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035425288801353174,
+      "loss": 1.1581,
+      "step": 3752
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035422617069856197,
+      "loss": 1.0533,
+      "step": 3753
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003541994465922408,
+      "loss": 1.0807,
+      "step": 3754
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035417271569574516,
+      "loss": 1.1643,
+      "step": 3755
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000354145978010252,
+      "loss": 1.2168,
+      "step": 3756
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035411923353693877,
+      "loss": 1.1985,
+      "step": 3757
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035409248227698324,
+      "loss": 1.1266,
+      "step": 3758
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003540657242315633,
+      "loss": 1.1595,
+      "step": 3759
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035403895940185726,
+      "loss": 1.2411,
+      "step": 3760
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035401218778904377,
+      "loss": 1.157,
+      "step": 3761
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003539854093943017,
+      "loss": 1.1928,
+      "step": 3762
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003539586242188102,
+      "loss": 1.1522,
+      "step": 3763
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035393183226374873,
+      "loss": 1.1584,
+      "step": 3764
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035390503353029714,
+      "loss": 1.1364,
+      "step": 3765
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035387822801963547,
+      "loss": 1.1436,
+      "step": 3766
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035385141573294413,
+      "loss": 1.177,
+      "step": 3767
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035382459667140383,
+      "loss": 1.1263,
+      "step": 3768
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003537977708361955,
+      "loss": 1.1312,
+      "step": 3769
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035377093822850037,
+      "loss": 1.1763,
+      "step": 3770
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035374409884950006,
+      "loss": 1.156,
+      "step": 3771
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035371725270037655,
+      "loss": 1.1971,
+      "step": 3772
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035369039978231184,
+      "loss": 1.1536,
+      "step": 3773
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035366354009648847,
+      "loss": 1.173,
+      "step": 3774
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003536366736440892,
+      "loss": 1.2142,
+      "step": 3775
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035360980042629714,
+      "loss": 1.2452,
+      "step": 3776
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035358292044429556,
+      "loss": 1.174,
+      "step": 3777
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003535560336992682,
+      "loss": 1.1332,
+      "step": 3778
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035352914019239897,
+      "loss": 1.1389,
+      "step": 3779
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003535022399248722,
+      "loss": 1.1556,
+      "step": 3780
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003534753328978723,
+      "loss": 1.2275,
+      "step": 3781
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00035344841911258423,
+      "loss": 1.1915,
+      "step": 3782
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00035342149857019324,
+      "loss": 1.1599,
+      "step": 3783
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003533945712718845,
+      "loss": 1.2035,
+      "step": 3784
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000353367637218844,
+      "loss": 1.0781,
+      "step": 3785
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00035334069641225767,
+      "loss": 1.1558,
+      "step": 3786
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003533137488533119,
+      "loss": 1.1369,
+      "step": 3787
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00035328679454319325,
+      "loss": 1.1655,
+      "step": 3788
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00035325983348308867,
+      "loss": 1.0861,
+      "step": 3789
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003532328656741855,
+      "loss": 1.1412,
+      "step": 3790
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003532058911176711,
+      "loss": 1.1454,
+      "step": 3791
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003531789098147335,
+      "loss": 1.194,
+      "step": 3792
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00035315192176656066,
+      "loss": 1.1728,
+      "step": 3793
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000353124926974341,
+      "loss": 1.2048,
+      "step": 3794
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00035309792543926333,
+      "loss": 1.2363,
+      "step": 3795
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003530709171625166,
+      "loss": 1.1206,
+      "step": 3796
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00035304390214529015,
+      "loss": 1.1771,
+      "step": 3797
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003530168803887735,
+      "loss": 1.135,
+      "step": 3798
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003529898518941567,
+      "loss": 1.1182,
+      "step": 3799
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003529628166626298,
+      "loss": 1.1565,
+      "step": 3800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00035293577469538344,
+      "loss": 1.1136,
+      "step": 3801
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003529087259936083,
+      "loss": 1.1419,
+      "step": 3802
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003528816705584955,
+      "loss": 1.2232,
+      "step": 3803
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003528546083912365,
+      "loss": 1.1804,
+      "step": 3804
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003528275394930229,
+      "loss": 1.1438,
+      "step": 3805
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003528004638650467,
+      "loss": 1.1894,
+      "step": 3806
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003527733815085002,
+      "loss": 1.1404,
+      "step": 3807
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003527462924245759,
+      "loss": 1.2759,
+      "step": 3808
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003527191966144667,
+      "loss": 1.153,
+      "step": 3809
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003526920940793658,
+      "loss": 1.1388,
+      "step": 3810
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003526649848204667,
+      "loss": 1.1689,
+      "step": 3811
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00035263786883896306,
+      "loss": 1.1411,
+      "step": 3812
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000352610746136049,
+      "loss": 1.1184,
+      "step": 3813
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003525836167129188,
+      "loss": 1.1476,
+      "step": 3814
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00035255648057076717,
+      "loss": 1.1936,
+      "step": 3815
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00035252933771078894,
+      "loss": 1.1251,
+      "step": 3816
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00035250218813417955,
+      "loss": 1.1928,
+      "step": 3817
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003524750318421344,
+      "loss": 1.1753,
+      "step": 3818
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003524478688358492,
+      "loss": 1.1132,
+      "step": 3819
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00035242069911652036,
+      "loss": 1.1811,
+      "step": 3820
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003523935226853441,
+      "loss": 1.2249,
+      "step": 3821
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003523663395435171,
+      "loss": 1.1854,
+      "step": 3822
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003523391496922366,
+      "loss": 1.1147,
+      "step": 3823
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003523119531326996,
+      "loss": 1.1912,
+      "step": 3824
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00035228474986610396,
+      "loss": 1.1417,
+      "step": 3825
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003522575398936474,
+      "loss": 1.1575,
+      "step": 3826
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003522303232165282,
+      "loss": 1.1454,
+      "step": 3827
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00035220309983594484,
+      "loss": 1.2554,
+      "step": 3828
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003521758697530961,
+      "loss": 1.1107,
+      "step": 3829
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000352148632969181,
+      "loss": 1.133,
+      "step": 3830
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00035212138948539896,
+      "loss": 1.2265,
+      "step": 3831
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003520941393029497,
+      "loss": 1.1047,
+      "step": 3832
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003520668824230331,
+      "loss": 1.1744,
+      "step": 3833
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003520396188468495,
+      "loss": 1.085,
+      "step": 3834
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003520123485755993,
+      "loss": 1.2127,
+      "step": 3835
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003519850716104835,
+      "loss": 1.1136,
+      "step": 3836
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00035195778795270316,
+      "loss": 1.1212,
+      "step": 3837
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00035193049760345977,
+      "loss": 1.248,
+      "step": 3838
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003519032005639551,
+      "loss": 1.1335,
+      "step": 3839
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003518758968353909,
+      "loss": 1.1503,
+      "step": 3840
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003518485864189699,
+      "loss": 1.2056,
+      "step": 3841
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00035182126931589445,
+      "loss": 1.1922,
+      "step": 3842
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003517939455273675,
+      "loss": 1.1291,
+      "step": 3843
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00035176661505459226,
+      "loss": 1.271,
+      "step": 3844
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00035173927789877225,
+      "loss": 1.1229,
+      "step": 3845
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003517119340611113,
+      "loss": 1.1572,
+      "step": 3846
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003516845835428134,
+      "loss": 1.1629,
+      "step": 3847
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00035165722634508304,
+      "loss": 1.0774,
+      "step": 3848
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003516298624691248,
+      "loss": 1.1829,
+      "step": 3849
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003516024919161437,
+      "loss": 1.209,
+      "step": 3850
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00035157511468734493,
+      "loss": 1.1505,
+      "step": 3851
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00035154773078393414,
+      "loss": 1.1589,
+      "step": 3852
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00035152034020711717,
+      "loss": 1.2246,
+      "step": 3853
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003514929429581002,
+      "loss": 1.1289,
+      "step": 3854
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003514655390380894,
+      "loss": 1.1458,
+      "step": 3855
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00035143812844829186,
+      "loss": 1.2281,
+      "step": 3856
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00035141071118991444,
+      "loss": 1.1251,
+      "step": 3857
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003513832872641645,
+      "loss": 1.1623,
+      "step": 3858
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00035135585667224955,
+      "loss": 1.1754,
+      "step": 3859
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003513284194153776,
+      "loss": 1.1123,
+      "step": 3860
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003513009754947569,
+      "loss": 1.1798,
+      "step": 3861
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003512735249115957,
+      "loss": 1.1547,
+      "step": 3862
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00035124606766710314,
+      "loss": 1.1052,
+      "step": 3863
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000351218603762488,
+      "loss": 1.2007,
+      "step": 3864
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00035119113319895983,
+      "loss": 1.1351,
+      "step": 3865
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003511636559777282,
+      "loss": 1.2526,
+      "step": 3866
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003511361721000031,
+      "loss": 1.1701,
+      "step": 3867
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003511086815669947,
+      "loss": 1.1987,
+      "step": 3868
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003510811843799138,
+      "loss": 1.2307,
+      "step": 3869
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000351053680539971,
+      "loss": 1.1139,
+      "step": 3870
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003510261700483775,
+      "loss": 1.148,
+      "step": 3871
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003509986529063447,
+      "loss": 1.1437,
+      "step": 3872
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00035097112911508434,
+      "loss": 1.1008,
+      "step": 3873
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003509435986758085,
+      "loss": 1.1613,
+      "step": 3874
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003509160615897294,
+      "loss": 1.1925,
+      "step": 3875
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003508885178580596,
+      "loss": 1.1025,
+      "step": 3876
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003508609674820121,
+      "loss": 1.1964,
+      "step": 3877
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00035083341046279997,
+      "loss": 1.1444,
+      "step": 3878
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003508058468016368,
+      "loss": 1.1801,
+      "step": 3879
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00035077827649973624,
+      "loss": 1.0672,
+      "step": 3880
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00035075069955831247,
+      "loss": 1.0891,
+      "step": 3881
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003507231159785797,
+      "loss": 1.1649,
+      "step": 3882
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003506955257617527,
+      "loss": 1.2125,
+      "step": 3883
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00035066792890904634,
+      "loss": 1.1425,
+      "step": 3884
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00035064032542167587,
+      "loss": 1.1266,
+      "step": 3885
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0003506127153008568,
+      "loss": 1.1546,
+      "step": 3886
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00035058509854780495,
+      "loss": 1.1225,
+      "step": 3887
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00035055747516373636,
+      "loss": 1.1006,
+      "step": 3888
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0003505298451498675,
+      "loss": 1.0933,
+      "step": 3889
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0003505022085074151,
+      "loss": 1.1654,
+      "step": 3890
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000350474565237596,
+      "loss": 1.1617,
+      "step": 3891
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0003504469153416276,
+      "loss": 1.149,
+      "step": 3892
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0003504192588207274,
+      "loss": 1.1296,
+      "step": 3893
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0003503915956761133,
+      "loss": 1.1068,
+      "step": 3894
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00035036392590900335,
+      "loss": 1.1708,
+      "step": 3895
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00035033624952061617,
+      "loss": 1.1254,
+      "step": 3896
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0003503085665121703,
+      "loss": 1.14,
+      "step": 3897
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00035028087688488487,
+      "loss": 1.1095,
+      "step": 3898
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00035025318063997907,
+      "loss": 1.1056,
+      "step": 3899
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00035022547777867276,
+      "loss": 1.125,
+      "step": 3900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00035019776830218555,
+      "loss": 1.1707,
+      "step": 3901
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00035017005221173777,
+      "loss": 1.1636,
+      "step": 3902
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00035014232950854985,
+      "loss": 1.1611,
+      "step": 3903
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00035011460019384265,
+      "loss": 1.1164,
+      "step": 3904
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0003500868642688371,
+      "loss": 1.2028,
+      "step": 3905
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00035005912173475466,
+      "loss": 1.1781,
+      "step": 3906
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00035003137259281695,
+      "loss": 1.1296,
+      "step": 3907
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0003500036168442459,
+      "loss": 1.0813,
+      "step": 3908
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0003499758544902637,
+      "loss": 1.1158,
+      "step": 3909
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00034994808553209285,
+      "loss": 1.0931,
+      "step": 3910
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0003499203099709562,
+      "loss": 1.1469,
+      "step": 3911
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00034989252780807684,
+      "loss": 1.0781,
+      "step": 3912
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0003498647390446782,
+      "loss": 1.1432,
+      "step": 3913
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0003498369436819839,
+      "loss": 1.1705,
+      "step": 3914
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0003498091417212179,
+      "loss": 1.2131,
+      "step": 3915
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0003497813331636045,
+      "loss": 1.0586,
+      "step": 3916
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0003497535180103682,
+      "loss": 1.091,
+      "step": 3917
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000349725696262734,
+      "loss": 1.1478,
+      "step": 3918
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0003496978679219268,
+      "loss": 1.1336,
+      "step": 3919
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00034967003298917216,
+      "loss": 1.154,
+      "step": 3920
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0003496421914656957,
+      "loss": 1.1164,
+      "step": 3921
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0003496143433527236,
+      "loss": 1.1165,
+      "step": 3922
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0003495864886514819,
+      "loss": 1.1584,
+      "step": 3923
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0003495586273631974,
+      "loss": 1.133,
+      "step": 3924
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00034953075948909685,
+      "loss": 1.196,
+      "step": 3925
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0003495028850304075,
+      "loss": 1.1874,
+      "step": 3926
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0003494750039883567,
+      "loss": 1.153,
+      "step": 3927
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0003494471163641722,
+      "loss": 1.2065,
+      "step": 3928
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0003494192221590822,
+      "loss": 1.1426,
+      "step": 3929
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00034939132137431486,
+      "loss": 1.0869,
+      "step": 3930
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0003493634140110988,
+      "loss": 1.0933,
+      "step": 3931
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000349335500070663,
+      "loss": 1.1414,
+      "step": 3932
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0003493075795542365,
+      "loss": 1.2653,
+      "step": 3933
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00034927965246304895,
+      "loss": 1.1993,
+      "step": 3934
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00034925171879833007,
+      "loss": 1.082,
+      "step": 3935
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00034922377856130984,
+      "loss": 1.1544,
+      "step": 3936
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00034919583175321866,
+      "loss": 1.2745,
+      "step": 3937
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00034916787837528727,
+      "loss": 1.1739,
+      "step": 3938
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0003491399184287465,
+      "loss": 1.1777,
+      "step": 3939
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0003491119519148275,
+      "loss": 1.1521,
+      "step": 3940
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00034908397883476187,
+      "loss": 1.1136,
+      "step": 3941
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0003490559991897814,
+      "loss": 1.1109,
+      "step": 3942
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0003490280129811181,
+      "loss": 1.1137,
+      "step": 3943
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0003490000202100045,
+      "loss": 1.0898,
+      "step": 3944
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00034897202087767304,
+      "loss": 1.112,
+      "step": 3945
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0003489440149853569,
+      "loss": 1.1593,
+      "step": 3946
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00034891600253428914,
+      "loss": 1.1548,
+      "step": 3947
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00034888798352570335,
+      "loss": 1.1921,
+      "step": 3948
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0003488599579608334,
+      "loss": 1.0506,
+      "step": 3949
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0003488319258409134,
+      "loss": 1.1937,
+      "step": 3950
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0003488038871671776,
+      "loss": 1.1194,
+      "step": 3951
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00034877584194086084,
+      "loss": 1.1206,
+      "step": 3952
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000348747790163198,
+      "loss": 1.213,
+      "step": 3953
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0003487197318354244,
+      "loss": 1.1908,
+      "step": 3954
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0003486916669587756,
+      "loss": 1.0901,
+      "step": 3955
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0003486635955344873,
+      "loss": 1.1534,
+      "step": 3956
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00034863551756379583,
+      "loss": 1.1898,
+      "step": 3957
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0003486074330479374,
+      "loss": 1.233,
+      "step": 3958
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0003485793419881489,
+      "loss": 1.2133,
+      "step": 3959
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0003485512443856672,
+      "loss": 1.1844,
+      "step": 3960
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0003485231402417296,
+      "loss": 1.1369,
+      "step": 3961
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0003484950295575737,
+      "loss": 1.0879,
+      "step": 3962
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0003484669123344373,
+      "loss": 1.1204,
+      "step": 3963
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0003484387885735586,
+      "loss": 1.165,
+      "step": 3964
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000348410658276176,
+      "loss": 1.1021,
+      "step": 3965
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00034838252144352817,
+      "loss": 1.1156,
+      "step": 3966
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00034835437807685427,
+      "loss": 1.1318,
+      "step": 3967
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0003483262281773934,
+      "loss": 1.165,
+      "step": 3968
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00034829807174638523,
+      "loss": 1.1416,
+      "step": 3969
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00034826990878506967,
+      "loss": 1.1468,
+      "step": 3970
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0003482417392946868,
+      "loss": 1.1367,
+      "step": 3971
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00034821356327647716,
+      "loss": 1.1365,
+      "step": 3972
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00034818538073168136,
+      "loss": 1.092,
+      "step": 3973
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0003481571916615405,
+      "loss": 1.2232,
+      "step": 3974
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0003481289960672959,
+      "loss": 1.1371,
+      "step": 3975
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0003481007939501891,
+      "loss": 1.2195,
+      "step": 3976
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000348072585311462,
+      "loss": 1.1068,
+      "step": 3977
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0003480443701523568,
+      "loss": 1.1853,
+      "step": 3978
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00034801614847411585,
+      "loss": 1.106,
+      "step": 3979
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00034798792027798203,
+      "loss": 1.129,
+      "step": 3980
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0003479596855651982,
+      "loss": 1.1817,
+      "step": 3981
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0003479314443370079,
+      "loss": 1.1298,
+      "step": 3982
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0003479031965946546,
+      "loss": 1.176,
+      "step": 3983
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0003478749423393821,
+      "loss": 1.0879,
+      "step": 3984
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0003478466815724348,
+      "loss": 1.1987,
+      "step": 3985
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00034781841429505693,
+      "loss": 1.1404,
+      "step": 3986
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00034779014050849346,
+      "loss": 1.1642,
+      "step": 3987
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00034776186021398927,
+      "loss": 1.1733,
+      "step": 3988
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0003477335734127897,
+      "loss": 1.1869,
+      "step": 3989
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00034770528010614045,
+      "loss": 1.0908,
+      "step": 3990
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0003476769802952873,
+      "loss": 1.2043,
+      "step": 3991
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00034764867398147663,
+      "loss": 1.0904,
+      "step": 3992
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0003476203611659546,
+      "loss": 1.1708,
+      "step": 3993
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00034759204184996824,
+      "loss": 1.0906,
+      "step": 3994
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0003475637160347645,
+      "loss": 1.12,
+      "step": 3995
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0003475353837215906,
+      "loss": 1.1463,
+      "step": 3996
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00034750704491169426,
+      "loss": 1.101,
+      "step": 3997
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00034747869960632336,
+      "loss": 1.037,
+      "step": 3998
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00034745034780672614,
+      "loss": 1.0875,
+      "step": 3999
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000347421989514151,
+      "loss": 1.1513,
+      "step": 4000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0003473936247298467,
+      "loss": 1.1672,
+      "step": 4001
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00034736525345506236,
+      "loss": 1.1924,
+      "step": 4002
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0003473368756910472,
+      "loss": 1.1651,
+      "step": 4003
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0003473084914390509,
+      "loss": 1.1386,
+      "step": 4004
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0003472801007003234,
+      "loss": 1.1065,
+      "step": 4005
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00034725170347611484,
+      "loss": 1.1109,
+      "step": 4006
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00034722329976767565,
+      "loss": 1.0959,
+      "step": 4007
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0003471948895762566,
+      "loss": 1.2102,
+      "step": 4008
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00034716647290310886,
+      "loss": 1.2279,
+      "step": 4009
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00034713804974948354,
+      "loss": 1.1184,
+      "step": 4010
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0003471096201166325,
+      "loss": 1.128,
+      "step": 4011
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00034708118400580745,
+      "loss": 1.1092,
+      "step": 4012
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0003470527414182606,
+      "loss": 1.1224,
+      "step": 4013
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0003470242923552445,
+      "loss": 1.1298,
+      "step": 4014
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0003469958368180118,
+      "loss": 1.1151,
+      "step": 4015
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00034696737480781564,
+      "loss": 1.1873,
+      "step": 4016
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0003469389063259094,
+      "loss": 1.165,
+      "step": 4017
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00034691043137354645,
+      "loss": 1.1995,
+      "step": 4018
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0003468819499519809,
+      "loss": 1.1695,
+      "step": 4019
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0003468534620624669,
+      "loss": 1.1062,
+      "step": 4020
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0003468249677062588,
+      "loss": 1.1607,
+      "step": 4021
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00034679646688461145,
+      "loss": 1.1631,
+      "step": 4022
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0003467679595987799,
+      "loss": 1.0627,
+      "step": 4023
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0003467394458500194,
+      "loss": 1.0111,
+      "step": 4024
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00034671092563958556,
+      "loss": 1.2511,
+      "step": 4025
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00034668239896873426,
+      "loss": 1.1356,
+      "step": 4026
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0003466538658387218,
+      "loss": 1.1308,
+      "step": 4027
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00034662532625080446,
+      "loss": 1.0626,
+      "step": 4028
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0003465967802062391,
+      "loss": 1.1707,
+      "step": 4029
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00034656822770628266,
+      "loss": 1.1468,
+      "step": 4030
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0003465396687521925,
+      "loss": 1.0542,
+      "step": 4031
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0003465111033452263,
+      "loss": 1.1885,
+      "step": 4032
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0003464825314866417,
+      "loss": 1.15,
+      "step": 4033
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00034645395317769713,
+      "loss": 1.1826,
+      "step": 4034
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00034642536841965083,
+      "loss": 1.0628,
+      "step": 4035
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0003463967772137617,
+      "loss": 1.1151,
+      "step": 4036
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0003463681795612886,
+      "loss": 1.1521,
+      "step": 4037
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00034633957546349086,
+      "loss": 1.1184,
+      "step": 4038
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0003463109649216282,
+      "loss": 1.1006,
+      "step": 4039
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0003462823479369603,
+      "loss": 1.138,
+      "step": 4040
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00034625372451074744,
+      "loss": 1.155,
+      "step": 4041
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00034622509464425004,
+      "loss": 1.181,
+      "step": 4042
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00034619645833872874,
+      "loss": 1.1618,
+      "step": 4043
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00034616781559544457,
+      "loss": 1.1208,
+      "step": 4044
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00034613916641565884,
+      "loss": 1.222,
+      "step": 4045
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0003461105108006331,
+      "loss": 1.0672,
+      "step": 4046
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00034608184875162924,
+      "loss": 1.1765,
+      "step": 4047
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00034605318026990927,
+      "loss": 1.1441,
+      "step": 4048
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00034602450535673574,
+      "loss": 1.2543,
+      "step": 4049
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00034599582401337135,
+      "loss": 1.2008,
+      "step": 4050
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000345967136241079,
+      "loss": 1.1083,
+      "step": 4051
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000345938442041122,
+      "loss": 1.1764,
+      "step": 4052
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0003459097414147639,
+      "loss": 1.16,
+      "step": 4053
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0003458810343632685,
+      "loss": 1.1692,
+      "step": 4054
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0003458523208878999,
+      "loss": 1.132,
+      "step": 4055
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0003458236009899226,
+      "loss": 1.1037,
+      "step": 4056
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00034579487467060126,
+      "loss": 1.1693,
+      "step": 4057
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00034576614193120074,
+      "loss": 1.1887,
+      "step": 4058
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00034573740277298634,
+      "loss": 1.1358,
+      "step": 4059
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0003457086571972237,
+      "loss": 1.1529,
+      "step": 4060
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00034567990520517844,
+      "loss": 1.1447,
+      "step": 4061
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0003456511467981168,
+      "loss": 1.151,
+      "step": 4062
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000345622381977305,
+      "loss": 1.095,
+      "step": 4063
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0003455936107440099,
+      "loss": 1.1626,
+      "step": 4064
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0003455648330994984,
+      "loss": 1.134,
+      "step": 4065
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00034553604904503753,
+      "loss": 1.1357,
+      "step": 4066
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000345507258581895,
+      "loss": 1.0539,
+      "step": 4067
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0003454784617113385,
+      "loss": 1.1898,
+      "step": 4068
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00034544965843463614,
+      "loss": 1.1819,
+      "step": 4069
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0003454208487530563,
+      "loss": 1.1516,
+      "step": 4070
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00034539203266786757,
+      "loss": 1.1373,
+      "step": 4071
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0003453632101803389,
+      "loss": 1.1787,
+      "step": 4072
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00034533438129173943,
+      "loss": 1.2127,
+      "step": 4073
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00034530554600333864,
+      "loss": 1.1057,
+      "step": 4074
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0003452767043164064,
+      "loss": 1.1889,
+      "step": 4075
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0003452478562322127,
+      "loss": 1.1599,
+      "step": 4076
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0003452190017520278,
+      "loss": 1.1408,
+      "step": 4077
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0003451901408771223,
+      "loss": 1.1442,
+      "step": 4078
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0003451612736087673,
+      "loss": 1.1943,
+      "step": 4079
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00034513239994823366,
+      "loss": 1.0902,
+      "step": 4080
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000345103519896793,
+      "loss": 1.1902,
+      "step": 4081
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0003450746334557171,
+      "loss": 1.2381,
+      "step": 4082
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0003450457406262779,
+      "loss": 1.1501,
+      "step": 4083
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0003450168414097478,
+      "loss": 1.1135,
+      "step": 4084
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0003449879358073991,
+      "loss": 1.0869,
+      "step": 4085
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000344959023820505,
+      "loss": 1.1838,
+      "step": 4086
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0003449301054503384,
+      "loss": 1.1957,
+      "step": 4087
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0003449011806981729,
+      "loss": 1.1759,
+      "step": 4088
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00034487224956528204,
+      "loss": 1.1237,
+      "step": 4089
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00034484331205293986,
+      "loss": 1.1596,
+      "step": 4090
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0003448143681624207,
+      "loss": 1.1469,
+      "step": 4091
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00034478541789499895,
+      "loss": 1.1184,
+      "step": 4092
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0003447564612519496,
+      "loss": 1.1752,
+      "step": 4093
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00034472749823454766,
+      "loss": 1.2087,
+      "step": 4094
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0003446985288440685,
+      "loss": 1.1032,
+      "step": 4095
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0003446695530817879,
+      "loss": 1.1076,
+      "step": 4096
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00034464057094898173,
+      "loss": 1.2452,
+      "step": 4097
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00034461158244692623,
+      "loss": 1.15,
+      "step": 4098
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00034458258757689785,
+      "loss": 1.1346,
+      "step": 4099
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00034455358634017346,
+      "loss": 1.0815,
+      "step": 4100
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00034452457873803007,
+      "loss": 1.1393,
+      "step": 4101
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00034449556477174505,
+      "loss": 1.1366,
+      "step": 4102
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0003444665444425961,
+      "loss": 1.2295,
+      "step": 4103
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0003444375177518611,
+      "loss": 1.1813,
+      "step": 4104
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0003444084847008181,
+      "loss": 1.1506,
+      "step": 4105
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00034437944529074575,
+      "loss": 1.1316,
+      "step": 4106
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00034435039952292277,
+      "loss": 1.1279,
+      "step": 4107
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0003443213473986281,
+      "loss": 1.1865,
+      "step": 4108
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0003442922889191411,
+      "loss": 1.2108,
+      "step": 4109
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00034426322408574137,
+      "loss": 1.0331,
+      "step": 4110
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00034423415289970886,
+      "loss": 1.1855,
+      "step": 4111
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0003442050753623236,
+      "loss": 1.1772,
+      "step": 4112
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00034417599147486597,
+      "loss": 1.1677,
+      "step": 4113
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0003441469012386168,
+      "loss": 1.1116,
+      "step": 4114
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00034411780465485713,
+      "loss": 1.138,
+      "step": 4115
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0003440887017248681,
+      "loss": 1.2607,
+      "step": 4116
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0003440595924499313,
+      "loss": 1.0509,
+      "step": 4117
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0003440304768313285,
+      "loss": 1.1856,
+      "step": 4118
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0003440013548703419,
+      "loss": 1.2222,
+      "step": 4119
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0003439722265682539,
+      "loss": 1.2851,
+      "step": 4120
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00034394309192634705,
+      "loss": 1.2459,
+      "step": 4121
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0003439139509459044,
+      "loss": 1.0856,
+      "step": 4122
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00034388480362820915,
+      "loss": 1.046,
+      "step": 4123
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00034385564997454473,
+      "loss": 1.1494,
+      "step": 4124
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000343826489986195,
+      "loss": 1.1253,
+      "step": 4125
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00034379732366444406,
+      "loss": 1.1843,
+      "step": 4126
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0003437681510105761,
+      "loss": 1.1155,
+      "step": 4127
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00034373897202587587,
+      "loss": 1.0594,
+      "step": 4128
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00034370978671162823,
+      "loss": 1.1648,
+      "step": 4129
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00034368059506911835,
+      "loss": 1.1216,
+      "step": 4130
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0003436513970996316,
+      "loss": 1.1939,
+      "step": 4131
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0003436221928044539,
+      "loss": 1.1409,
+      "step": 4132
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00034359298218487107,
+      "loss": 1.2941,
+      "step": 4133
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0003435637652421695,
+      "loss": 1.2075,
+      "step": 4134
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00034353454197763583,
+      "loss": 1.1813,
+      "step": 4135
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00034350531239255684,
+      "loss": 1.0742,
+      "step": 4136
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0003434760764882195,
+      "loss": 1.15,
+      "step": 4137
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0003434468342659114,
+      "loss": 1.0369,
+      "step": 4138
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00034341758572692025,
+      "loss": 1.2205,
+      "step": 4139
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0003433883308725339,
+      "loss": 1.07,
+      "step": 4140
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00034335906970404056,
+      "loss": 1.1346,
+      "step": 4141
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00034332980222272886,
+      "loss": 1.1396,
+      "step": 4142
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00034330052842988757,
+      "loss": 1.2024,
+      "step": 4143
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00034327124832680565,
+      "loss": 1.2135,
+      "step": 4144
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00034324196191477254,
+      "loss": 1.1764,
+      "step": 4145
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00034321266919507795,
+      "loss": 1.1324,
+      "step": 4146
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0003431833701690116,
+      "loss": 1.2773,
+      "step": 4147
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0003431540648378638,
+      "loss": 1.1839,
+      "step": 4148
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.000343124753202925,
+      "loss": 1.1473,
+      "step": 4149
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0003430954352654859,
+      "loss": 1.1422,
+      "step": 4150
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00034306611102683757,
+      "loss": 1.1506,
+      "step": 4151
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0003430367804882712,
+      "loss": 1.1305,
+      "step": 4152
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00034300744365107843,
+      "loss": 1.2064,
+      "step": 4153
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0003429781005165512,
+      "loss": 1.0851,
+      "step": 4154
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0003429487510859814,
+      "loss": 1.2263,
+      "step": 4155
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00034291939536066165,
+      "loss": 1.1913,
+      "step": 4156
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0003428900333418846,
+      "loss": 1.1056,
+      "step": 4157
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0003428606650309431,
+      "loss": 1.1175,
+      "step": 4158
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0003428312904291304,
+      "loss": 1.1794,
+      "step": 4159
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0003428019095377401,
+      "loss": 1.1923,
+      "step": 4160
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00034277252235806596,
+      "loss": 1.2005,
+      "step": 4161
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00034274312889140195,
+      "loss": 1.1547,
+      "step": 4162
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00034271372913904257,
+      "loss": 1.1721,
+      "step": 4163
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0003426843231022823,
+      "loss": 1.2367,
+      "step": 4164
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00034265491078241614,
+      "loss": 1.1572,
+      "step": 4165
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00034262549218073914,
+      "loss": 1.1565,
+      "step": 4166
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0003425960672985469,
+      "loss": 1.2154,
+      "step": 4167
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00034256663613713503,
+      "loss": 1.1437,
+      "step": 4168
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0003425371986977996,
+      "loss": 1.2128,
+      "step": 4169
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00034250775498183685,
+      "loss": 1.084,
+      "step": 4170
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00034247830499054336,
+      "loss": 1.1589,
+      "step": 4171
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0003424488487252159,
+      "loss": 1.0969,
+      "step": 4172
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0003424193861871516,
+      "loss": 1.1455,
+      "step": 4173
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0003423899173776479,
+      "loss": 1.1912,
+      "step": 4174
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00034236044229800236,
+      "loss": 1.1189,
+      "step": 4175
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00034233096094951306,
+      "loss": 1.1492,
+      "step": 4176
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0003423014733334781,
+      "loss": 1.1653,
+      "step": 4177
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.000342271979451196,
+      "loss": 1.1163,
+      "step": 4178
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0003422424793039655,
+      "loss": 1.1838,
+      "step": 4179
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00034221297289308564,
+      "loss": 1.1361,
+      "step": 4180
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0003421834602198558,
+      "loss": 1.1032,
+      "step": 4181
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0003421539412855755,
+      "loss": 1.1641,
+      "step": 4182
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0003421244160915446,
+      "loss": 1.1789,
+      "step": 4183
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0003420948846390633,
+      "loss": 1.0417,
+      "step": 4184
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0003420653469294321,
+      "loss": 1.2625,
+      "step": 4185
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00034203580296395143,
+      "loss": 1.1103,
+      "step": 4186
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0003420062527439225,
+      "loss": 1.211,
+      "step": 4187
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0003419766962706464,
+      "loss": 1.0982,
+      "step": 4188
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0003419471335454248,
+      "loss": 1.1239,
+      "step": 4189
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0003419175645695593,
+      "loss": 1.1697,
+      "step": 4190
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00034188798934435224,
+      "loss": 1.0991,
+      "step": 4191
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0003418584078711057,
+      "loss": 1.1763,
+      "step": 4192
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00034182882015112244,
+      "loss": 1.1186,
+      "step": 4193
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0003417992261857053,
+      "loss": 1.2302,
+      "step": 4194
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0003417696259761575,
+      "loss": 1.1517,
+      "step": 4195
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00034174001952378243,
+      "loss": 1.2683,
+      "step": 4196
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0003417104068298839,
+      "loss": 1.1638,
+      "step": 4197
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0003416807878957658,
+      "loss": 1.1158,
+      "step": 4198
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0003416511627227324,
+      "loss": 1.1391,
+      "step": 4199
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0003416215313120884,
+      "loss": 1.1403,
+      "step": 4200
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00034159189366513845,
+      "loss": 1.1629,
+      "step": 4201
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00034156224978318774,
+      "loss": 1.012,
+      "step": 4202
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00034153259966754156,
+      "loss": 1.0502,
+      "step": 4203
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00034150294331950566,
+      "loss": 1.1565,
+      "step": 4204
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0003414732807403858,
+      "loss": 1.1045,
+      "step": 4205
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0003414436119314883,
+      "loss": 1.1821,
+      "step": 4206
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0003414139368941196,
+      "loss": 1.105,
+      "step": 4207
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0003413842556295864,
+      "loss": 1.124,
+      "step": 4208
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0003413545681391959,
+      "loss": 1.1781,
+      "step": 4209
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0003413248744242551,
+      "loss": 1.0802,
+      "step": 4210
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00034129517448607173,
+      "loss": 1.1367,
+      "step": 4211
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0003412654683259536,
+      "loss": 1.1797,
+      "step": 4212
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00034123575594520884,
+      "loss": 1.0445,
+      "step": 4213
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0003412060373451458,
+      "loss": 1.1309,
+      "step": 4214
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0003411763125270731,
+      "loss": 1.1065,
+      "step": 4215
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0003411465814922998,
+      "loss": 1.2134,
+      "step": 4216
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.000341116844242135,
+      "loss": 1.1592,
+      "step": 4217
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0003410871007778882,
+      "loss": 1.1771,
+      "step": 4218
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00034105735110086916,
+      "loss": 1.1487,
+      "step": 4219
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.000341027595212388,
+      "loss": 1.0569,
+      "step": 4220
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00034099783311375485,
+      "loss": 1.1656,
+      "step": 4221
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0003409680648062804,
+      "loss": 1.1413,
+      "step": 4222
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0003409382902912755,
+      "loss": 1.1457,
+      "step": 4223
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0003409085095700512,
+      "loss": 1.1314,
+      "step": 4224
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00034087872264391885,
+      "loss": 1.109,
+      "step": 4225
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0003408489295141903,
+      "loss": 1.0412,
+      "step": 4226
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0003408191301821774,
+      "loss": 1.2238,
+      "step": 4227
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0003407893246491923,
+      "loss": 1.1236,
+      "step": 4228
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00034075951291654755,
+      "loss": 1.1436,
+      "step": 4229
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00034072969498555593,
+      "loss": 1.1251,
+      "step": 4230
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00034069987085753043,
+      "loss": 1.1915,
+      "step": 4231
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0003406700405337844,
+      "loss": 1.1053,
+      "step": 4232
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00034064020401563134,
+      "loss": 1.1594,
+      "step": 4233
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0003406103613043852,
+      "loss": 1.1234,
+      "step": 4234
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00034058051240136005,
+      "loss": 1.1164,
+      "step": 4235
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00034055065730787025,
+      "loss": 1.17,
+      "step": 4236
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0003405207960252305,
+      "loss": 1.1166,
+      "step": 4237
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0003404909285547558,
+      "loss": 1.1474,
+      "step": 4238
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00034046105489776127,
+      "loss": 1.1447,
+      "step": 4239
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00034043117505556253,
+      "loss": 1.1096,
+      "step": 4240
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00034040128902947515,
+      "loss": 1.1283,
+      "step": 4241
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00034037139682081537,
+      "loss": 1.1354,
+      "step": 4242
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00034034149843089934,
+      "loss": 1.1517,
+      "step": 4243
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0003403115938610436,
+      "loss": 1.2549,
+      "step": 4244
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00034028168311256527,
+      "loss": 1.082,
+      "step": 4245
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00034025176618678114,
+      "loss": 1.1306,
+      "step": 4246
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0003402218430850088,
+      "loss": 1.0449,
+      "step": 4247
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0003401919138085658,
+      "loss": 1.1245,
+      "step": 4248
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00034016197835877015,
+      "loss": 1.1869,
+      "step": 4249
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0003401320367369401,
+      "loss": 1.211,
+      "step": 4250
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00034010208894439406,
+      "loss": 1.1306,
+      "step": 4251
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0003400721349824508,
+      "loss": 1.1404,
+      "step": 4252
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0003400421748524293,
+      "loss": 1.1985,
+      "step": 4253
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00034001220855564893,
+      "loss": 1.0801,
+      "step": 4254
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00033998223609342916,
+      "loss": 1.1788,
+      "step": 4255
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00033995225746708994,
+      "loss": 1.145,
+      "step": 4256
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00033992227267795133,
+      "loss": 1.2023,
+      "step": 4257
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0003398922817273337,
+      "loss": 1.1294,
+      "step": 4258
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00033986228461655766,
+      "loss": 1.1737,
+      "step": 4259
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00033983228134694426,
+      "loss": 1.15,
+      "step": 4260
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0003398022719198145,
+      "loss": 1.1558,
+      "step": 4261
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0003397722563364901,
+      "loss": 1.175,
+      "step": 4262
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0003397422345982926,
+      "loss": 1.092,
+      "step": 4263
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.000339712206706544,
+      "loss": 1.1243,
+      "step": 4264
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00033968217266256676,
+      "loss": 1.1239,
+      "step": 4265
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0003396521324676832,
+      "loss": 1.1005,
+      "step": 4266
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0003396220861232163,
+      "loss": 1.1413,
+      "step": 4267
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0003395920336304891,
+      "loss": 1.0598,
+      "step": 4268
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00033956197499082505,
+      "loss": 1.098,
+      "step": 4269
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0003395319102055476,
+      "loss": 1.1957,
+      "step": 4270
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0003395018392759807,
+      "loss": 1.1079,
+      "step": 4271
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00033947176220344873,
+      "loss": 1.0789,
+      "step": 4272
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0003394416789892759,
+      "loss": 1.1698,
+      "step": 4273
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00033941158963478703,
+      "loss": 1.1544,
+      "step": 4274
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.000339381494141307,
+      "loss": 1.0838,
+      "step": 4275
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0003393513925101613,
+      "loss": 1.076,
+      "step": 4276
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00033932128474267514,
+      "loss": 1.1332,
+      "step": 4277
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0003392911708401745,
+      "loss": 1.159,
+      "step": 4278
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0003392610508039855,
+      "loss": 1.1926,
+      "step": 4279
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0003392309246354344,
+      "loss": 1.1966,
+      "step": 4280
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00033920079233584776,
+      "loss": 1.0968,
+      "step": 4281
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0003391706539065525,
+      "loss": 1.1646,
+      "step": 4282
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0003391405093488758,
+      "loss": 1.0974,
+      "step": 4283
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00033911035866414504,
+      "loss": 1.1363,
+      "step": 4284
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0003390802018536879,
+      "loss": 1.1728,
+      "step": 4285
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0003390500389188324,
+      "loss": 1.0611,
+      "step": 4286
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0003390198698609066,
+      "loss": 1.1204,
+      "step": 4287
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00033898969468123917,
+      "loss": 1.1724,
+      "step": 4288
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0003389595133811588,
+      "loss": 1.1746,
+      "step": 4289
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00033892932596199456,
+      "loss": 1.2275,
+      "step": 4290
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0003388991324250757,
+      "loss": 1.0991,
+      "step": 4291
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0003388689327717318,
+      "loss": 1.1628,
+      "step": 4292
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00033883872700329274,
+      "loss": 1.0647,
+      "step": 4293
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0003388085151210886,
+      "loss": 1.1532,
+      "step": 4294
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0003387782971264498,
+      "loss": 1.1955,
+      "step": 4295
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00033874807302070704,
+      "loss": 1.1629,
+      "step": 4296
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.000338717842805191,
+      "loss": 1.0945,
+      "step": 4297
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00033868760648123313,
+      "loss": 1.2339,
+      "step": 4298
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00033865736405016476,
+      "loss": 1.2647,
+      "step": 4299
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00033862711551331764,
+      "loss": 1.1336,
+      "step": 4300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0003385968608720238,
+      "loss": 1.251,
+      "step": 4301
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00033856660012761546,
+      "loss": 1.1356,
+      "step": 4302
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00033853633328142515,
+      "loss": 1.1901,
+      "step": 4303
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00033850606033478564,
+      "loss": 1.1607,
+      "step": 4304
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00033847578128903014,
+      "loss": 1.1133,
+      "step": 4305
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0003384454961454918,
+      "loss": 1.1939,
+      "step": 4306
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0003384152049055044,
+      "loss": 1.1365,
+      "step": 4307
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0003383849075704017,
+      "loss": 1.0944,
+      "step": 4308
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00033835460414151787,
+      "loss": 1.0708,
+      "step": 4309
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00033832429462018733,
+      "loss": 1.1615,
+      "step": 4310
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00033829397900774475,
+      "loss": 1.1127,
+      "step": 4311
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0003382636573055251,
+      "loss": 1.0802,
+      "step": 4312
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00033823332951486365,
+      "loss": 1.169,
+      "step": 4313
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0003382029956370957,
+      "loss": 1.0692,
+      "step": 4314
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0003381726556735572,
+      "loss": 1.1493,
+      "step": 4315
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00033814230962558406,
+      "loss": 1.2178,
+      "step": 4316
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00033811195749451265,
+      "loss": 1.1857,
+      "step": 4317
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00033808159928167946,
+      "loss": 1.1461,
+      "step": 4318
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00033805123498842127,
+      "loss": 1.1272,
+      "step": 4319
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00033802086461607526,
+      "loss": 1.0823,
+      "step": 4320
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00033799048816597876,
+      "loss": 1.1893,
+      "step": 4321
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00033796010563946945,
+      "loss": 1.1656,
+      "step": 4322
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00033792971703788517,
+      "loss": 1.1319,
+      "step": 4323
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00033789932236256406,
+      "loss": 1.2147,
+      "step": 4324
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00033786892161484456,
+      "loss": 1.1853,
+      "step": 4325
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00033783851479606545,
+      "loss": 1.1213,
+      "step": 4326
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0003378081019075656,
+      "loss": 1.1503,
+      "step": 4327
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0003377776829506842,
+      "loss": 1.1582,
+      "step": 4328
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0003377472579267609,
+      "loss": 1.1122,
+      "step": 4329
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0003377168268371354,
+      "loss": 1.1302,
+      "step": 4330
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0003376863896831477,
+      "loss": 1.1475,
+      "step": 4331
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0003376559464661381,
+      "loss": 1.1231,
+      "step": 4332
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0003376254971874472,
+      "loss": 1.1299,
+      "step": 4333
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00033759504184841584,
+      "loss": 1.1848,
+      "step": 4334
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00033756458045038507,
+      "loss": 1.1889,
+      "step": 4335
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00033753411299469635,
+      "loss": 1.1614,
+      "step": 4336
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00033750363948269126,
+      "loss": 1.1284,
+      "step": 4337
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0003374731599157116,
+      "loss": 1.1612,
+      "step": 4338
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00033744267429509976,
+      "loss": 1.1822,
+      "step": 4339
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00033741218262219795,
+      "loss": 1.1446,
+      "step": 4340
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00033738168489834907,
+      "loss": 1.1907,
+      "step": 4341
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00033735118112489585,
+      "loss": 1.193,
+      "step": 4342
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0003373206713031818,
+      "loss": 1.1458,
+      "step": 4343
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00033729015543455023,
+      "loss": 1.1808,
+      "step": 4344
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.000337259633520345,
+      "loss": 1.2061,
+      "step": 4345
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00033722910556191005,
+      "loss": 1.1082,
+      "step": 4346
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0003371985715605897,
+      "loss": 1.0641,
+      "step": 4347
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00033716803151772863,
+      "loss": 1.09,
+      "step": 4348
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00033713748543467157,
+      "loss": 1.1987,
+      "step": 4349
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0003371069333127636,
+      "loss": 1.1899,
+      "step": 4350
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0003370763751533502,
+      "loss": 1.1422,
+      "step": 4351
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00033704581095777685,
+      "loss": 1.1382,
+      "step": 4352
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00033701524072738954,
+      "loss": 1.1062,
+      "step": 4353
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0003369846644635344,
+      "loss": 1.1824,
+      "step": 4354
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00033695408216755785,
+      "loss": 1.1889,
+      "step": 4355
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00033692349384080656,
+      "loss": 1.1382,
+      "step": 4356
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0003368928994846276,
+      "loss": 1.1131,
+      "step": 4357
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00033686229910036806,
+      "loss": 1.0888,
+      "step": 4358
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0003368316926893755,
+      "loss": 1.1339,
+      "step": 4359
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0003368010802529977,
+      "loss": 1.159,
+      "step": 4360
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0003367704617925825,
+      "loss": 1.0791,
+      "step": 4361
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00033673983730947844,
+      "loss": 1.1233,
+      "step": 4362
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0003367092068050339,
+      "loss": 1.1524,
+      "step": 4363
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00033667857028059773,
+      "loss": 1.1571,
+      "step": 4364
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00033664792773751905,
+      "loss": 1.131,
+      "step": 4365
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00033661727917714716,
+      "loss": 1.2011,
+      "step": 4366
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00033658662460083166,
+      "loss": 1.0826,
+      "step": 4367
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00033655596400992255,
+      "loss": 1.0959,
+      "step": 4368
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0003365252974057698,
+      "loss": 1.1895,
+      "step": 4369
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00033649462478972387,
+      "loss": 1.1881,
+      "step": 4370
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0003364639461631355,
+      "loss": 1.1362,
+      "step": 4371
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00033643326152735546,
+      "loss": 1.1383,
+      "step": 4372
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00033640257088373516,
+      "loss": 1.1208,
+      "step": 4373
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0003363718742336259,
+      "loss": 1.1422,
+      "step": 4374
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0003363411715783794,
+      "loss": 1.1362,
+      "step": 4375
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00033631046291934786,
+      "loss": 1.1209,
+      "step": 4376
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00033627974825788337,
+      "loss": 1.2298,
+      "step": 4377
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0003362490275953384,
+      "loss": 1.0371,
+      "step": 4378
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0003362183009330659,
+      "loss": 1.0981,
+      "step": 4379
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00033618756827241876,
+      "loss": 1.1942,
+      "step": 4380
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0003361568296147504,
+      "loss": 1.1228,
+      "step": 4381
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00033612608496141433,
+      "loss": 1.062,
+      "step": 4382
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0003360953343137644,
+      "loss": 1.1656,
+      "step": 4383
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00033606457767315477,
+      "loss": 1.1406,
+      "step": 4384
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00033603381504093975,
+      "loss": 1.1111,
+      "step": 4385
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0003360030464184741,
+      "loss": 1.0704,
+      "step": 4386
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00033597227180711244,
+      "loss": 1.1192,
+      "step": 4387
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0003359414912082102,
+      "loss": 1.1998,
+      "step": 4388
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0003359107046231227,
+      "loss": 1.0797,
+      "step": 4389
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0003358799120532056,
+      "loss": 1.1466,
+      "step": 4390
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00033584911349981495,
+      "loss": 1.169,
+      "step": 4391
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0003358183089643069,
+      "loss": 1.1665,
+      "step": 4392
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0003357874984480379,
+      "loss": 1.1638,
+      "step": 4393
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0003357566819523647,
+      "loss": 1.0903,
+      "step": 4394
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00033572585947864433,
+      "loss": 1.167,
+      "step": 4395
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0003356950310282341,
+      "loss": 1.1108,
+      "step": 4396
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0003356641966024914,
+      "loss": 1.0617,
+      "step": 4397
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0003356333562027742,
+      "loss": 1.1514,
+      "step": 4398
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00033560250983044045,
+      "loss": 1.1196,
+      "step": 4399
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00033557165748684855,
+      "loss": 1.1063,
+      "step": 4400
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00033554079917335694,
+      "loss": 1.1011,
+      "step": 4401
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0003355099348913246,
+      "loss": 1.138,
+      "step": 4402
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00033547906464211064,
+      "loss": 1.1239,
+      "step": 4403
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0003354481884270743,
+      "loss": 1.1015,
+      "step": 4404
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0003354173062475754,
+      "loss": 1.1845,
+      "step": 4405
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0003353864181049737,
+      "loss": 1.1862,
+      "step": 4406
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0003353555240006293,
+      "loss": 1.1258,
+      "step": 4407
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00033532462393590285,
+      "loss": 1.2019,
+      "step": 4408
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00033529371791215487,
+      "loss": 1.1308,
+      "step": 4409
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0003352628059307463,
+      "loss": 1.1044,
+      "step": 4410
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00033523188799303844,
+      "loss": 1.1097,
+      "step": 4411
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00033520096410039274,
+      "loss": 1.2142,
+      "step": 4412
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0003351700342541708,
+      "loss": 1.1549,
+      "step": 4413
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0003351390984557348,
+      "loss": 1.1289,
+      "step": 4414
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0003351081567064469,
+      "loss": 1.2543,
+      "step": 4415
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0003350772090076697,
+      "loss": 1.2069,
+      "step": 4416
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0003350462553607659,
+      "loss": 1.0792,
+      "step": 4417
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0003350152957670985,
+      "loss": 1.1422,
+      "step": 4418
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00033498433022803093,
+      "loss": 1.1378,
+      "step": 4419
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00033495335874492675,
+      "loss": 1.0986,
+      "step": 4420
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0003349223813191497,
+      "loss": 1.1214,
+      "step": 4421
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00033489139795206386,
+      "loss": 1.1096,
+      "step": 4422
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0003348604086450337,
+      "loss": 1.1397,
+      "step": 4423
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00033482941339942377,
+      "loss": 1.0717,
+      "step": 4424
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0003347984122165989,
+      "loss": 1.1326,
+      "step": 4425
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0003347674050979243,
+      "loss": 1.1528,
+      "step": 4426
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0003347363920447654,
+      "loss": 1.1726,
+      "step": 4427
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0003347053730584877,
+      "loss": 1.2039,
+      "step": 4428
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00033467434814045735,
+      "loss": 1.1224,
+      "step": 4429
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0003346433172920403,
+      "loss": 1.2512,
+      "step": 4430
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00033461228051460316,
+      "loss": 1.1287,
+      "step": 4431
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00033458123780951256,
+      "loss": 1.1234,
+      "step": 4432
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00033455018917813553,
+      "loss": 1.2096,
+      "step": 4433
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00033451913462183915,
+      "loss": 1.1079,
+      "step": 4434
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0003344880741419911,
+      "loss": 1.1165,
+      "step": 4435
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.000334457007739959,
+      "loss": 1.1366,
+      "step": 4436
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00033442593541711086,
+      "loss": 1.0758,
+      "step": 4437
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.000334394857174815,
+      "loss": 1.1978,
+      "step": 4438
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00033436377301443996,
+      "loss": 1.1031,
+      "step": 4439
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00033433268293735446,
+      "loss": 1.0845,
+      "step": 4440
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00033430158694492773,
+      "loss": 1.2407,
+      "step": 4441
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00033427048503852883,
+      "loss": 1.1972,
+      "step": 4442
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00033423937721952743,
+      "loss": 1.0833,
+      "step": 4443
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00033420826348929354,
+      "loss": 1.1879,
+      "step": 4444
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00033417714384919694,
+      "loss": 1.1617,
+      "step": 4445
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0003341460183006082,
+      "loss": 1.1034,
+      "step": 4446
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0003341148868448979,
+      "loss": 1.0815,
+      "step": 4447
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00033408374948343695,
+      "loss": 1.1195,
+      "step": 4448
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00033405260621759635,
+      "loss": 1.0987,
+      "step": 4449
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0003340214570487476,
+      "loss": 1.1622,
+      "step": 4450
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00033399030197826227,
+      "loss": 1.1976,
+      "step": 4451
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00033395914100751235,
+      "loss": 1.1833,
+      "step": 4452
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00033392797413787,
+      "loss": 1.113,
+      "step": 4453
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00033389680137070767,
+      "loss": 1.1606,
+      "step": 4454
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00033386562270739805,
+      "loss": 1.2375,
+      "step": 4455
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00033383443814931406,
+      "loss": 1.1518,
+      "step": 4456
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0003338032476978289,
+      "loss": 1.1893,
+      "step": 4457
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0003337720513543161,
+      "loss": 1.149,
+      "step": 4458
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0003337408491201493,
+      "loss": 1.1468,
+      "step": 4459
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0003337096409967026,
+      "loss": 1.2722,
+      "step": 4460
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0003336784269853502,
+      "loss": 1.1691,
+      "step": 4461
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0003336472070874666,
+      "loss": 1.1158,
+      "step": 4462
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00033361598130442655,
+      "loss": 1.1521,
+      "step": 4463
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0003335847496376052,
+      "loss": 1.1721,
+      "step": 4464
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0003335535120883776,
+      "loss": 1.143,
+      "step": 4465
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0003335222686581195,
+      "loss": 1.154,
+      "step": 4466
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0003334910193482067,
+      "loss": 1.123,
+      "step": 4467
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0003334597641600151,
+      "loss": 1.1703,
+      "step": 4468
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00033342850309492117,
+      "loss": 1.2455,
+      "step": 4469
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0003333972361543015,
+      "loss": 1.1679,
+      "step": 4470
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00033336596333953284,
+      "loss": 1.1142,
+      "step": 4471
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00033333468465199234,
+      "loss": 1.0605,
+      "step": 4472
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00033330340009305733,
+      "loss": 1.1028,
+      "step": 4473
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00033327210966410545,
+      "loss": 1.1274,
+      "step": 4474
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0003332408133665146,
+      "loss": 1.1834,
+      "step": 4475
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0003332095112016628,
+      "loss": 1.0449,
+      "step": 4476
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0003331782031709286,
+      "loss": 1.147,
+      "step": 4477
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0003331468892756905,
+      "loss": 1.0534,
+      "step": 4478
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00033311556951732754,
+      "loss": 1.1556,
+      "step": 4479
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0003330842438972188,
+      "loss": 1.1097,
+      "step": 4480
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00033305291241674374,
+      "loss": 1.1189,
+      "step": 4481
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00033302157507728204,
+      "loss": 1.1347,
+      "step": 4482
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0003329902318802137,
+      "loss": 1.1033,
+      "step": 4483
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00033295888282691867,
+      "loss": 1.1489,
+      "step": 4484
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0003329275279187777,
+      "loss": 1.1519,
+      "step": 4485
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00033289616715717136,
+      "loss": 1.1874,
+      "step": 4486
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0003328648005434806,
+      "loss": 1.1192,
+      "step": 4487
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0003328334280790868,
+      "loss": 1.106,
+      "step": 4488
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0003328020497653713,
+      "loss": 1.2377,
+      "step": 4489
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0003327706656037159,
+      "loss": 1.0739,
+      "step": 4490
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0003327392755955026,
+      "loss": 1.1483,
+      "step": 4491
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00033270787974211364,
+      "loss": 1.1504,
+      "step": 4492
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0003326764780449315,
+      "loss": 1.1707,
+      "step": 4493
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0003326450705053391,
+      "loss": 1.1668,
+      "step": 4494
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00033261365712471927,
+      "loss": 1.1492,
+      "step": 4495
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00033258223790445546,
+      "loss": 1.1582,
+      "step": 4496
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00033255081284593117,
+      "loss": 1.1626,
+      "step": 4497
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00033251938195053016,
+      "loss": 1.0754,
+      "step": 4498
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0003324879452196365,
+      "loss": 1.0858,
+      "step": 4499
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0003324565026546346,
+      "loss": 1.2144,
+      "step": 4500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00033242505425690893,
+      "loss": 1.1263,
+      "step": 4501
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0003323936000278444,
+      "loss": 1.059,
+      "step": 4502
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00033236213996882597,
+      "loss": 1.1214,
+      "step": 4503
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00033233067408123905,
+      "loss": 1.1327,
+      "step": 4504
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0003322992023664694,
+      "loss": 1.1031,
+      "step": 4505
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00033226772482590255,
+      "loss": 1.1772,
+      "step": 4506
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00033223624146092484,
+      "loss": 1.1736,
+      "step": 4507
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00033220475227292267,
+      "loss": 1.0856,
+      "step": 4508
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00033217325726328257,
+      "loss": 1.0568,
+      "step": 4509
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00033214175643339145,
+      "loss": 1.1281,
+      "step": 4510
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00033211024978463643,
+      "loss": 1.1625,
+      "step": 4511
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0003320787373184048,
+      "loss": 1.0997,
+      "step": 4512
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0003320472190360845,
+      "loss": 1.2125,
+      "step": 4513
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0003320156949390633,
+      "loss": 1.1415,
+      "step": 4514
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0003319841650287292,
+      "loss": 1.0957,
+      "step": 4515
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00033195262930647086,
+      "loss": 1.1153,
+      "step": 4516
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0003319210877736768,
+      "loss": 1.1607,
+      "step": 4517
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00033188954043173596,
+      "loss": 1.1207,
+      "step": 4518
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0003318579872820376,
+      "loss": 1.1308,
+      "step": 4519
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00033182642832597117,
+      "loss": 1.1605,
+      "step": 4520
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0003317948635649263,
+      "loss": 1.1053,
+      "step": 4521
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00033176329300029303,
+      "loss": 1.089,
+      "step": 4522
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0003317317166334615,
+      "loss": 1.1341,
+      "step": 4523
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0003317001344658221,
+      "loss": 1.1526,
+      "step": 4524
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0003316685464987657,
+      "loss": 1.1556,
+      "step": 4525
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0003316369527336833,
+      "loss": 1.1566,
+      "step": 4526
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0003316053531719659,
+      "loss": 1.1866,
+      "step": 4527
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0003315737478150053,
+      "loss": 1.0636,
+      "step": 4528
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00033154213666419303,
+      "loss": 1.163,
+      "step": 4529
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00033151051972092106,
+      "loss": 1.1164,
+      "step": 4530
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0003314788969865818,
+      "loss": 1.1985,
+      "step": 4531
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0003314472684625677,
+      "loss": 1.1648,
+      "step": 4532
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00033141563415027146,
+      "loss": 1.139,
+      "step": 4533
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0003313839940510861,
+      "loss": 1.1538,
+      "step": 4534
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.000331352348166405,
+      "loss": 1.0823,
+      "step": 4535
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0003313206964976216,
+      "loss": 1.1453,
+      "step": 4536
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0003312890390461297,
+      "loss": 1.1354,
+      "step": 4537
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0003312573758133233,
+      "loss": 1.1578,
+      "step": 4538
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00033122570680059677,
+      "loss": 1.1023,
+      "step": 4539
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00033119403200934454,
+      "loss": 1.2184,
+      "step": 4540
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00033116235144096154,
+      "loss": 1.1549,
+      "step": 4541
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0003311306650968428,
+      "loss": 1.0877,
+      "step": 4542
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00033109897297838357,
+      "loss": 1.1136,
+      "step": 4543
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00033106727508697945,
+      "loss": 1.1601,
+      "step": 4544
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00033103557142402624,
+      "loss": 1.0812,
+      "step": 4545
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0003310038619909199,
+      "loss": 1.0655,
+      "step": 4546
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00033097214678905705,
+      "loss": 1.0419,
+      "step": 4547
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.000330940425819834,
+      "loss": 1.1923,
+      "step": 4548
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0003309086990846477,
+      "loss": 1.1565,
+      "step": 4549
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0003308769665848952,
+      "loss": 1.1474,
+      "step": 4550
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00033084522832197387,
+      "loss": 1.1016,
+      "step": 4551
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00033081348429728127,
+      "loss": 1.2168,
+      "step": 4552
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00033078173451221525,
+      "loss": 1.1344,
+      "step": 4553
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0003307499789681739,
+      "loss": 1.1004,
+      "step": 4554
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0003307182176665557,
+      "loss": 1.1763,
+      "step": 4555
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0003306864506087591,
+      "loss": 1.1024,
+      "step": 4556
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.000330654677796183,
+      "loss": 1.1189,
+      "step": 4557
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0003306228992302266,
+      "loss": 1.1279,
+      "step": 4558
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0003305911149122892,
+      "loss": 1.1229,
+      "step": 4559
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0003305593248437704,
+      "loss": 1.2084,
+      "step": 4560
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00033052752902607013,
+      "loss": 1.1455,
+      "step": 4561
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0003304957274605885,
+      "loss": 1.1362,
+      "step": 4562
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0003304639201487259,
+      "loss": 1.1245,
+      "step": 4563
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00033043210709188297,
+      "loss": 1.0906,
+      "step": 4564
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00033040028829146054,
+      "loss": 1.2154,
+      "step": 4565
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00033036846374885984,
+      "loss": 1.1339,
+      "step": 4566
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0003303366334654822,
+      "loss": 1.0839,
+      "step": 4567
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00033030479744272927,
+      "loss": 1.1531,
+      "step": 4568
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.000330272955682003,
+      "loss": 1.133,
+      "step": 4569
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0003302411081847055,
+      "loss": 1.1007,
+      "step": 4570
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00033020925495223915,
+      "loss": 1.168,
+      "step": 4571
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0003301773959860067,
+      "loss": 1.1063,
+      "step": 4572
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.000330145531287411,
+      "loss": 1.1704,
+      "step": 4573
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0003301136608578552,
+      "loss": 1.1171,
+      "step": 4574
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0003300817846987427,
+      "loss": 1.0921,
+      "step": 4575
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0003300499028114772,
+      "loss": 1.1729,
+      "step": 4576
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0003300180151974626,
+      "loss": 1.1212,
+      "step": 4577
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00032998612185810314,
+      "loss": 1.1641,
+      "step": 4578
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0003299542227948032,
+      "loss": 1.1721,
+      "step": 4579
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0003299223180089674,
+      "loss": 1.0175,
+      "step": 4580
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00032989040750200073,
+      "loss": 1.1092,
+      "step": 4581
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0003298584912753084,
+      "loss": 1.1147,
+      "step": 4582
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0003298265693302957,
+      "loss": 1.1272,
+      "step": 4583
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00032979464166836846,
+      "loss": 1.1717,
+      "step": 4584
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0003297627082909326,
+      "loss": 1.1547,
+      "step": 4585
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0003297307691993942,
+      "loss": 1.091,
+      "step": 4586
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00032969882439515984,
+      "loss": 1.053,
+      "step": 4587
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00032966687387963613,
+      "loss": 1.1812,
+      "step": 4588
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00032963491765423,
+      "loss": 1.0641,
+      "step": 4589
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0003296029557203486,
+      "loss": 1.1175,
+      "step": 4590
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0003295709880793995,
+      "loss": 1.0488,
+      "step": 4591
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0003295390147327903,
+      "loss": 1.1067,
+      "step": 4592
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.000329507035681929,
+      "loss": 1.0616,
+      "step": 4593
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0003294750509282238,
+      "loss": 1.1453,
+      "step": 4594
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0003294430604730831,
+      "loss": 1.256,
+      "step": 4595
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0003294110643179156,
+      "loss": 1.0631,
+      "step": 4596
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00032937906246413036,
+      "loss": 1.1786,
+      "step": 4597
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00032934705491313645,
+      "loss": 1.194,
+      "step": 4598
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00032931504166634334,
+      "loss": 1.1604,
+      "step": 4599
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00032928302272516077,
+      "loss": 1.1298,
+      "step": 4600
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0003292509980909987,
+      "loss": 1.1017,
+      "step": 4601
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00032921896776526733,
+      "loss": 1.1817,
+      "step": 4602
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00032918693174937717,
+      "loss": 1.1529,
+      "step": 4603
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00032915489004473877,
+      "loss": 1.0614,
+      "step": 4604
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0003291228426527632,
+      "loss": 1.1766,
+      "step": 4605
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00032909078957486175,
+      "loss": 1.0908,
+      "step": 4606
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00032905873081244574,
+      "loss": 1.1661,
+      "step": 4607
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0003290266663669268,
+      "loss": 1.1205,
+      "step": 4608
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00032899459623971715,
+      "loss": 1.1686,
+      "step": 4609
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00032896252043222875,
+      "loss": 1.0413,
+      "step": 4610
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00032893043894587423,
+      "loss": 1.1012,
+      "step": 4611
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0003288983517820662,
+      "loss": 1.0663,
+      "step": 4612
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00032886625894221764,
+      "loss": 1.1034,
+      "step": 4613
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00032883416042774185,
+      "loss": 1.1328,
+      "step": 4614
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0003288020562400521,
+      "loss": 1.1535,
+      "step": 4615
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00032876994638056226,
+      "loss": 1.0736,
+      "step": 4616
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0003287378308506862,
+      "loss": 1.1991,
+      "step": 4617
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00032870570965183815,
+      "loss": 1.1979,
+      "step": 4618
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0003286735827854326,
+      "loss": 1.1358,
+      "step": 4619
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0003286414502528843,
+      "loss": 1.1032,
+      "step": 4620
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0003286093120556081,
+      "loss": 1.1791,
+      "step": 4621
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0003285771681950192,
+      "loss": 1.0904,
+      "step": 4622
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00032854501867253316,
+      "loss": 1.1305,
+      "step": 4623
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00032851286348956554,
+      "loss": 1.1979,
+      "step": 4624
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0003284807026475325,
+      "loss": 1.1044,
+      "step": 4625
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0003284485361478501,
+      "loss": 1.0779,
+      "step": 4626
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00032841636399193473,
+      "loss": 1.117,
+      "step": 4627
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00032838418618120325,
+      "loss": 1.1661,
+      "step": 4628
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00032835200271707257,
+      "loss": 1.1489,
+      "step": 4629
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00032831981360095975,
+      "loss": 1.0928,
+      "step": 4630
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0003282876188342824,
+      "loss": 1.1266,
+      "step": 4631
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00032825541841845815,
+      "loss": 1.1481,
+      "step": 4632
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.000328223212354905,
+      "loss": 1.2615,
+      "step": 4633
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00032819100064504107,
+      "loss": 1.1302,
+      "step": 4634
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0003281587832902849,
+      "loss": 1.1588,
+      "step": 4635
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00032812656029205503,
+      "loss": 1.0964,
+      "step": 4636
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0003280943316517705,
+      "loss": 1.204,
+      "step": 4637
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00032806209737085054,
+      "loss": 1.1084,
+      "step": 4638
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0003280298574507145,
+      "loss": 1.14,
+      "step": 4639
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0003279976118927821,
+      "loss": 1.1177,
+      "step": 4640
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00032796536069847325,
+      "loss": 1.1515,
+      "step": 4641
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0003279331038692082,
+      "loss": 1.0595,
+      "step": 4642
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00032790084140640733,
+      "loss": 1.1954,
+      "step": 4643
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00032786857331149126,
+      "loss": 1.1202,
+      "step": 4644
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00032783629958588103,
+      "loss": 1.0354,
+      "step": 4645
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00032780402023099777,
+      "loss": 1.0402,
+      "step": 4646
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0003277717352482628,
+      "loss": 1.1758,
+      "step": 4647
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.000327739444639098,
+      "loss": 1.1241,
+      "step": 4648
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00032770714840492507,
+      "loss": 1.1799,
+      "step": 4649
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00032767484654716637,
+      "loss": 1.1097,
+      "step": 4650
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0003276425390672442,
+      "loss": 1.0946,
+      "step": 4651
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0003276102259665812,
+      "loss": 1.0674,
+      "step": 4652
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00032757790724660034,
+      "loss": 1.0805,
+      "step": 4653
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00032754558290872476,
+      "loss": 1.1476,
+      "step": 4654
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0003275132529543778,
+      "loss": 1.0451,
+      "step": 4655
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0003274809173849832,
+      "loss": 1.1443,
+      "step": 4656
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0003274485762019648,
+      "loss": 1.1676,
+      "step": 4657
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00032741622940674677,
+      "loss": 1.1693,
+      "step": 4658
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0003273838770007536,
+      "loss": 1.1736,
+      "step": 4659
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0003273515189854097,
+      "loss": 1.1199,
+      "step": 4660
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0003273191553621401,
+      "loss": 1.1132,
+      "step": 4661
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00032728678613236996,
+      "loss": 1.137,
+      "step": 4662
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00032725441129752454,
+      "loss": 1.1627,
+      "step": 4663
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00032722203085902964,
+      "loss": 1.1193,
+      "step": 4664
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0003271896448183109,
+      "loss": 1.1281,
+      "step": 4665
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0003271572531767947,
+      "loss": 1.0392,
+      "step": 4666
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00032712485593590726,
+      "loss": 1.1728,
+      "step": 4667
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0003270924530970751,
+      "loss": 1.17,
+      "step": 4668
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00032706004466172533,
+      "loss": 1.1563,
+      "step": 4669
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00032702763063128485,
+      "loss": 1.124,
+      "step": 4670
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00032699521100718103,
+      "loss": 1.2078,
+      "step": 4671
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0003269627857908416,
+      "loss": 1.1537,
+      "step": 4672
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0003269303549836943,
+      "loss": 1.0924,
+      "step": 4673
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00032689791858716725,
+      "loss": 1.0766,
+      "step": 4674
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00032686547660268876,
+      "loss": 1.0822,
+      "step": 4675
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00032683302903168743,
+      "loss": 1.2227,
+      "step": 4676
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0003268005758755921,
+      "loss": 1.0389,
+      "step": 4677
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00032676811713583186,
+      "loss": 1.1019,
+      "step": 4678
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.000326735652813836,
+      "loss": 1.1289,
+      "step": 4679
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0003267031829110341,
+      "loss": 1.1615,
+      "step": 4680
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00032667070742885593,
+      "loss": 1.1518,
+      "step": 4681
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00032663822636873157,
+      "loss": 1.1268,
+      "step": 4682
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0003266057397320914,
+      "loss": 1.0768,
+      "step": 4683
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0003265732475203659,
+      "loss": 1.0906,
+      "step": 4684
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0003265407497349859,
+      "loss": 1.1834,
+      "step": 4685
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00032650824637738235,
+      "loss": 1.2114,
+      "step": 4686
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00032647573744898665,
+      "loss": 1.1565,
+      "step": 4687
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0003264432229512302,
+      "loss": 1.0964,
+      "step": 4688
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.000326410702885545,
+      "loss": 1.0325,
+      "step": 4689
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0003263781772533628,
+      "loss": 1.0534,
+      "step": 4690
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00032634564605611607,
+      "loss": 1.1328,
+      "step": 4691
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00032631310929523725,
+      "loss": 1.1375,
+      "step": 4692
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00032628056697215903,
+      "loss": 1.1552,
+      "step": 4693
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0003262480190883145,
+      "loss": 1.0896,
+      "step": 4694
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00032621546564513696,
+      "loss": 1.1337,
+      "step": 4695
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00032618290664405974,
+      "loss": 1.0651,
+      "step": 4696
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00032615034208651665,
+      "loss": 1.118,
+      "step": 4697
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00032611777197394176,
+      "loss": 1.0831,
+      "step": 4698
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00032608519630776913,
+      "loss": 1.122,
+      "step": 4699
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00032605261508943334,
+      "loss": 1.122,
+      "step": 4700
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0003260200283203691,
+      "loss": 1.2217,
+      "step": 4701
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0003259874360020113,
+      "loss": 1.1256,
+      "step": 4702
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0003259548381357953,
+      "loss": 1.0603,
+      "step": 4703
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0003259222347231564,
+      "loss": 1.1113,
+      "step": 4704
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00032588962576553026,
+      "loss": 1.2601,
+      "step": 4705
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00032585701126435293,
+      "loss": 1.0849,
+      "step": 4706
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0003258243912210606,
+      "loss": 1.1919,
+      "step": 4707
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00032579176563708956,
+      "loss": 1.1897,
+      "step": 4708
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0003257591345138766,
+      "loss": 1.1498,
+      "step": 4709
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0003257264978528586,
+      "loss": 1.0831,
+      "step": 4710
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0003256938556554727,
+      "loss": 1.0214,
+      "step": 4711
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0003256612079231562,
+      "loss": 1.2447,
+      "step": 4712
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.000325628554657347,
+      "loss": 1.1492,
+      "step": 4713
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00032559589585948276,
+      "loss": 1.1661,
+      "step": 4714
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00032556323153100174,
+      "loss": 1.1895,
+      "step": 4715
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00032553056167334226,
+      "loss": 1.0978,
+      "step": 4716
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0003254978862879429,
+      "loss": 1.1232,
+      "step": 4717
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00032546520537624253,
+      "loss": 1.1555,
+      "step": 4718
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00032543251893968035,
+      "loss": 1.08,
+      "step": 4719
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00032539982697969564,
+      "loss": 1.1183,
+      "step": 4720
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00032536712949772793,
+      "loss": 1.1741,
+      "step": 4721
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00032533442649521717,
+      "loss": 1.1038,
+      "step": 4722
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00032530171797360343,
+      "loss": 1.0794,
+      "step": 4723
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00032526900393432687,
+      "loss": 1.0748,
+      "step": 4724
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00032523628437882817,
+      "loss": 1.2446,
+      "step": 4725
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00032520355930854824,
+      "loss": 1.0865,
+      "step": 4726
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00032517082872492793,
+      "loss": 1.1708,
+      "step": 4727
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00032513809262940864,
+      "loss": 1.069,
+      "step": 4728
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00032510535102343196,
+      "loss": 1.1184,
+      "step": 4729
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0003250726039084395,
+      "loss": 1.1358,
+      "step": 4730
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00032503985128587345,
+      "loss": 1.1692,
+      "step": 4731
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00032500709315717597,
+      "loss": 1.0708,
+      "step": 4732
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00032497432952378956,
+      "loss": 1.1445,
+      "step": 4733
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00032494156038715707,
+      "loss": 1.1472,
+      "step": 4734
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0003249087857487213,
+      "loss": 1.1203,
+      "step": 4735
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0003248760056099257,
+      "loss": 1.1243,
+      "step": 4736
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00032484321997221363,
+      "loss": 1.2038,
+      "step": 4737
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0003248104288370288,
+      "loss": 1.1915,
+      "step": 4738
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0003247776322058152,
+      "loss": 1.1697,
+      "step": 4739
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.000324744830080017,
+      "loss": 1.1104,
+      "step": 4740
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0003247120224610787,
+      "loss": 1.1521,
+      "step": 4741
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0003246792093504449,
+      "loss": 1.1206,
+      "step": 4742
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0003246463907495605,
+      "loss": 1.0912,
+      "step": 4743
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00032461356665987085,
+      "loss": 1.1195,
+      "step": 4744
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0003245807370828213,
+      "loss": 1.0612,
+      "step": 4745
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0003245479020198573,
+      "loss": 1.1508,
+      "step": 4746
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0003245150614724249,
+      "loss": 1.2189,
+      "step": 4747
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0003244822154419703,
+      "loss": 1.1011,
+      "step": 4748
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00032444936392993977,
+      "loss": 1.1042,
+      "step": 4749
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00032441650693777995,
+      "loss": 1.1506,
+      "step": 4750
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00032438364446693774,
+      "loss": 1.1776,
+      "step": 4751
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00032435077651886017,
+      "loss": 1.2191,
+      "step": 4752
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00032431790309499453,
+      "loss": 1.1084,
+      "step": 4753
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00032428502419678867,
+      "loss": 1.0587,
+      "step": 4754
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00032425213982569006,
+      "loss": 1.1131,
+      "step": 4755
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.000324219249983147,
+      "loss": 1.161,
+      "step": 4756
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00032418635467060773,
+      "loss": 1.0271,
+      "step": 4757
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0003241534538895208,
+      "loss": 1.109,
+      "step": 4758
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00032412054764133496,
+      "loss": 1.1598,
+      "step": 4759
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0003240876359274993,
+      "loss": 1.113,
+      "step": 4760
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.000324054718749463,
+      "loss": 1.182,
+      "step": 4761
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00032402179610867564,
+      "loss": 1.0914,
+      "step": 4762
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0003239888680065869,
+      "loss": 1.187,
+      "step": 4763
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0003239559344446469,
+      "loss": 1.0907,
+      "step": 4764
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0003239229954243058,
+      "loss": 1.0865,
+      "step": 4765
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.000323890050947014,
+      "loss": 1.1277,
+      "step": 4766
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00032385710101422225,
+      "loss": 1.129,
+      "step": 4767
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00032382414562738164,
+      "loss": 1.1233,
+      "step": 4768
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0003237911847879431,
+      "loss": 1.2267,
+      "step": 4769
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0003237582184973583,
+      "loss": 1.1315,
+      "step": 4770
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0003237252467570787,
+      "loss": 1.2191,
+      "step": 4771
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00032369226956855643,
+      "loss": 1.0563,
+      "step": 4772
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00032365928693324356,
+      "loss": 1.1325,
+      "step": 4773
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00032362629885259234,
+      "loss": 1.1779,
+      "step": 4774
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0003235933053280556,
+      "loss": 1.0886,
+      "step": 4775
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0003235603063610861,
+      "loss": 1.1101,
+      "step": 4776
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00032352730195313697,
+      "loss": 1.1419,
+      "step": 4777
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0003234942921056615,
+      "loss": 1.084,
+      "step": 4778
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0003234612768201135,
+      "loss": 1.1153,
+      "step": 4779
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0003234282560979465,
+      "loss": 1.2038,
+      "step": 4780
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0003233952299406148,
+      "loss": 1.0798,
+      "step": 4781
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0003233621983495726,
+      "loss": 1.1696,
+      "step": 4782
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0003233291613262744,
+      "loss": 1.191,
+      "step": 4783
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00032329611887217517,
+      "loss": 1.1693,
+      "step": 4784
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0003232630709887297,
+      "loss": 1.112,
+      "step": 4785
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0003232300176773935,
+      "loss": 1.1109,
+      "step": 4786
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0003231969589396218,
+      "loss": 1.0428,
+      "step": 4787
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00032316389477687055,
+      "loss": 1.1505,
+      "step": 4788
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00032313082519059563,
+      "loss": 1.1998,
+      "step": 4789
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00032309775018225334,
+      "loss": 1.1412,
+      "step": 4790
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00032306466975330004,
+      "loss": 1.2196,
+      "step": 4791
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0003230315839051925,
+      "loss": 1.1767,
+      "step": 4792
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00032299849263938774,
+      "loss": 1.116,
+      "step": 4793
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0003229653959573427,
+      "loss": 1.1663,
+      "step": 4794
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00032293229386051503,
+      "loss": 1.0462,
+      "step": 4795
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0003228991863503622,
+      "loss": 1.0555,
+      "step": 4796
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0003228660734283421,
+      "loss": 1.0408,
+      "step": 4797
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00032283295509591306,
+      "loss": 1.2012,
+      "step": 4798
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0003227998313545333,
+      "loss": 1.0992,
+      "step": 4799
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0003227667022056614,
+      "loss": 1.116,
+      "step": 4800
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0003227335676507563,
+      "loss": 1.0729,
+      "step": 4801
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0003227004276912769,
+      "loss": 1.0286,
+      "step": 4802
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00032266728232868274,
+      "loss": 1.1291,
+      "step": 4803
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00032263413156443326,
+      "loss": 1.051,
+      "step": 4804
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00032260097539998835,
+      "loss": 1.116,
+      "step": 4805
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0003225678138368079,
+      "loss": 1.1744,
+      "step": 4806
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00032253464687635226,
+      "loss": 1.0272,
+      "step": 4807
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00032250147452008186,
+      "loss": 1.0822,
+      "step": 4808
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0003224682967694576,
+      "loss": 1.185,
+      "step": 4809
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00032243511362594035,
+      "loss": 1.1675,
+      "step": 4810
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00032240192509099127,
+      "loss": 1.149,
+      "step": 4811
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00032236873116607204,
+      "loss": 1.1455,
+      "step": 4812
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00032233553185264405,
+      "loss": 1.0793,
+      "step": 4813
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00032230232715216956,
+      "loss": 1.0981,
+      "step": 4814
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0003222691170661105,
+      "loss": 1.0534,
+      "step": 4815
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00032223590159592935,
+      "loss": 1.1983,
+      "step": 4816
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00032220268074308875,
+      "loss": 1.1497,
+      "step": 4817
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0003221694545090516,
+      "loss": 1.1126,
+      "step": 4818
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.000322136222895281,
+      "loss": 1.1332,
+      "step": 4819
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00032210298590324034,
+      "loss": 1.1365,
+      "step": 4820
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0003220697435343931,
+      "loss": 1.1037,
+      "step": 4821
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00032203649579020326,
+      "loss": 1.0804,
+      "step": 4822
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0003220032426721348,
+      "loss": 1.1062,
+      "step": 4823
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00032196998418165205,
+      "loss": 1.1068,
+      "step": 4824
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0003219367203202195,
+      "loss": 1.1497,
+      "step": 4825
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.000321903451089302,
+      "loss": 1.1789,
+      "step": 4826
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00032187017649036446,
+      "loss": 1.0987,
+      "step": 4827
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0003218368965248722,
+      "loss": 1.2097,
+      "step": 4828
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0003218036111942907,
+      "loss": 1.0866,
+      "step": 4829
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0003217703205000857,
+      "loss": 1.1408,
+      "step": 4830
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0003217370244437231,
+      "loss": 1.1464,
+      "step": 4831
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0003217037230266691,
+      "loss": 1.1568,
+      "step": 4832
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0003216704162503902,
+      "loss": 1.055,
+      "step": 4833
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0003216371041163529,
+      "loss": 1.1321,
+      "step": 4834
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0003216037866260243,
+      "loss": 1.1807,
+      "step": 4835
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0003215704637808715,
+      "loss": 1.0494,
+      "step": 4836
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00032153713558236174,
+      "loss": 1.0744,
+      "step": 4837
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0003215038020319627,
+      "loss": 1.1235,
+      "step": 4838
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00032147046313114224,
+      "loss": 1.1568,
+      "step": 4839
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00032143711888136845,
+      "loss": 0.9939,
+      "step": 4840
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0003214037692841096,
+      "loss": 1.1613,
+      "step": 4841
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00032137041434083423,
+      "loss": 1.1488,
+      "step": 4842
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00032133705405301124,
+      "loss": 1.1608,
+      "step": 4843
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0003213036884221095,
+      "loss": 1.1254,
+      "step": 4844
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00032127031744959837,
+      "loss": 1.1382,
+      "step": 4845
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00032123694113694737,
+      "loss": 1.1123,
+      "step": 4846
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00032120355948562606,
+      "loss": 1.0663,
+      "step": 4847
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0003211701724971045,
+      "loss": 1.1251,
+      "step": 4848
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00032113678017285294,
+      "loss": 1.1176,
+      "step": 4849
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00032110338251434175,
+      "loss": 1.1263,
+      "step": 4850
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0003210699795230416,
+      "loss": 1.1651,
+      "step": 4851
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0003210365712004234,
+      "loss": 1.1294,
+      "step": 4852
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0003210031575479583,
+      "loss": 1.1414,
+      "step": 4853
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0003209697385671176,
+      "loss": 1.1451,
+      "step": 4854
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00032093631425937305,
+      "loss": 1.1298,
+      "step": 4855
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0003209028846261964,
+      "loss": 1.1379,
+      "step": 4856
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0003208694496690597,
+      "loss": 1.1619,
+      "step": 4857
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0003208360093894352,
+      "loss": 1.124,
+      "step": 4858
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0003208025637887956,
+      "loss": 1.1196,
+      "step": 4859
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00032076911286861364,
+      "loss": 1.1585,
+      "step": 4860
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0003207356566303622,
+      "loss": 1.0624,
+      "step": 4861
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0003207021950755147,
+      "loss": 1.1073,
+      "step": 4862
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00032066872820554456,
+      "loss": 1.1526,
+      "step": 4863
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0003206352560219254,
+      "loss": 1.2349,
+      "step": 4864
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0003206017785261313,
+      "loss": 1.1426,
+      "step": 4865
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00032056829571963636,
+      "loss": 1.058,
+      "step": 4866
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00032053480760391504,
+      "loss": 1.1076,
+      "step": 4867
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0003205013141804419,
+      "loss": 1.1367,
+      "step": 4868
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.000320467815450692,
+      "loss": 1.1337,
+      "step": 4869
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0003204343114161403,
+      "loss": 1.1188,
+      "step": 4870
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0003204008020782622,
+      "loss": 1.0797,
+      "step": 4871
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00032036728743853327,
+      "loss": 1.1002,
+      "step": 4872
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0003203337674984294,
+      "loss": 1.1156,
+      "step": 4873
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00032030024225942653,
+      "loss": 1.164,
+      "step": 4874
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.000320266711723001,
+      "loss": 1.08,
+      "step": 4875
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0003202331758906294,
+      "loss": 1.0983,
+      "step": 4876
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0003201996347637883,
+      "loss": 1.148,
+      "step": 4877
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00032016608834395486,
+      "loss": 1.2053,
+      "step": 4878
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00032013253663260623,
+      "loss": 1.2554,
+      "step": 4879
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00032009897963121984,
+      "loss": 1.1505,
+      "step": 4880
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0003200654173412734,
+      "loss": 1.1039,
+      "step": 4881
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0003200318497642448,
+      "loss": 1.0754,
+      "step": 4882
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00031999827690161226,
+      "loss": 1.1343,
+      "step": 4883
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00031996469875485406,
+      "loss": 1.1098,
+      "step": 4884
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0003199311153254488,
+      "loss": 1.1678,
+      "step": 4885
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0003198975266148755,
+      "loss": 1.0977,
+      "step": 4886
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00031986393262461305,
+      "loss": 1.0629,
+      "step": 4887
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0003198303333561409,
+      "loss": 1.0567,
+      "step": 4888
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0003197967288109385,
+      "loss": 1.196,
+      "step": 4889
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0003197631189904857,
+      "loss": 1.144,
+      "step": 4890
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0003197295038962624,
+      "loss": 1.1431,
+      "step": 4891
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.000319695883529749,
+      "loss": 1.1225,
+      "step": 4892
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00031966225789242575,
+      "loss": 1.0965,
+      "step": 4893
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0003196286269857736,
+      "loss": 1.1176,
+      "step": 4894
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00031959499081127334,
+      "loss": 1.1392,
+      "step": 4895
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0003195613493704062,
+      "loss": 1.1721,
+      "step": 4896
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00031952770266465346,
+      "loss": 1.1465,
+      "step": 4897
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.000319494050695497,
+      "loss": 1.0847,
+      "step": 4898
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0003194603934644184,
+      "loss": 1.0591,
+      "step": 4899
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0003194267309729,
+      "loss": 1.1019,
+      "step": 4900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.000319393063222424,
+      "loss": 1.1744,
+      "step": 4901
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00031935939021447295,
+      "loss": 1.1946,
+      "step": 4902
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00031932571195052964,
+      "loss": 1.2018,
+      "step": 4903
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00031929202843207716,
+      "loss": 1.0396,
+      "step": 4904
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00031925833966059874,
+      "loss": 1.1411,
+      "step": 4905
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0003192246456375778,
+      "loss": 1.1954,
+      "step": 4906
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00031919094636449817,
+      "loss": 1.1121,
+      "step": 4907
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0003191572418428437,
+      "loss": 1.1147,
+      "step": 4908
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00031912353207409863,
+      "loss": 1.0433,
+      "step": 4909
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00031908981705974733,
+      "loss": 1.1848,
+      "step": 4910
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0003190560968012744,
+      "loss": 1.158,
+      "step": 4911
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00031902237130016485,
+      "loss": 1.2432,
+      "step": 4912
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0003189886405579038,
+      "loss": 1.1669,
+      "step": 4913
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0003189549045759763,
+      "loss": 1.1095,
+      "step": 4914
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00031892116335586817,
+      "loss": 1.0592,
+      "step": 4915
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00031888741689906514,
+      "loss": 1.1518,
+      "step": 4916
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00031885366520705325,
+      "loss": 1.1728,
+      "step": 4917
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0003188199082813188,
+      "loss": 1.2745,
+      "step": 4918
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00031878614612334806,
+      "loss": 1.1012,
+      "step": 4919
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00031875237873462797,
+      "loss": 1.1075,
+      "step": 4920
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0003187186061166454,
+      "loss": 1.1838,
+      "step": 4921
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00031868482827088755,
+      "loss": 1.1467,
+      "step": 4922
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00031865104519884183,
+      "loss": 1.1537,
+      "step": 4923
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00031861725690199587,
+      "loss": 1.1473,
+      "step": 4924
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00031858346338183745,
+      "loss": 1.0823,
+      "step": 4925
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0003185496646398548,
+      "loss": 1.0757,
+      "step": 4926
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0003185158606775362,
+      "loss": 1.1554,
+      "step": 4927
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00031848205149637024,
+      "loss": 1.1496,
+      "step": 4928
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0003184482370978456,
+      "loss": 1.1312,
+      "step": 4929
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00031841441748345145,
+      "loss": 1.1113,
+      "step": 4930
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0003183805926546769,
+      "loss": 1.1191,
+      "step": 4931
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00031834676261301154,
+      "loss": 1.1508,
+      "step": 4932
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0003183129273599449,
+      "loss": 1.0848,
+      "step": 4933
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00031827908689696714,
+      "loss": 1.1171,
+      "step": 4934
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0003182452412255683,
+      "loss": 1.1782,
+      "step": 4935
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0003182113903472388,
+      "loss": 1.1524,
+      "step": 4936
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00031817753426346924,
+      "loss": 1.1398,
+      "step": 4937
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0003181436729757505,
+      "loss": 1.1765,
+      "step": 4938
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00031810980648557365,
+      "loss": 1.1518,
+      "step": 4939
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00031807593479443,
+      "loss": 1.1852,
+      "step": 4940
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00031804205790381113,
+      "loss": 1.0462,
+      "step": 4941
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00031800817581520874,
+      "loss": 1.0061,
+      "step": 4942
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00031797428853011484,
+      "loss": 1.0973,
+      "step": 4943
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0003179403960500217,
+      "loss": 1.1634,
+      "step": 4944
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0003179064983764218,
+      "loss": 1.1575,
+      "step": 4945
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0003178725955108078,
+      "loss": 1.119,
+      "step": 4946
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00031783868745467254,
+      "loss": 1.1449,
+      "step": 4947
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00031780477420950915,
+      "loss": 1.2047,
+      "step": 4948
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00031777085577681117,
+      "loss": 1.1218,
+      "step": 4949
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0003177369321580721,
+      "loss": 1.0994,
+      "step": 4950
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0003177030033547857,
+      "loss": 1.1745,
+      "step": 4951
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00031766906936844607,
+      "loss": 1.0911,
+      "step": 4952
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0003176351302005476,
+      "loss": 1.1354,
+      "step": 4953
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0003176011858525846,
+      "loss": 1.1217,
+      "step": 4954
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00031756723632605203,
+      "loss": 1.1186,
+      "step": 4955
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0003175332816224447,
+      "loss": 1.2047,
+      "step": 4956
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00031749932174325785,
+      "loss": 1.0796,
+      "step": 4957
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00031746535668998696,
+      "loss": 1.1266,
+      "step": 4958
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00031743138646412754,
+      "loss": 1.1442,
+      "step": 4959
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0003173974110671756,
+      "loss": 1.1681,
+      "step": 4960
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00031736343050062726,
+      "loss": 1.0274,
+      "step": 4961
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0003173294447659787,
+      "loss": 1.1184,
+      "step": 4962
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00031729545386472666,
+      "loss": 1.0743,
+      "step": 4963
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00031726145779836784,
+      "loss": 1.217,
+      "step": 4964
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00031722745656839925,
+      "loss": 1.1623,
+      "step": 4965
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00031719345017631816,
+      "loss": 1.1441,
+      "step": 4966
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0003171594386236221,
+      "loss": 1.1087,
+      "step": 4967
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0003171254219118086,
+      "loss": 1.1153,
+      "step": 4968
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0003170914000423758,
+      "loss": 1.0991,
+      "step": 4969
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0003170573730168216,
+      "loss": 1.0734,
+      "step": 4970
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00031702334083664465,
+      "loss": 1.1965,
+      "step": 4971
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00031698930350334344,
+      "loss": 1.1429,
+      "step": 4972
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00031695526101841674,
+      "loss": 1.0825,
+      "step": 4973
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00031692121338336367,
+      "loss": 1.1081,
+      "step": 4974
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0003168871605996835,
+      "loss": 1.1326,
+      "step": 4975
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00031685310266887587,
+      "loss": 1.1262,
+      "step": 4976
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0003168190395924403,
+      "loss": 1.1056,
+      "step": 4977
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0003167849713718769,
+      "loss": 1.1526,
+      "step": 4978
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0003167508980086859,
+      "loss": 1.1473,
+      "step": 4979
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00031671681950436756,
+      "loss": 1.1455,
+      "step": 4980
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0003166827358604227,
+      "loss": 1.1185,
+      "step": 4981
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00031664864707835205,
+      "loss": 1.1363,
+      "step": 4982
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0003166145531596569,
+      "loss": 1.1428,
+      "step": 4983
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00031658045410583834,
+      "loss": 1.1703,
+      "step": 4984
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00031654634991839804,
+      "loss": 1.1225,
+      "step": 4985
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00031651224059883786,
+      "loss": 1.0971,
+      "step": 4986
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00031647812614865963,
+      "loss": 1.1157,
+      "step": 4987
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0003164440065693657,
+      "loss": 1.1553,
+      "step": 4988
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0003164098818624585,
+      "loss": 1.1,
+      "step": 4989
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0003163757520294407,
+      "loss": 1.1183,
+      "step": 4990
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00031634161707181523,
+      "loss": 1.0882,
+      "step": 4991
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0003163074769910852,
+      "loss": 1.1186,
+      "step": 4992
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00031627333178875397,
+      "loss": 1.1097,
+      "step": 4993
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00031623918146632515,
+      "loss": 1.0686,
+      "step": 4994
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00031620502602530253,
+      "loss": 1.0511,
+      "step": 4995
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0003161708654671901,
+      "loss": 1.0893,
+      "step": 4996
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00031613669979349226,
+      "loss": 1.1724,
+      "step": 4997
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00031610252900571333,
+      "loss": 1.1675,
+      "step": 4998
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00031606835310535813,
+      "loss": 1.149,
+      "step": 4999
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00031603417209393153,
+      "loss": 1.1833,
+      "step": 5000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00031599998597293873,
+      "loss": 1.1524,
+      "step": 5001
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00031596579474388515,
+      "loss": 1.1257,
+      "step": 5002
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0003159315984082763,
+      "loss": 1.1482,
+      "step": 5003
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0003158973969676181,
+      "loss": 1.1032,
+      "step": 5004
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00031586319042341656,
+      "loss": 1.1444,
+      "step": 5005
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00031582897877717805,
+      "loss": 1.1556,
+      "step": 5006
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.000315794762030409,
+      "loss": 1.1464,
+      "step": 5007
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0003157605401846161,
+      "loss": 1.1419,
+      "step": 5008
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00031572631324130643,
+      "loss": 1.071,
+      "step": 5009
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00031569208120198715,
+      "loss": 1.1451,
+      "step": 5010
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0003156578440681656,
+      "loss": 1.1935,
+      "step": 5011
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00031562360184134945,
+      "loss": 1.1064,
+      "step": 5012
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0003155893545230466,
+      "loss": 1.147,
+      "step": 5013
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00031555510211476507,
+      "loss": 1.1846,
+      "step": 5014
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00031552084461801313,
+      "loss": 1.183,
+      "step": 5015
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00031548658203429947,
+      "loss": 1.2222,
+      "step": 5016
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0003154523143651327,
+      "loss": 1.0723,
+      "step": 5017
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00031541804161202175,
+      "loss": 1.1108,
+      "step": 5018
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00031538376377647596,
+      "loss": 1.1618,
+      "step": 5019
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0003153494808600047,
+      "loss": 1.0648,
+      "step": 5020
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00031531519286411757,
+      "loss": 1.1166,
+      "step": 5021
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00031528089979032464,
+      "loss": 1.0491,
+      "step": 5022
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00031524660164013575,
+      "loss": 1.1438,
+      "step": 5023
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00031521229841506137,
+      "loss": 1.0553,
+      "step": 5024
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0003151779901166119,
+      "loss": 1.1184,
+      "step": 5025
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0003151436767462983,
+      "loss": 1.0863,
+      "step": 5026
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0003151093583056314,
+      "loss": 1.0654,
+      "step": 5027
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0003150750347961226,
+      "loss": 1.1074,
+      "step": 5028
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0003150407062192831,
+      "loss": 1.1638,
+      "step": 5029
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0003150063725766247,
+      "loss": 1.0753,
+      "step": 5030
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00031497203386965925,
+      "loss": 1.0687,
+      "step": 5031
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0003149376900998989,
+      "loss": 1.15,
+      "step": 5032
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0003149033412688559,
+      "loss": 1.1786,
+      "step": 5033
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0003148689873780429,
+      "loss": 1.0999,
+      "step": 5034
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00031483462842897254,
+      "loss": 1.1955,
+      "step": 5035
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0003148002644231579,
+      "loss": 1.178,
+      "step": 5036
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0003147658953621122,
+      "loss": 1.1389,
+      "step": 5037
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00031473152124734895,
+      "loss": 1.1427,
+      "step": 5038
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00031469714208038165,
+      "loss": 1.127,
+      "step": 5039
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0003146627578627243,
+      "loss": 1.0816,
+      "step": 5040
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00031462836859589096,
+      "loss": 1.1732,
+      "step": 5041
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.000314593974281396,
+      "loss": 1.0877,
+      "step": 5042
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00031455957492075397,
+      "loss": 1.0727,
+      "step": 5043
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0003145251705154797,
+      "loss": 1.1266,
+      "step": 5044
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.000314490761067088,
+      "loss": 1.0975,
+      "step": 5045
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00031445634657709424,
+      "loss": 1.1356,
+      "step": 5046
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0003144219270470139,
+      "loss": 1.1518,
+      "step": 5047
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00031438750247836264,
+      "loss": 1.1062,
+      "step": 5048
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00031435307287265617,
+      "loss": 1.183,
+      "step": 5049
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00031431863823141076,
+      "loss": 1.1387,
+      "step": 5050
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0003142841985561427,
+      "loss": 1.1239,
+      "step": 5051
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00031424975384836855,
+      "loss": 1.1008,
+      "step": 5052
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.000314215304109605,
+      "loss": 1.149,
+      "step": 5053
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0003141808493413692,
+      "loss": 1.1596,
+      "step": 5054
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0003141463895451783,
+      "loss": 1.101,
+      "step": 5055
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0003141119247225496,
+      "loss": 1.1666,
+      "step": 5056
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00031407745487500096,
+      "loss": 1.1606,
+      "step": 5057
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0003140429800040502,
+      "loss": 1.0941,
+      "step": 5058
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0003140085001112153,
+      "loss": 1.1296,
+      "step": 5059
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0003139740151980148,
+      "loss": 1.0682,
+      "step": 5060
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0003139395252659671,
+      "loss": 1.1118,
+      "step": 5061
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00031390503031659094,
+      "loss": 1.0861,
+      "step": 5062
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00031387053035140537,
+      "loss": 1.2117,
+      "step": 5063
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00031383602537192966,
+      "loss": 1.0923,
+      "step": 5064
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00031380151537968305,
+      "loss": 1.1288,
+      "step": 5065
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0003137670003761854,
+      "loss": 1.1481,
+      "step": 5066
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0003137324803629564,
+      "loss": 1.1046,
+      "step": 5067
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0003136979553415162,
+      "loss": 1.1663,
+      "step": 5068
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00031366342531338515,
+      "loss": 1.1013,
+      "step": 5069
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0003136288902800838,
+      "loss": 1.1105,
+      "step": 5070
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0003135943502431328,
+      "loss": 1.1119,
+      "step": 5071
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0003135598052040532,
+      "loss": 1.1182,
+      "step": 5072
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0003135252551643662,
+      "loss": 1.1376,
+      "step": 5073
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00031349070012559317,
+      "loss": 1.0618,
+      "step": 5074
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0003134561400892557,
+      "loss": 1.0446,
+      "step": 5075
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0003134215750568758,
+      "loss": 1.1287,
+      "step": 5076
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0003133870050299753,
+      "loss": 1.0936,
+      "step": 5077
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0003133524300100767,
+      "loss": 1.1256,
+      "step": 5078
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0003133178499987024,
+      "loss": 1.1528,
+      "step": 5079
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00031328326499737526,
+      "loss": 1.0433,
+      "step": 5080
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00031324867500761807,
+      "loss": 1.1054,
+      "step": 5081
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.000313214080030954,
+      "loss": 1.0254,
+      "step": 5082
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00031317948006890663,
+      "loss": 1.1519,
+      "step": 5083
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0003131448751229994,
+      "loss": 1.1749,
+      "step": 5084
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0003131102651947562,
+      "loss": 1.1746,
+      "step": 5085
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0003130756502857011,
+      "loss": 1.0205,
+      "step": 5086
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00031304103039735826,
+      "loss": 1.056,
+      "step": 5087
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0003130064055312523,
+      "loss": 1.1359,
+      "step": 5088
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0003129717756889079,
+      "loss": 1.2061,
+      "step": 5089
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0003129371408718499,
+      "loss": 1.1808,
+      "step": 5090
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0003129025010816035,
+      "loss": 1.0497,
+      "step": 5091
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00031286785631969407,
+      "loss": 1.0758,
+      "step": 5092
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0003128332065876472,
+      "loss": 1.163,
+      "step": 5093
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0003127985518869887,
+      "loss": 1.0868,
+      "step": 5094
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0003127638922192445,
+      "loss": 1.161,
+      "step": 5095
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0003127292275859409,
+      "loss": 1.0584,
+      "step": 5096
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00031269455798860446,
+      "loss": 1.1831,
+      "step": 5097
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00031265988342876175,
+      "loss": 1.137,
+      "step": 5098
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0003126252039079396,
+      "loss": 1.1962,
+      "step": 5099
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00031259051942766523,
+      "loss": 1.1261,
+      "step": 5100
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00031255582998946594,
+      "loss": 1.2096,
+      "step": 5101
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0003125211355948692,
+      "loss": 1.1177,
+      "step": 5102
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00031248643624540303,
+      "loss": 1.1429,
+      "step": 5103
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0003124517319425952,
+      "loss": 1.089,
+      "step": 5104
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00031241702268797384,
+      "loss": 1.1085,
+      "step": 5105
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0003123823084830676,
+      "loss": 1.1213,
+      "step": 5106
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.000312347589329405,
+      "loss": 1.0758,
+      "step": 5107
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00031231286522851484,
+      "loss": 1.097,
+      "step": 5108
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0003122781361819263,
+      "loss": 1.0977,
+      "step": 5109
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00031224340219116863,
+      "loss": 1.2011,
+      "step": 5110
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0003122086632577714,
+      "loss": 1.0971,
+      "step": 5111
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00031217391938326424,
+      "loss": 1.1132,
+      "step": 5112
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0003121391705691772,
+      "loss": 1.1341,
+      "step": 5113
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00031210441681704036,
+      "loss": 1.0647,
+      "step": 5114
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00031206965812838414,
+      "loss": 1.1703,
+      "step": 5115
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00031203489450473904,
+      "loss": 1.1231,
+      "step": 5116
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0003120001259476361,
+      "loss": 1.1435,
+      "step": 5117
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0003119653524586062,
+      "loss": 1.0743,
+      "step": 5118
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00031193057403918064,
+      "loss": 1.209,
+      "step": 5119
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0003118957906908908,
+      "loss": 1.1589,
+      "step": 5120
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0003118610024152684,
+      "loss": 1.1151,
+      "step": 5121
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0003118262092138455,
+      "loss": 1.0725,
+      "step": 5122
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0003117914110881539,
+      "loss": 1.1654,
+      "step": 5123
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00031175660803972626,
+      "loss": 1.1232,
+      "step": 5124
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.000311721800070095,
+      "loss": 1.0652,
+      "step": 5125
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0003116869871807929,
+      "loss": 1.1323,
+      "step": 5126
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00031165216937335284,
+      "loss": 1.1002,
+      "step": 5127
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00031161734664930825,
+      "loss": 1.0779,
+      "step": 5128
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00031158251901019227,
+      "loss": 1.1288,
+      "step": 5129
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0003115476864575388,
+      "loss": 1.1721,
+      "step": 5130
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0003115128489928815,
+      "loss": 1.0953,
+      "step": 5131
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0003114780066177545,
+      "loss": 1.1169,
+      "step": 5132
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0003114431593336922,
+      "loss": 0.9992,
+      "step": 5133
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00031140830714222886,
+      "loss": 1.1117,
+      "step": 5134
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0003113734500448994,
+      "loss": 1.167,
+      "step": 5135
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0003113385880432387,
+      "loss": 1.0493,
+      "step": 5136
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00031130372113878185,
+      "loss": 1.0668,
+      "step": 5137
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00031126884933306424,
+      "loss": 1.1337,
+      "step": 5138
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00031123397262762157,
+      "loss": 1.0985,
+      "step": 5139
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00031119909102398947,
+      "loss": 1.1307,
+      "step": 5140
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00031116420452370397,
+      "loss": 1.1712,
+      "step": 5141
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0003111293131283015,
+      "loss": 1.1487,
+      "step": 5142
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00031109441683931815,
+      "loss": 1.1538,
+      "step": 5143
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00031105951565829084,
+      "loss": 1.097,
+      "step": 5144
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0003110246095867565,
+      "loss": 1.0643,
+      "step": 5145
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00031098969862625197,
+      "loss": 1.1539,
+      "step": 5146
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0003109547827783147,
+      "loss": 1.129,
+      "step": 5147
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00031091986204448223,
+      "loss": 1.0149,
+      "step": 5148
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00031088493642629224,
+      "loss": 1.0718,
+      "step": 5149
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0003108500059252827,
+      "loss": 1.0422,
+      "step": 5150
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0003108150705429918,
+      "loss": 1.0454,
+      "step": 5151
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0003107801302809578,
+      "loss": 1.1087,
+      "step": 5152
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0003107451851407195,
+      "loss": 1.1103,
+      "step": 5153
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00031071023512381556,
+      "loss": 1.0669,
+      "step": 5154
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00031067528023178505,
+      "loss": 1.1273,
+      "step": 5155
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0003106403204661672,
+      "loss": 1.0185,
+      "step": 5156
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0003106053558285014,
+      "loss": 1.1243,
+      "step": 5157
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0003105703863203275,
+      "loss": 1.1464,
+      "step": 5158
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00031053541194318515,
+      "loss": 1.0822,
+      "step": 5159
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00031050043269861464,
+      "loss": 1.0679,
+      "step": 5160
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0003104654485881562,
+      "loss": 1.098,
+      "step": 5161
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0003104304596133503,
+      "loss": 1.0839,
+      "step": 5162
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0003103954657757378,
+      "loss": 1.1087,
+      "step": 5163
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00031036046707685964,
+      "loss": 1.0984,
+      "step": 5164
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00031032546351825684,
+      "loss": 1.0316,
+      "step": 5165
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0003102904551014709,
+      "loss": 1.0948,
+      "step": 5166
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00031025544182804345,
+      "loss": 1.0421,
+      "step": 5167
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00031022042369951617,
+      "loss": 1.0534,
+      "step": 5168
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00031018540071743123,
+      "loss": 1.0147,
+      "step": 5169
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0003101503728833308,
+      "loss": 1.0336,
+      "step": 5170
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0003101153401987573,
+      "loss": 1.0004,
+      "step": 5171
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0003100803026652535,
+      "loss": 1.0522,
+      "step": 5172
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00031004526028436216,
+      "loss": 0.9504,
+      "step": 5173
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00031001021305762644,
+      "loss": 1.0584,
+      "step": 5174
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0003099751609865896,
+      "loss": 1.0288,
+      "step": 5175
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00030994010407279517,
+      "loss": 1.0052,
+      "step": 5176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0003099050423177869,
+      "loss": 1.059,
+      "step": 5177
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0003098699757231087,
+      "loss": 1.0818,
+      "step": 5178
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00030983490429030484,
+      "loss": 1.0256,
+      "step": 5179
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0003097998280209195,
+      "loss": 1.1158,
+      "step": 5180
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0003097647469164975,
+      "loss": 1.0449,
+      "step": 5181
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00030972966097858335,
+      "loss": 1.1083,
+      "step": 5182
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0003096945702087223,
+      "loss": 1.0558,
+      "step": 5183
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0003096594746084595,
+      "loss": 1.059,
+      "step": 5184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0003096243741793404,
+      "loss": 1.0351,
+      "step": 5185
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0003095892689229106,
+      "loss": 1.0268,
+      "step": 5186
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00030955415884071594,
+      "loss": 1.0953,
+      "step": 5187
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00030951904393430256,
+      "loss": 1.125,
+      "step": 5188
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00030948392420521674,
+      "loss": 1.0493,
+      "step": 5189
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00030944879965500496,
+      "loss": 1.0725,
+      "step": 5190
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0003094136702852139,
+      "loss": 1.0228,
+      "step": 5191
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00030937853609739057,
+      "loss": 1.1238,
+      "step": 5192
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00030934339709308195,
+      "loss": 1.0757,
+      "step": 5193
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0003093082532738355,
+      "loss": 1.0752,
+      "step": 5194
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00030927310464119885,
+      "loss": 1.0896,
+      "step": 5195
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0003092379511967196,
+      "loss": 1.0118,
+      "step": 5196
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00030920279294194584,
+      "loss": 1.0319,
+      "step": 5197
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00030916762987842573,
+      "loss": 1.1237,
+      "step": 5198
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0003091324620077077,
+      "loss": 1.1383,
+      "step": 5199
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0003090972893313404,
+      "loss": 0.9826,
+      "step": 5200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0003090621118508725,
+      "loss": 1.0755,
+      "step": 5201
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0003090269295678532,
+      "loss": 1.0882,
+      "step": 5202
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0003089917424838317,
+      "loss": 1.0526,
+      "step": 5203
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00030895655060035745,
+      "loss": 1.0964,
+      "step": 5204
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00030892135391898015,
+      "loss": 1.1007,
+      "step": 5205
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0003088861524412497,
+      "loss": 1.0437,
+      "step": 5206
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0003088509461687162,
+      "loss": 1.1204,
+      "step": 5207
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0003088157351029299,
+      "loss": 1.1145,
+      "step": 5208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0003087805192454414,
+      "loss": 1.0455,
+      "step": 5209
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0003087452985978013,
+      "loss": 1.0509,
+      "step": 5210
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00030871007316156065,
+      "loss": 1.0511,
+      "step": 5211
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0003086748429382706,
+      "loss": 1.0748,
+      "step": 5212
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00030863960792948245,
+      "loss": 1.1363,
+      "step": 5213
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0003086043681367479,
+      "loss": 1.057,
+      "step": 5214
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0003085691235616186,
+      "loss": 1.0904,
+      "step": 5215
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0003085338742056466,
+      "loss": 1.0711,
+      "step": 5216
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00030849862007038413,
+      "loss": 1.0122,
+      "step": 5217
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00030846336115738354,
+      "loss": 1.0498,
+      "step": 5218
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00030842809746819754,
+      "loss": 1.0741,
+      "step": 5219
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0003083928290043789,
+      "loss": 1.1196,
+      "step": 5220
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0003083575557674807,
+      "loss": 1.0558,
+      "step": 5221
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00030832227775905613,
+      "loss": 1.0919,
+      "step": 5222
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0003082869949806588,
+      "loss": 1.081,
+      "step": 5223
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0003082517074338422,
+      "loss": 1.0704,
+      "step": 5224
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0003082164151201604,
+      "loss": 1.0983,
+      "step": 5225
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0003081811180411674,
+      "loss": 1.0483,
+      "step": 5226
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0003081458161984175,
+      "loss": 1.0686,
+      "step": 5227
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0003081105095934653,
+      "loss": 1.0126,
+      "step": 5228
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00030807519822786543,
+      "loss": 1.1136,
+      "step": 5229
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00030803988210317286,
+      "loss": 1.0917,
+      "step": 5230
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0003080045612209427,
+      "loss": 1.1463,
+      "step": 5231
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0003079692355827304,
+      "loss": 1.0987,
+      "step": 5232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0003079339051900915,
+      "loss": 1.0214,
+      "step": 5233
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0003078985700445817,
+      "loss": 1.1024,
+      "step": 5234
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00030786323014775704,
+      "loss": 1.1394,
+      "step": 5235
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00030782788550117373,
+      "loss": 1.1179,
+      "step": 5236
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0003077925361063881,
+      "loss": 1.0715,
+      "step": 5237
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0003077571819649568,
+      "loss": 1.1143,
+      "step": 5238
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00030772182307843674,
+      "loss": 1.0782,
+      "step": 5239
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00030768645944838477,
+      "loss": 1.0366,
+      "step": 5240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0003076510910763582,
+      "loss": 1.012,
+      "step": 5241
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00030761571796391457,
+      "loss": 1.0722,
+      "step": 5242
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0003075803401126114,
+      "loss": 1.0905,
+      "step": 5243
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00030754495752400673,
+      "loss": 1.0033,
+      "step": 5244
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00030750957019965846,
+      "loss": 1.058,
+      "step": 5245
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0003074741781411249,
+      "loss": 1.152,
+      "step": 5246
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00030743878134996454,
+      "loss": 1.0283,
+      "step": 5247
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0003074033798277362,
+      "loss": 1.0541,
+      "step": 5248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0003073679735759987,
+      "loss": 1.0041,
+      "step": 5249
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00030733256259631107,
+      "loss": 1.0621,
+      "step": 5250
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0003072971468902328,
+      "loss": 1.0099,
+      "step": 5251
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0003072617264593233,
+      "loss": 1.0609,
+      "step": 5252
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00030722630130514226,
+      "loss": 1.0474,
+      "step": 5253
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00030719087142924977,
+      "loss": 1.1073,
+      "step": 5254
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.000307155436833206,
+      "loss": 1.0945,
+      "step": 5255
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0003071199975185712,
+      "loss": 1.129,
+      "step": 5256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00030708455348690594,
+      "loss": 1.1049,
+      "step": 5257
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00030704910473977105,
+      "loss": 0.9793,
+      "step": 5258
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00030701365127872753,
+      "loss": 1.0983,
+      "step": 5259
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00030697819310533653,
+      "loss": 1.037,
+      "step": 5260
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00030694273022115946,
+      "loss": 1.026,
+      "step": 5261
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00030690726262775795,
+      "loss": 1.0008,
+      "step": 5262
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0003068717903266938,
+      "loss": 1.0288,
+      "step": 5263
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.000306836313319529,
+      "loss": 1.0404,
+      "step": 5264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0003068008316078259,
+      "loss": 1.0675,
+      "step": 5265
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0003067653451931468,
+      "loss": 1.1112,
+      "step": 5266
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0003067298540770544,
+      "loss": 1.092,
+      "step": 5267
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00030669435826111163,
+      "loss": 1.1377,
+      "step": 5268
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00030665885774688135,
+      "loss": 1.0948,
+      "step": 5269
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00030662335253592705,
+      "loss": 1.0781,
+      "step": 5270
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.000306587842629812,
+      "loss": 1.0293,
+      "step": 5271
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0003065523280301,
+      "loss": 1.0479,
+      "step": 5272
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.000306516808738355,
+      "loss": 1.0682,
+      "step": 5273
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00030648128475614086,
+      "loss": 1.1027,
+      "step": 5274
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00030644575608502214,
+      "loss": 1.0619,
+      "step": 5275
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00030641022272656323,
+      "loss": 1.0876,
+      "step": 5276
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00030637468468232884,
+      "loss": 1.096,
+      "step": 5277
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0003063391419538838,
+      "loss": 1.1396,
+      "step": 5278
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00030630359454279337,
+      "loss": 1.1034,
+      "step": 5279
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0003062680424506228,
+      "loss": 1.0504,
+      "step": 5280
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0003062324856789378,
+      "loss": 1.1283,
+      "step": 5281
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0003061969242293039,
+      "loss": 1.0392,
+      "step": 5282
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00030616135810328704,
+      "loss": 1.1042,
+      "step": 5283
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00030612578730245356,
+      "loss": 1.0485,
+      "step": 5284
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00030609021182836964,
+      "loss": 1.144,
+      "step": 5285
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.000306054631682602,
+      "loss": 1.0731,
+      "step": 5286
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0003060190468667173,
+      "loss": 1.0944,
+      "step": 5287
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00030598345738228256,
+      "loss": 1.0336,
+      "step": 5288
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0003059478632308649,
+      "loss": 1.0865,
+      "step": 5289
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00030591226441403194,
+      "loss": 1.1004,
+      "step": 5290
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0003058766609333509,
+      "loss": 1.0268,
+      "step": 5291
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0003058410527903899,
+      "loss": 1.0677,
+      "step": 5292
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00030580543998671685,
+      "loss": 1.0831,
+      "step": 5293
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00030576982252389985,
+      "loss": 1.0676,
+      "step": 5294
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0003057342004035075,
+      "loss": 1.0451,
+      "step": 5295
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00030569857362710824,
+      "loss": 1.0575,
+      "step": 5296
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00030566294219627093,
+      "loss": 1.1163,
+      "step": 5297
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00030562730611256476,
+      "loss": 1.0453,
+      "step": 5298
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00030559166537755885,
+      "loss": 1.0657,
+      "step": 5299
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00030555601999282257,
+      "loss": 1.1535,
+      "step": 5300
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00030552036995992575,
+      "loss": 1.0657,
+      "step": 5301
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.000305484715280438,
+      "loss": 1.0334,
+      "step": 5302
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0003054490559559295,
+      "loss": 1.1204,
+      "step": 5303
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00030541339198797066,
+      "loss": 1.1584,
+      "step": 5304
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00030537772337813166,
+      "loss": 1.118,
+      "step": 5305
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0003053420501279834,
+      "loss": 1.0706,
+      "step": 5306
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0003053063722390966,
+      "loss": 1.0536,
+      "step": 5307
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0003052706897130424,
+      "loss": 1.1627,
+      "step": 5308
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0003052350025513921,
+      "loss": 0.9961,
+      "step": 5309
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0003051993107557171,
+      "loss": 1.0994,
+      "step": 5310
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00030516361432758913,
+      "loss": 1.0743,
+      "step": 5311
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0003051279132685802,
+      "loss": 1.1619,
+      "step": 5312
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0003050922075802623,
+      "loss": 1.0918,
+      "step": 5313
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0003050564972642076,
+      "loss": 1.1,
+      "step": 5314
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00030502078232198885,
+      "loss": 1.0781,
+      "step": 5315
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00030498506275517857,
+      "loss": 1.1864,
+      "step": 5316
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0003049493385653498,
+      "loss": 1.0485,
+      "step": 5317
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00030491360975407565,
+      "loss": 1.0676,
+      "step": 5318
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00030487787632292937,
+      "loss": 1.1015,
+      "step": 5319
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0003048421382734845,
+      "loss": 1.0915,
+      "step": 5320
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0003048063956073147,
+      "loss": 1.0229,
+      "step": 5321
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00030477064832599407,
+      "loss": 1.0179,
+      "step": 5322
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0003047348964310966,
+      "loss": 1.0575,
+      "step": 5323
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0003046991399241966,
+      "loss": 1.1066,
+      "step": 5324
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00030466337880686873,
+      "loss": 1.0325,
+      "step": 5325
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00030462761308068775,
+      "loss": 1.0776,
+      "step": 5326
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0003045918427472284,
+      "loss": 1.0012,
+      "step": 5327
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00030455606780806595,
+      "loss": 1.0599,
+      "step": 5328
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00030452028826477584,
+      "loss": 1.0513,
+      "step": 5329
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00030448450411893355,
+      "loss": 1.0807,
+      "step": 5330
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00030444871537211476,
+      "loss": 1.0618,
+      "step": 5331
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0003044129220258956,
+      "loss": 1.118,
+      "step": 5332
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.000304377124081852,
+      "loss": 1.1005,
+      "step": 5333
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0003043413215415605,
+      "loss": 1.0986,
+      "step": 5334
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0003043055144065976,
+      "loss": 1.2012,
+      "step": 5335
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00030426970267854,
+      "loss": 1.1827,
+      "step": 5336
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0003042338863589649,
+      "loss": 1.0759,
+      "step": 5337
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0003041980654494493,
+      "loss": 1.0621,
+      "step": 5338
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0003041622399515705,
+      "loss": 1.019,
+      "step": 5339
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00030412640986690625,
+      "loss": 1.0455,
+      "step": 5340
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0003040905751970342,
+      "loss": 1.0965,
+      "step": 5341
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00030405473594353235,
+      "loss": 1.0066,
+      "step": 5342
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00030401889210797894,
+      "loss": 1.1376,
+      "step": 5343
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0003039830436919524,
+      "loss": 1.0483,
+      "step": 5344
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00030394719069703124,
+      "loss": 1.0747,
+      "step": 5345
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0003039113331247941,
+      "loss": 1.065,
+      "step": 5346
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0003038754709768202,
+      "loss": 1.0876,
+      "step": 5347
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0003038396042546887,
+      "loss": 1.0367,
+      "step": 5348
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00030380373295997885,
+      "loss": 1.0478,
+      "step": 5349
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0003037678570942704,
+      "loss": 0.9923,
+      "step": 5350
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00030373197665914305,
+      "loss": 1.1078,
+      "step": 5351
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0003036960916561768,
+      "loss": 1.0759,
+      "step": 5352
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0003036602020869519,
+      "loss": 1.1071,
+      "step": 5353
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0003036243079530487,
+      "loss": 1.0866,
+      "step": 5354
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0003035884092560478,
+      "loss": 1.0632,
+      "step": 5355
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00030355250599753,
+      "loss": 1.0299,
+      "step": 5356
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00030351659817907635,
+      "loss": 1.0712,
+      "step": 5357
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.000303480685802268,
+      "loss": 1.0148,
+      "step": 5358
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00030344476886868643,
+      "loss": 1.0279,
+      "step": 5359
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00030340884737991306,
+      "loss": 0.9311,
+      "step": 5360
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00030337292133752986,
+      "loss": 1.0741,
+      "step": 5361
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00030333699074311886,
+      "loss": 1.0505,
+      "step": 5362
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0003033010555982621,
+      "loss": 1.0765,
+      "step": 5363
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0003032651159045421,
+      "loss": 1.094,
+      "step": 5364
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0003032291716635415,
+      "loss": 1.0425,
+      "step": 5365
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0003031932228768429,
+      "loss": 1.1456,
+      "step": 5366
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00030315726954602956,
+      "loss": 1.0779,
+      "step": 5367
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0003031213116726845,
+      "loss": 1.0598,
+      "step": 5368
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0003030853492583913,
+      "loss": 1.1474,
+      "step": 5369
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00030304938230473345,
+      "loss": 1.1001,
+      "step": 5370
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00030301341081329477,
+      "loss": 1.0506,
+      "step": 5371
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00030297743478565915,
+      "loss": 1.1697,
+      "step": 5372
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00030294145422341104,
+      "loss": 1.0776,
+      "step": 5373
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00030290546912813463,
+      "loss": 1.1138,
+      "step": 5374
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0003028694795014147,
+      "loss": 1.1548,
+      "step": 5375
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00030283348534483595,
+      "loss": 1.1353,
+      "step": 5376
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00030279748665998336,
+      "loss": 1.0991,
+      "step": 5377
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0003027614834484421,
+      "loss": 1.0144,
+      "step": 5378
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00030272547571179775,
+      "loss": 1.0242,
+      "step": 5379
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0003026894634516357,
+      "loss": 1.017,
+      "step": 5380
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.000302653446669542,
+      "loss": 1.1205,
+      "step": 5381
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0003026174253671024,
+      "loss": 1.0543,
+      "step": 5382
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0003025813995459032,
+      "loss": 1.0612,
+      "step": 5383
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0003025453692075308,
+      "loss": 1.0629,
+      "step": 5384
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00030250933435357186,
+      "loss": 1.084,
+      "step": 5385
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0003024732949856131,
+      "loss": 1.1095,
+      "step": 5386
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00030243725110524144,
+      "loss": 1.1112,
+      "step": 5387
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0003024012027140442,
+      "loss": 1.1026,
+      "step": 5388
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00030236514981360875,
+      "loss": 1.0895,
+      "step": 5389
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00030232909240552263,
+      "loss": 1.0614,
+      "step": 5390
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00030229303049137367,
+      "loss": 1.0372,
+      "step": 5391
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00030225696407274984,
+      "loss": 1.0775,
+      "step": 5392
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0003022208931512393,
+      "loss": 1.0484,
+      "step": 5393
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0003021848177284305,
+      "loss": 1.1066,
+      "step": 5394
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.000302148737805912,
+      "loss": 1.111,
+      "step": 5395
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00030211265338527253,
+      "loss": 1.1042,
+      "step": 5396
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0003020765644681011,
+      "loss": 1.0384,
+      "step": 5397
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00030204047105598686,
+      "loss": 1.0868,
+      "step": 5398
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0003020043731505193,
+      "loss": 1.0371,
+      "step": 5399
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0003019682707532878,
+      "loss": 1.0867,
+      "step": 5400
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0003019321638658823,
+      "loss": 1.0665,
+      "step": 5401
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0003018960524898927,
+      "loss": 1.1823,
+      "step": 5402
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0003018599366269091,
+      "loss": 1.072,
+      "step": 5403
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.000301823816278522,
+      "loss": 1.0447,
+      "step": 5404
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00030178769144632183,
+      "loss": 1.1233,
+      "step": 5405
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0003017515621318994,
+      "loss": 1.0166,
+      "step": 5406
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00030171542833684574,
+      "loss": 1.0247,
+      "step": 5407
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00030167929006275193,
+      "loss": 1.0953,
+      "step": 5408
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00030164314731120917,
+      "loss": 1.0175,
+      "step": 5409
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00030160700008380933,
+      "loss": 1.0278,
+      "step": 5410
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0003015708483821439,
+      "loss": 1.1075,
+      "step": 5411
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00030153469220780486,
+      "loss": 1.0965,
+      "step": 5412
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00030149853156238453,
+      "loss": 1.0435,
+      "step": 5413
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00030146236644747497,
+      "loss": 1.0919,
+      "step": 5414
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00030142619686466897,
+      "loss": 1.0932,
+      "step": 5415
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.000301390022815559,
+      "loss": 1.078,
+      "step": 5416
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00030135384430173814,
+      "loss": 1.0718,
+      "step": 5417
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0003013176613247995,
+      "loss": 1.0646,
+      "step": 5418
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00030128147388633645,
+      "loss": 1.0554,
+      "step": 5419
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0003012452819879424,
+      "loss": 1.1206,
+      "step": 5420
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0003012090856312111,
+      "loss": 1.0686,
+      "step": 5421
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0003011728848177365,
+      "loss": 1.1144,
+      "step": 5422
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00030113667954911254,
+      "loss": 0.9876,
+      "step": 5423
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00030110046982693376,
+      "loss": 1.1018,
+      "step": 5424
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0003010642556527945,
+      "loss": 1.0642,
+      "step": 5425
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0003010280370282895,
+      "loss": 1.1411,
+      "step": 5426
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0003009918139550136,
+      "loss": 1.011,
+      "step": 5427
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0003009555864345619,
+      "loss": 1.0738,
+      "step": 5428
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00030091935446852975,
+      "loss": 1.1129,
+      "step": 5429
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00030088311805851256,
+      "loss": 1.0607,
+      "step": 5430
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00030084687720610605,
+      "loss": 1.0502,
+      "step": 5431
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.000300810631912906,
+      "loss": 1.0987,
+      "step": 5432
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0003007743821805085,
+      "loss": 1.0696,
+      "step": 5433
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0003007381280105099,
+      "loss": 1.0111,
+      "step": 5434
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00030070186940450656,
+      "loss": 0.9993,
+      "step": 5435
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0003006656063640951,
+      "loss": 1.0945,
+      "step": 5436
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00030062933889087246,
+      "loss": 1.0447,
+      "step": 5437
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00030059306698643564,
+      "loss": 1.0047,
+      "step": 5438
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00030055679065238184,
+      "loss": 1.1142,
+      "step": 5439
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0003005205098903086,
+      "loss": 1.0926,
+      "step": 5440
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0003004842247018133,
+      "loss": 1.0895,
+      "step": 5441
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.000300447935088494,
+      "loss": 1.0652,
+      "step": 5442
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00030041164105194866,
+      "loss": 1.1083,
+      "step": 5443
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00030037534259377553,
+      "loss": 1.102,
+      "step": 5444
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0003003390397155728,
+      "loss": 0.9995,
+      "step": 5445
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00030030273241893926,
+      "loss": 1.0014,
+      "step": 5446
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00030026642070547365,
+      "loss": 1.0641,
+      "step": 5447
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.000300230104576775,
+      "loss": 1.1587,
+      "step": 5448
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0003001937840344424,
+      "loss": 1.0867,
+      "step": 5449
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0003001574590800753,
+      "loss": 1.1003,
+      "step": 5450
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0003001211297152732,
+      "loss": 1.0674,
+      "step": 5451
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.000300084795941636,
+      "loss": 1.0417,
+      "step": 5452
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0003000484577607635,
+      "loss": 1.1256,
+      "step": 5453
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.000300012115174256,
+      "loss": 0.9795,
+      "step": 5454
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00029997576818371365,
+      "loss": 1.0623,
+      "step": 5455
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00029993941679073717,
+      "loss": 1.1312,
+      "step": 5456
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002999030609969273,
+      "loss": 1.058,
+      "step": 5457
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002998667008038848,
+      "loss": 1.0874,
+      "step": 5458
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00029983033621321087,
+      "loss": 1.0061,
+      "step": 5459
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002997939672265069,
+      "loss": 1.0866,
+      "step": 5460
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00029975759384537435,
+      "loss": 1.0184,
+      "step": 5461
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002997212160714149,
+      "loss": 1.1094,
+      "step": 5462
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002996848339062305,
+      "loss": 1.074,
+      "step": 5463
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00029964844735142317,
+      "loss": 1.1338,
+      "step": 5464
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002996120564085952,
+      "loss": 1.0808,
+      "step": 5465
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002995756610793491,
+      "loss": 1.0139,
+      "step": 5466
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002995392613652876,
+      "loss": 0.9977,
+      "step": 5467
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002995028572680134,
+      "loss": 1.0302,
+      "step": 5468
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002994664487891297,
+      "loss": 1.0826,
+      "step": 5469
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002994300359302397,
+      "loss": 1.0682,
+      "step": 5470
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00029939361869294677,
+      "loss": 1.1214,
+      "step": 5471
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00029935719707885463,
+      "loss": 1.0817,
+      "step": 5472
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002993207710895671,
+      "loss": 1.0383,
+      "step": 5473
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00029928434072668816,
+      "loss": 1.0573,
+      "step": 5474
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002992479059918221,
+      "loss": 1.0994,
+      "step": 5475
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002992114668865732,
+      "loss": 1.0171,
+      "step": 5476
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002991750234125462,
+      "loss": 1.192,
+      "step": 5477
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00029913857557134575,
+      "loss": 1.1468,
+      "step": 5478
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00029910212336457695,
+      "loss": 1.0338,
+      "step": 5479
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002990656667938448,
+      "loss": 1.1111,
+      "step": 5480
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00029902920586075496,
+      "loss": 1.1302,
+      "step": 5481
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002989927405669127,
+      "loss": 1.0175,
+      "step": 5482
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002989562709139239,
+      "loss": 1.0798,
+      "step": 5483
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002989197969033945,
+      "loss": 1.0618,
+      "step": 5484
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00029888331853693056,
+      "loss": 1.0925,
+      "step": 5485
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002988468358161385,
+      "loss": 1.1083,
+      "step": 5486
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002988103487426248,
+      "loss": 1.0747,
+      "step": 5487
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00029877385731799617,
+      "loss": 1.0148,
+      "step": 5488
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00029873736154385945,
+      "loss": 1.0168,
+      "step": 5489
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00029870086142182184,
+      "loss": 1.1568,
+      "step": 5490
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002986643569534906,
+      "loss": 1.04,
+      "step": 5491
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00029862784814047315,
+      "loss": 0.9825,
+      "step": 5492
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00029859133498437717,
+      "loss": 1.1366,
+      "step": 5493
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00029855481748681053,
+      "loss": 1.0806,
+      "step": 5494
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00029851829564938134,
+      "loss": 1.1359,
+      "step": 5495
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002984817694736977,
+      "loss": 1.0327,
+      "step": 5496
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002984452389613681,
+      "loss": 1.0834,
+      "step": 5497
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00029840870411400127,
+      "loss": 0.9555,
+      "step": 5498
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002983721649332059,
+      "loss": 1.0328,
+      "step": 5499
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00029833562142059103,
+      "loss": 1.0796,
+      "step": 5500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002982990735777658,
+      "loss": 1.0796,
+      "step": 5501
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002982625214063397,
+      "loss": 0.9926,
+      "step": 5502
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002982259649079222,
+      "loss": 1.0452,
+      "step": 5503
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002981894040841231,
+      "loss": 1.0459,
+      "step": 5504
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002981528389365525,
+      "loss": 1.0592,
+      "step": 5505
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00029811626946682035,
+      "loss": 1.0599,
+      "step": 5506
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002980796956765371,
+      "loss": 1.1146,
+      "step": 5507
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00029804311756731315,
+      "loss": 1.0052,
+      "step": 5508
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002980065351407593,
+      "loss": 1.0514,
+      "step": 5509
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002979699483984865,
+      "loss": 0.9967,
+      "step": 5510
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002979333573421058,
+      "loss": 1.0576,
+      "step": 5511
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002978967619732285,
+      "loss": 1.0837,
+      "step": 5512
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002978601622934661,
+      "loss": 1.0471,
+      "step": 5513
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00029782355830443015,
+      "loss": 1.0662,
+      "step": 5514
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00029778695000773266,
+      "loss": 1.1534,
+      "step": 5515
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002977503374049856,
+      "loss": 0.9897,
+      "step": 5516
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002977137204978012,
+      "loss": 1.0738,
+      "step": 5517
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00029767709928779195,
+      "loss": 1.0296,
+      "step": 5518
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00029764047377657044,
+      "loss": 1.0583,
+      "step": 5519
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00029760384396574935,
+      "loss": 1.1149,
+      "step": 5520
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002975672098569419,
+      "loss": 1.0457,
+      "step": 5521
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002975305714517611,
+      "loss": 0.9727,
+      "step": 5522
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002974939287518203,
+      "loss": 1.1031,
+      "step": 5523
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00029745728175873326,
+      "loss": 1.0332,
+      "step": 5524
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00029742063047411357,
+      "loss": 1.0864,
+      "step": 5525
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00029738397489957515,
+      "loss": 1.0519,
+      "step": 5526
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00029734731503673227,
+      "loss": 1.042,
+      "step": 5527
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002973106508871991,
+      "loss": 1.0631,
+      "step": 5528
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002972739824525902,
+      "loss": 1.1675,
+      "step": 5529
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002972373097345203,
+      "loss": 1.0978,
+      "step": 5530
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002972006327346043,
+      "loss": 1.0944,
+      "step": 5531
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002971639514544572,
+      "loss": 1.0947,
+      "step": 5532
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00029712726589569435,
+      "loss": 1.1323,
+      "step": 5533
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00029709057605993106,
+      "loss": 1.0638,
+      "step": 5534
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002970538819487831,
+      "loss": 1.0575,
+      "step": 5535
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002970171835638662,
+      "loss": 1.0652,
+      "step": 5536
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00029698048090679645,
+      "loss": 1.0438,
+      "step": 5537
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00029694377397919,
+      "loss": 1.0772,
+      "step": 5538
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002969070627826633,
+      "loss": 1.076,
+      "step": 5539
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002968703473188328,
+      "loss": 1.0849,
+      "step": 5540
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00029683362758931546,
+      "loss": 1.0633,
+      "step": 5541
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.000296796903595728,
+      "loss": 1.0966,
+      "step": 5542
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002967601753396878,
+      "loss": 1.1068,
+      "step": 5543
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00029672344282281204,
+      "loss": 1.1324,
+      "step": 5544
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00029668670604671827,
+      "loss": 1.0761,
+      "step": 5545
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00029664996501302425,
+      "loss": 1.1117,
+      "step": 5546
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002966132197233478,
+      "loss": 1.0343,
+      "step": 5547
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002965764701793069,
+      "loss": 1.0751,
+      "step": 5548
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002965397163825201,
+      "loss": 1.0933,
+      "step": 5549
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00029650295833460566,
+      "loss": 1.1028,
+      "step": 5550
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00029646619603718215,
+      "loss": 1.0832,
+      "step": 5551
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002964294294918686,
+      "loss": 1.0488,
+      "step": 5552
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00029639265870028386,
+      "loss": 1.0391,
+      "step": 5553
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002963558836640472,
+      "loss": 1.0888,
+      "step": 5554
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002963191043847781,
+      "loss": 1.0252,
+      "step": 5555
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.000296282320864096,
+      "loss": 1.0929,
+      "step": 5556
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00029624553310362065,
+      "loss": 1.0362,
+      "step": 5557
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002962087411049721,
+      "loss": 1.0376,
+      "step": 5558
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00029617194486977045,
+      "loss": 1.0757,
+      "step": 5559
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00029613514439963596,
+      "loss": 1.057,
+      "step": 5560
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00029609833969618926,
+      "loss": 1.0028,
+      "step": 5561
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00029606153076105096,
+      "loss": 1.0935,
+      "step": 5562
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.000296024717595842,
+      "loss": 1.1054,
+      "step": 5563
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002959879002021833,
+      "loss": 1.0851,
+      "step": 5564
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00029595107858169624,
+      "loss": 0.9981,
+      "step": 5565
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002959142527360023,
+      "loss": 1.0911,
+      "step": 5566
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00029587742266672307,
+      "loss": 1.0083,
+      "step": 5567
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002958405883754803,
+      "loss": 0.9628,
+      "step": 5568
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002958037498638961,
+      "loss": 1.0478,
+      "step": 5569
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002957669071335925,
+      "loss": 1.0979,
+      "step": 5570
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.000295730060186192,
+      "loss": 1.0398,
+      "step": 5571
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002956932090233171,
+      "loss": 1.1326,
+      "step": 5572
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002956563536465905,
+      "loss": 1.0523,
+      "step": 5573
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00029561949405763525,
+      "loss": 1.0407,
+      "step": 5574
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00029558263025807437,
+      "loss": 1.1433,
+      "step": 5575
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002955457622495312,
+      "loss": 1.0745,
+      "step": 5576
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002955088900336292,
+      "loss": 0.9916,
+      "step": 5577
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.000295472013611992,
+      "loss": 1.1112,
+      "step": 5578
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002954351329862435,
+      "loss": 1.1364,
+      "step": 5579
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002953982481580078,
+      "loss": 1.0304,
+      "step": 5580
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00029536135912890907,
+      "loss": 1.1158,
+      "step": 5581
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00029532446590057156,
+      "loss": 1.0608,
+      "step": 5582
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00029528756847462017,
+      "loss": 1.1068,
+      "step": 5583
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002952506668526794,
+      "loss": 1.0858,
+      "step": 5584
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002952137610363744,
+      "loss": 1.0435,
+      "step": 5585
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002951768510273302,
+      "loss": 1.0621,
+      "step": 5586
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002951399368271723,
+      "loss": 1.0873,
+      "step": 5587
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00029510301843752604,
+      "loss": 1.0737,
+      "step": 5588
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00029506609586001716,
+      "loss": 1.1061,
+      "step": 5589
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002950291690962716,
+      "loss": 1.0182,
+      "step": 5590
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00029499223814791534,
+      "loss": 1.0501,
+      "step": 5591
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00029495530301657473,
+      "loss": 1.0621,
+      "step": 5592
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002949183637038762,
+      "loss": 1.156,
+      "step": 5593
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002948814202114463,
+      "loss": 1.0524,
+      "step": 5594
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002948444725409119,
+      "loss": 1.1195,
+      "step": 5595
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00029480752069389996,
+      "loss": 1.11,
+      "step": 5596
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002947705646720377,
+      "loss": 1.0947,
+      "step": 5597
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002947336044769524,
+      "loss": 1.0956,
+      "step": 5598
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002946966401102717,
+      "loss": 1.0681,
+      "step": 5599
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00029465967157362333,
+      "loss": 1.0701,
+      "step": 5600
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.000294622698868635,
+      "loss": 1.1286,
+      "step": 5601
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002945857219969351,
+      "loss": 1.0506,
+      "step": 5602
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002945487409601517,
+      "loss": 1.0455,
+      "step": 5603
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00029451175575991327,
+      "loss": 1.0453,
+      "step": 5604
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002944747663978486,
+      "loss": 1.0982,
+      "step": 5605
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002944377728755863,
+      "loss": 1.0796,
+      "step": 5606
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002944007751947556,
+      "loss": 1.0748,
+      "step": 5607
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00029436377335698556,
+      "loss": 1.0152,
+      "step": 5608
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002943267673639055,
+      "loss": 1.0877,
+      "step": 5609
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002942897572171452,
+      "loss": 1.0792,
+      "step": 5610
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002942527429183342,
+      "loss": 0.9693,
+      "step": 5611
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00029421572446910256,
+      "loss": 1.1378,
+      "step": 5612
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002941787018710803,
+      "loss": 1.1178,
+      "step": 5613
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002941416751258977,
+      "loss": 1.1021,
+      "step": 5614
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002941046442351853,
+      "loss": 1.0582,
+      "step": 5615
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00029406760920057374,
+      "loss": 1.0434,
+      "step": 5616
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00029403057002369375,
+      "loss": 1.04,
+      "step": 5617
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002939935267061765,
+      "loss": 1.0945,
+      "step": 5618
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002939564792496532,
+      "loss": 1.106,
+      "step": 5619
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00029391942765575506,
+      "loss": 1.0731,
+      "step": 5620
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002938823719261138,
+      "loss": 1.0508,
+      "step": 5621
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002938453120623612,
+      "loss": 1.1493,
+      "step": 5622
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00029380824806612895,
+      "loss": 1.0748,
+      "step": 5623
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00029377117993904946,
+      "loss": 1.0807,
+      "step": 5624
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002937341076827549,
+      "loss": 1.0341,
+      "step": 5625
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002936970312988777,
+      "loss": 1.0593,
+      "step": 5626
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002936599507890506,
+      "loss": 1.0138,
+      "step": 5627
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00029362286615490637,
+      "loss": 1.042,
+      "step": 5628
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00029358577739807806,
+      "loss": 1.0002,
+      "step": 5629
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00029354868452019896,
+      "loss": 1.0934,
+      "step": 5630
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002935115875229023,
+      "loss": 1.0466,
+      "step": 5631
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00029347448640782174,
+      "loss": 1.1118,
+      "step": 5632
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.000293437381176591,
+      "loss": 1.0639,
+      "step": 5633
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.000293400271830844,
+      "loss": 1.1075,
+      "step": 5634
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002933631583722149,
+      "loss": 1.0937,
+      "step": 5635
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00029332604080233797,
+      "loss": 1.0706,
+      "step": 5636
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002932889191228477,
+      "loss": 1.1093,
+      "step": 5637
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002932517933353787,
+      "loss": 1.0952,
+      "step": 5638
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002932146634415658,
+      "loss": 1.0178,
+      "step": 5639
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002931775294430441,
+      "loss": 1.0612,
+      "step": 5640
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002931403913414486,
+      "loss": 1.016,
+      "step": 5641
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00029310324913841494,
+      "loss": 1.1058,
+      "step": 5642
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002930661028355785,
+      "loss": 1.0192,
+      "step": 5643
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00029302895243457514,
+      "loss": 1.1141,
+      "step": 5644
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002929917979370406,
+      "loss": 1.1297,
+      "step": 5645
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00029295463934461105,
+      "loss": 1.0567,
+      "step": 5646
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002929174766589229,
+      "loss": 0.9979,
+      "step": 5647
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002928803098816125,
+      "loss": 1.0149,
+      "step": 5648
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00029284313901431654,
+      "loss": 1.0653,
+      "step": 5649
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00029280596405867175,
+      "loss": 1.0415,
+      "step": 5650
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00029276878501631514,
+      "loss": 1.0048,
+      "step": 5651
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00029273160188888396,
+      "loss": 1.0786,
+      "step": 5652
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002926944146780156,
+      "loss": 1.0084,
+      "step": 5653
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002926572233853475,
+      "loss": 1.0224,
+      "step": 5654
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002926200280125174,
+      "loss": 1.0335,
+      "step": 5655
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00029258282856116325,
+      "loss": 1.0613,
+      "step": 5656
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.000292545625032923,
+      "loss": 1.0829,
+      "step": 5657
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002925084174294351,
+      "loss": 1.0638,
+      "step": 5658
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00029247120575233784,
+      "loss": 1.0816,
+      "step": 5659
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002924339900032699,
+      "loss": 1.1096,
+      "step": 5660
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00029239677018387013,
+      "loss": 1.0362,
+      "step": 5661
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002923595462957774,
+      "loss": 1.1048,
+      "step": 5662
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00029232231834063084,
+      "loss": 1.1344,
+      "step": 5663
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002922850863200699,
+      "loss": 1.0574,
+      "step": 5664
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.000292247850235734,
+      "loss": 1.0395,
+      "step": 5665
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002922106100892629,
+      "loss": 1.1308,
+      "step": 5666
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002921733658822965,
+      "loss": 1.0926,
+      "step": 5667
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002921361176164747,
+      "loss": 1.0254,
+      "step": 5668
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002920988652934379,
+      "loss": 1.0676,
+      "step": 5669
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00029206160891482637,
+      "loss": 1.1448,
+      "step": 5670
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00029202434848228073,
+      "loss": 1.0509,
+      "step": 5671
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00029198708399744177,
+      "loss": 1.0213,
+      "step": 5672
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002919498154619504,
+      "loss": 1.1483,
+      "step": 5673
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002919125428774478,
+      "loss": 1.0878,
+      "step": 5674
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002918752662455753,
+      "loss": 1.042,
+      "step": 5675
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00029183798556797427,
+      "loss": 1.071,
+      "step": 5676
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002918007008462863,
+      "loss": 1.0331,
+      "step": 5677
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00029176341208215345,
+      "loss": 1.0973,
+      "step": 5678
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00029172611927721757,
+      "loss": 1.0538,
+      "step": 5679
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002916888224331209,
+      "loss": 1.0709,
+      "step": 5680
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00029165152155150584,
+      "loss": 1.0411,
+      "step": 5681
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002916142166340148,
+      "loss": 1.0432,
+      "step": 5682
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00029157690768229065,
+      "loss": 1.1345,
+      "step": 5683
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002915395946979762,
+      "loss": 1.066,
+      "step": 5684
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00029150227768271457,
+      "loss": 1.0166,
+      "step": 5685
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.000291464956638149,
+      "loss": 1.1032,
+      "step": 5686
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.000291427631565923,
+      "loss": 1.1132,
+      "step": 5687
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00029139030246768,
+      "loss": 1.1012,
+      "step": 5688
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.000291352969345064,
+      "loss": 1.1033,
+      "step": 5689
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002913156321997188,
+      "loss": 1.1159,
+      "step": 5690
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00029127829103328857,
+      "loss": 1.0498,
+      "step": 5691
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002912409458474177,
+      "loss": 1.0221,
+      "step": 5692
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00029120359664375076,
+      "loss": 1.0947,
+      "step": 5693
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002911662434239322,
+      "loss": 1.0165,
+      "step": 5694
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002911288861896069,
+      "loss": 1.1116,
+      "step": 5695
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002910915249424201,
+      "loss": 1.1578,
+      "step": 5696
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002910541596840168,
+      "loss": 1.0865,
+      "step": 5697
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00029101679041604247,
+      "loss": 1.0275,
+      "step": 5698
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00029097941714014265,
+      "loss": 1.161,
+      "step": 5699
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00029094203985796305,
+      "loss": 1.0874,
+      "step": 5700
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00029090465857114963,
+      "loss": 1.1175,
+      "step": 5701
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002908672732813484,
+      "loss": 1.072,
+      "step": 5702
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002908298839902057,
+      "loss": 1.1326,
+      "step": 5703
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002907924906993679,
+      "loss": 1.1157,
+      "step": 5704
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002907550934104816,
+      "loss": 1.1182,
+      "step": 5705
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00029071769212519376,
+      "loss": 1.0197,
+      "step": 5706
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002906802868451511,
+      "loss": 1.0288,
+      "step": 5707
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.000290642877572001,
+      "loss": 1.0176,
+      "step": 5708
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00029060546430739054,
+      "loss": 1.0033,
+      "step": 5709
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00029056804705296737,
+      "loss": 1.0509,
+      "step": 5710
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00029053062581037916,
+      "loss": 1.0351,
+      "step": 5711
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00029049320058127377,
+      "loss": 1.064,
+      "step": 5712
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002904557713672991,
+      "loss": 1.0567,
+      "step": 5713
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00029041833817010343,
+      "loss": 1.0882,
+      "step": 5714
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00029038090099133506,
+      "loss": 1.0494,
+      "step": 5715
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002903434598326426,
+      "loss": 1.0552,
+      "step": 5716
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00029030601469567485,
+      "loss": 1.0931,
+      "step": 5717
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00029026856558208057,
+      "loss": 1.0358,
+      "step": 5718
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002902311124935089,
+      "loss": 1.0667,
+      "step": 5719
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002901936554316091,
+      "loss": 1.0615,
+      "step": 5720
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00029015619439803046,
+      "loss": 1.0821,
+      "step": 5721
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002901187293944228,
+      "loss": 1.0449,
+      "step": 5722
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002900812604224357,
+      "loss": 0.9653,
+      "step": 5723
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002900437874837192,
+      "loss": 1.1087,
+      "step": 5724
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002900063105799234,
+      "loss": 1.0238,
+      "step": 5725
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002899688297126986,
+      "loss": 1.0266,
+      "step": 5726
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002899313448836952,
+      "loss": 1.0274,
+      "step": 5727
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00028989385609456405,
+      "loss": 1.0982,
+      "step": 5728
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002898563633469557,
+      "loss": 1.0723,
+      "step": 5729
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00028981886664252143,
+      "loss": 1.0696,
+      "step": 5730
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00028978136598291213,
+      "loss": 0.9629,
+      "step": 5731
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002897438613697793,
+      "loss": 1.1207,
+      "step": 5732
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00028970635280477436,
+      "loss": 1.0309,
+      "step": 5733
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00028966884028954913,
+      "loss": 1.0173,
+      "step": 5734
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002896313238257554,
+      "loss": 1.0779,
+      "step": 5735
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002895938034150452,
+      "loss": 1.1022,
+      "step": 5736
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002895562790590708,
+      "loss": 1.0538,
+      "step": 5737
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002895187507594844,
+      "loss": 1.0803,
+      "step": 5738
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002894812185179389,
+      "loss": 1.1157,
+      "step": 5739
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002894436823360866,
+      "loss": 1.0647,
+      "step": 5740
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002894061422155808,
+      "loss": 1.0794,
+      "step": 5741
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002893685981580744,
+      "loss": 0.9592,
+      "step": 5742
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002893310501652207,
+      "loss": 1.092,
+      "step": 5743
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.000289293498238673,
+      "loss": 1.0595,
+      "step": 5744
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002892559423800851,
+      "loss": 1.0402,
+      "step": 5745
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002892183825911106,
+      "loss": 1.1882,
+      "step": 5746
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00028918081887340354,
+      "loss": 1.0609,
+      "step": 5747
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.000289143251228618,
+      "loss": 1.1453,
+      "step": 5748
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00028910567965840833,
+      "loss": 1.0872,
+      "step": 5749
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002890681041644289,
+      "loss": 1.0663,
+      "step": 5750
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002890305247483344,
+      "loss": 1.1009,
+      "step": 5751
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002889929414117796,
+      "loss": 1.1009,
+      "step": 5752
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002889553541564196,
+      "loss": 1.028,
+      "step": 5753
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002889177629839094,
+      "loss": 1.0932,
+      "step": 5754
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002888801678959045,
+      "loss": 1.0192,
+      "step": 5755
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002888425688940603,
+      "loss": 1.0595,
+      "step": 5756
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002888049659800324,
+      "loss": 1.1158,
+      "step": 5757
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00028876735915547673,
+      "loss": 1.0201,
+      "step": 5758
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002887297484220493,
+      "loss": 1.0977,
+      "step": 5759
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002886921337814063,
+      "loss": 1.1314,
+      "step": 5760
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00028865451523520417,
+      "loss": 1.0975,
+      "step": 5761
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002886168927850993,
+      "loss": 1.0374,
+      "step": 5762
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00028857926643274844,
+      "loss": 1.1353,
+      "step": 5763
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002885416361798085,
+      "loss": 1.0904,
+      "step": 5764
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002885040020279365,
+      "loss": 1.0169,
+      "step": 5765
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002884663639787896,
+      "loss": 1.0199,
+      "step": 5766
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002884287220340254,
+      "loss": 1.0724,
+      "step": 5767
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002883910761953013,
+      "loss": 1.1443,
+      "step": 5768
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.000288353426464275,
+      "loss": 1.0961,
+      "step": 5769
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00028831577284260456,
+      "loss": 1.0766,
+      "step": 5770
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00028827811533194787,
+      "loss": 1.1321,
+      "step": 5771
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00028824045393396327,
+      "loss": 1.1199,
+      "step": 5772
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00028820278865030927,
+      "loss": 1.0205,
+      "step": 5773
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002881651194826444,
+      "loss": 1.1339,
+      "step": 5774
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00028812744643262736,
+      "loss": 1.0528,
+      "step": 5775
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00028808976950191705,
+      "loss": 1.1332,
+      "step": 5776
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002880520886921727,
+      "loss": 1.1127,
+      "step": 5777
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00028801440400505354,
+      "loss": 1.0368,
+      "step": 5778
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.000287976715442219,
+      "loss": 1.0465,
+      "step": 5779
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00028793902300532873,
+      "loss": 1.1059,
+      "step": 5780
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00028790132669604247,
+      "loss": 1.1058,
+      "step": 5781
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002878636265160202,
+      "loss": 1.1321,
+      "step": 5782
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.000287825922466922,
+      "loss": 1.1377,
+      "step": 5783
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00028778821455040823,
+      "loss": 1.0641,
+      "step": 5784
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002877505027681394,
+      "loss": 1.1218,
+      "step": 5785
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.000287712787121776,
+      "loss": 1.0345,
+      "step": 5786
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00028767506761297906,
+      "loss": 1.1437,
+      "step": 5787
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002876373442434093,
+      "loss": 1.0929,
+      "step": 5788
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002875996170147281,
+      "loss": 0.9977,
+      "step": 5789
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002875618859285966,
+      "loss": 1.0094,
+      "step": 5790
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00028752415098667633,
+      "loss": 1.0077,
+      "step": 5791
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002874864121906291,
+      "loss": 1.1126,
+      "step": 5792
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002874486695421166,
+      "loss": 1.0693,
+      "step": 5793
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002874109230428008,
+      "loss": 1.0466,
+      "step": 5794
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.000287373172694344,
+      "loss": 1.0578,
+      "step": 5795
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00028733541849840834,
+      "loss": 1.0267,
+      "step": 5796
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00028729766045665654,
+      "loss": 1.0497,
+      "step": 5797
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002872598985707511,
+      "loss": 1.1386,
+      "step": 5798
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.000287222132842355,
+      "loss": 1.1196,
+      "step": 5799
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002871843632731312,
+      "loss": 1.0477,
+      "step": 5800
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00028714658986474294,
+      "loss": 1.0145,
+      "step": 5801
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002871088126188534,
+      "loss": 1.0441,
+      "step": 5802
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002870710315371262,
+      "loss": 1.0819,
+      "step": 5803
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00028703324662122514,
+      "loss": 1.0117,
+      "step": 5804
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002869954578728139,
+      "loss": 1.0383,
+      "step": 5805
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00028695766529355675,
+      "loss": 1.0695,
+      "step": 5806
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00028691986888511756,
+      "loss": 1.0435,
+      "step": 5807
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002868820686491609,
+      "loss": 1.0415,
+      "step": 5808
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00028684426458735134,
+      "loss": 1.1094,
+      "step": 5809
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00028680645670135346,
+      "loss": 1.0826,
+      "step": 5810
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002867686449928321,
+      "loss": 1.1225,
+      "step": 5811
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002867308294634526,
+      "loss": 1.0513,
+      "step": 5812
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002866930101148797,
+      "loss": 1.0667,
+      "step": 5813
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00028665518694877924,
+      "loss": 1.0455,
+      "step": 5814
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002866173599668164,
+      "loss": 1.1155,
+      "step": 5815
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.000286579529170657,
+      "loss": 1.0009,
+      "step": 5816
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00028654169456196715,
+      "loss": 1.1035,
+      "step": 5817
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00028650385614241257,
+      "loss": 1.0523,
+      "step": 5818
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00028646601391365964,
+      "loss": 1.0244,
+      "step": 5819
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00028642816787737474,
+      "loss": 1.0993,
+      "step": 5820
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002863903180352244,
+      "loss": 1.2024,
+      "step": 5821
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00028635246438887525,
+      "loss": 1.0379,
+      "step": 5822
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002863146069399943,
+      "loss": 1.0607,
+      "step": 5823
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00028627674569024856,
+      "loss": 0.9806,
+      "step": 5824
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002862388806413053,
+      "loss": 1.0784,
+      "step": 5825
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00028620101179483185,
+      "loss": 1.0215,
+      "step": 5826
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00028616313915249576,
+      "loss": 1.132,
+      "step": 5827
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002861252627159648,
+      "loss": 1.0718,
+      "step": 5828
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002860873824869067,
+      "loss": 1.0242,
+      "step": 5829
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002860494984669897,
+      "loss": 1.0516,
+      "step": 5830
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002860116106578821,
+      "loss": 1.0713,
+      "step": 5831
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002859737190612521,
+      "loss": 0.9699,
+      "step": 5832
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00028593582367876817,
+      "loss": 1.0049,
+      "step": 5833
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002858979245120994,
+      "loss": 1.0049,
+      "step": 5834
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00028586002156291425,
+      "loss": 1.0033,
+      "step": 5835
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00028582211483288206,
+      "loss": 1.1019,
+      "step": 5836
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.000285784204323672,
+      "loss": 1.0589,
+      "step": 5837
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00028574629003695346,
+      "loss": 1.1123,
+      "step": 5838
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002857083719743959,
+      "loss": 1.0993,
+      "step": 5839
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002856704501376692,
+      "loss": 1.1529,
+      "step": 5840
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002856325245284431,
+      "loss": 1.0057,
+      "step": 5841
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002855945951483878,
+      "loss": 1.0443,
+      "step": 5842
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002855566619991733,
+      "loss": 1.0106,
+      "step": 5843
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00028551872508247026,
+      "loss": 0.9921,
+      "step": 5844
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002854807843999491,
+      "loss": 1.114,
+      "step": 5845
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002854428399532805,
+      "loss": 1.0987,
+      "step": 5846
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002854048917441353,
+      "loss": 1.071,
+      "step": 5847
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00028536693977418475,
+      "loss": 1.0666,
+      "step": 5848
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002853289840450999,
+      "loss": 1.0256,
+      "step": 5849
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00028529102455855215,
+      "loss": 1.1176,
+      "step": 5850
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002852530613162132,
+      "loss": 1.0222,
+      "step": 5851
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00028521509431975444,
+      "loss": 1.1013,
+      "step": 5852
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00028517712357084806,
+      "loss": 1.1246,
+      "step": 5853
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.000285139149071166,
+      "loss": 1.0847,
+      "step": 5854
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002851011708223804,
+      "loss": 1.0468,
+      "step": 5855
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002850631888261637,
+      "loss": 1.0976,
+      "step": 5856
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002850252030841885,
+      "loss": 1.0347,
+      "step": 5857
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00028498721359812736,
+      "loss": 1.1134,
+      "step": 5858
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00028494922036965324,
+      "loss": 1.07,
+      "step": 5859
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002849112234004391,
+      "loss": 1.0276,
+      "step": 5860
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00028487322269215817,
+      "loss": 1.1021,
+      "step": 5861
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002848352182464839,
+      "loss": 0.9942,
+      "step": 5862
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002847972100650897,
+      "loss": 1.0052,
+      "step": 5863
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002847591981496493,
+      "loss": 1.0212,
+      "step": 5864
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002847211825018366,
+      "loss": 1.0609,
+      "step": 5865
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002846831631233255,
+      "loss": 1.1162,
+      "step": 5866
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002846451400157903,
+      "loss": 1.0256,
+      "step": 5867
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00028460711318090533,
+      "loss": 1.0273,
+      "step": 5868
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002845690826203451,
+      "loss": 1.0521,
+      "step": 5869
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002845310483357843,
+      "loss": 1.0811,
+      "step": 5870
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00028449301032889765,
+      "loss": 0.979,
+      "step": 5871
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002844549686013602,
+      "loss": 1.0252,
+      "step": 5872
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002844169231548472,
+      "loss": 1.0168,
+      "step": 5873
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.000284378873991034,
+      "loss": 1.0142,
+      "step": 5874
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.000284340821111596,
+      "loss": 0.989,
+      "step": 5875
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00028430276451820894,
+      "loss": 1.0482,
+      "step": 5876
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002842647042125485,
+      "loss": 1.0163,
+      "step": 5877
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00028422664019629084,
+      "loss": 1.0238,
+      "step": 5878
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.000284188572471112,
+      "loss": 1.0784,
+      "step": 5879
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00028415050103868834,
+      "loss": 1.043,
+      "step": 5880
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002841124259006963,
+      "loss": 1.1099,
+      "step": 5881
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002840743470588126,
+      "loss": 1.0576,
+      "step": 5882
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00028403626451471394,
+      "loss": 1.0425,
+      "step": 5883
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00028399817827007734,
+      "loss": 1.1111,
+      "step": 5884
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002839600883265799,
+      "loss": 1.0718,
+      "step": 5885
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00028392199468589894,
+      "loss": 1.0597,
+      "step": 5886
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00028388389734971197,
+      "loss": 1.1291,
+      "step": 5887
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00028384579631969645,
+      "loss": 1.084,
+      "step": 5888
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00028380769159753024,
+      "loss": 1.0812,
+      "step": 5889
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00028376958318489134,
+      "loss": 1.0494,
+      "step": 5890
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002837314710834578,
+      "loss": 1.1043,
+      "step": 5891
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002836933552949078,
+      "loss": 1.1149,
+      "step": 5892
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00028365523582092,
+      "loss": 1.0871,
+      "step": 5893
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002836171126631728,
+      "loss": 1.0828,
+      "step": 5894
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00028357898582334504,
+      "loss": 1.0083,
+      "step": 5895
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00028354085530311554,
+      "loss": 1.0398,
+      "step": 5896
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00028350272110416345,
+      "loss": 1.0483,
+      "step": 5897
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.000283464583228168,
+      "loss": 1.0,
+      "step": 5898
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002834264416768086,
+      "loss": 1.1156,
+      "step": 5899
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00028338829645176486,
+      "loss": 1.0578,
+      "step": 5900
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002833501475547164,
+      "loss": 1.0891,
+      "step": 5901
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002833119949873432,
+      "loss": 1.1875,
+      "step": 5902
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00028327383875132513,
+      "loss": 1.0387,
+      "step": 5903
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00028323567884834267,
+      "loss": 1.148,
+      "step": 5904
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.000283197515280076,
+      "loss": 1.0514,
+      "step": 5905
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00028315934804820577,
+      "loss": 1.0713,
+      "step": 5906
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00028312117715441256,
+      "loss": 1.0421,
+      "step": 5907
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002830830026003773,
+      "loss": 1.0138,
+      "step": 5908
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.000283044824387781,
+      "loss": 1.1192,
+      "step": 5909
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002830066425183048,
+      "loss": 1.1283,
+      "step": 5910
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002829684569936301,
+      "loss": 1.0197,
+      "step": 5911
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00028293026781543837,
+      "loss": 1.1148,
+      "step": 5912
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00028289207498541125,
+      "loss": 1.0843,
+      "step": 5913
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00028285387850523054,
+      "loss": 1.1388,
+      "step": 5914
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00028281567837657834,
+      "loss": 1.1226,
+      "step": 5915
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00028277747460113666,
+      "loss": 1.0378,
+      "step": 5916
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002827392671805878,
+      "loss": 1.0219,
+      "step": 5917
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00028270105611661445,
+      "loss": 1.0785,
+      "step": 5918
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.000282662841410899,
+      "loss": 1.0754,
+      "step": 5919
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00028262462306512427,
+      "loss": 1.1969,
+      "step": 5920
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002825864010809732,
+      "loss": 1.0946,
+      "step": 5921
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.000282548175460129,
+      "loss": 1.0598,
+      "step": 5922
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00028250994620427483,
+      "loss": 1.0645,
+      "step": 5923
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002824717133150942,
+      "loss": 0.9853,
+      "step": 5924
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00028243347679427063,
+      "loss": 1.054,
+      "step": 5925
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00028239523664348786,
+      "loss": 1.0862,
+      "step": 5926
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002823569928644298,
+      "loss": 1.0863,
+      "step": 5927
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00028231874545878063,
+      "loss": 1.1249,
+      "step": 5928
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002822804944282244,
+      "loss": 1.1404,
+      "step": 5929
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002822422397744456,
+      "loss": 1.0819,
+      "step": 5930
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002822039814991288,
+      "loss": 1.0289,
+      "step": 5931
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00028216571960395864,
+      "loss": 1.1205,
+      "step": 5932
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00028212745409062,
+      "loss": 1.0967,
+      "step": 5933
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002820891849607978,
+      "loss": 1.1118,
+      "step": 5934
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00028205091221617743,
+      "loss": 1.1283,
+      "step": 5935
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002820126358584441,
+      "loss": 1.0941,
+      "step": 5936
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002819743558892833,
+      "loss": 0.993,
+      "step": 5937
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00028193607231038077,
+      "loss": 1.0696,
+      "step": 5938
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002818977851234222,
+      "loss": 1.0748,
+      "step": 5939
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002818594943300937,
+      "loss": 1.0911,
+      "step": 5940
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00028182119993208133,
+      "loss": 0.9665,
+      "step": 5941
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002817829019310714,
+      "loss": 0.9848,
+      "step": 5942
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002817446003287504,
+      "loss": 1.0572,
+      "step": 5943
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00028170629512680494,
+      "loss": 1.0514,
+      "step": 5944
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002816679863269217,
+      "loss": 1.0823,
+      "step": 5945
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00028162967393078773,
+      "loss": 1.1315,
+      "step": 5946
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00028159135794009,
+      "loss": 1.0685,
+      "step": 5947
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002815530383565158,
+      "loss": 1.0151,
+      "step": 5948
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00028151471518175254,
+      "loss": 1.1193,
+      "step": 5949
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00028147638841748786,
+      "loss": 1.0495,
+      "step": 5950
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00028143805806540937,
+      "loss": 1.1109,
+      "step": 5951
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00028139972412720497,
+      "loss": 1.0326,
+      "step": 5952
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00028136138660456265,
+      "loss": 1.179,
+      "step": 5953
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002813230454991707,
+      "loss": 1.0761,
+      "step": 5954
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002812847008127175,
+      "loss": 1.0502,
+      "step": 5955
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002812463525468915,
+      "loss": 1.0375,
+      "step": 5956
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002812080007033813,
+      "loss": 1.0849,
+      "step": 5957
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002811696452838757,
+      "loss": 1.0887,
+      "step": 5958
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00028113128629006387,
+      "loss": 1.0773,
+      "step": 5959
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00028109292372363484,
+      "loss": 1.0089,
+      "step": 5960
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00028105455758627786,
+      "loss": 0.9874,
+      "step": 5961
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002810161878796825,
+      "loss": 1.1083,
+      "step": 5962
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00028097781460553826,
+      "loss": 1.0738,
+      "step": 5963
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002809394377655349,
+      "loss": 1.0314,
+      "step": 5964
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00028090105736136247,
+      "loss": 1.1397,
+      "step": 5965
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002808626733947109,
+      "loss": 1.0611,
+      "step": 5966
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00028082428586727055,
+      "loss": 1.1111,
+      "step": 5967
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00028078589478073174,
+      "loss": 1.0794,
+      "step": 5968
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00028074750013678515,
+      "loss": 1.0656,
+      "step": 5969
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002807091019371213,
+      "loss": 1.0513,
+      "step": 5970
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00028067070018343115,
+      "loss": 1.0713,
+      "step": 5971
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00028063229487740576,
+      "loss": 1.0958,
+      "step": 5972
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00028059388602073615,
+      "loss": 1.0618,
+      "step": 5973
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00028055547361511386,
+      "loss": 1.0818,
+      "step": 5974
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00028051705766223034,
+      "loss": 1.0731,
+      "step": 5975
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002804786381637771,
+      "loss": 1.0505,
+      "step": 5976
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00028044021512144607,
+      "loss": 1.0288,
+      "step": 5977
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002804017885369291,
+      "loss": 1.1247,
+      "step": 5978
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00028036335841191847,
+      "loss": 1.0918,
+      "step": 5979
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002803249247481063,
+      "loss": 1.0444,
+      "step": 5980
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002802864875471851,
+      "loss": 1.1122,
+      "step": 5981
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00028024804681084744,
+      "loss": 1.1325,
+      "step": 5982
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00028020960254078594,
+      "loss": 1.031,
+      "step": 5983
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002801711547386937,
+      "loss": 1.0527,
+      "step": 5984
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002801327034062637,
+      "loss": 0.9556,
+      "step": 5985
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.000280094248545189,
+      "loss": 1.023,
+      "step": 5986
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00028005579015716314,
+      "loss": 0.9852,
+      "step": 5987
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00028001732824387953,
+      "loss": 1.0626,
+      "step": 5988
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002799788628070319,
+      "loss": 1.0659,
+      "step": 5989
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00027994039384831405,
+      "loss": 1.0856,
+      "step": 5990
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00027990192136941993,
+      "loss": 1.1466,
+      "step": 5991
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002798634453720437,
+      "loss": 1.1005,
+      "step": 5992
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00027982496585787976,
+      "loss": 1.0823,
+      "step": 5993
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00027978648282862243,
+      "loss": 1.0778,
+      "step": 5994
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00027974799628596627,
+      "loss": 1.0844,
+      "step": 5995
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00027970950623160617,
+      "loss": 1.0742,
+      "step": 5996
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00027967101266723694,
+      "loss": 1.0037,
+      "step": 5997
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00027963251559455363,
+      "loss": 1.0857,
+      "step": 5998
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002795940150152516,
+      "loss": 0.9781,
+      "step": 5999
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00027955551093102613,
+      "loss": 1.0241,
+      "step": 6000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00027951700334357267,
+      "loss": 1.1039,
+      "step": 6001
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002794784922545871,
+      "loss": 1.0392,
+      "step": 6002
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.000279439977665765,
+      "loss": 1.0931,
+      "step": 6003
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002794014595788026,
+      "loss": 1.0045,
+      "step": 6004
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002793629379953959,
+      "loss": 1.0297,
+      "step": 6005
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00027932441291724126,
+      "loss": 1.0903,
+      "step": 6006
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002792858843460351,
+      "loss": 0.9758,
+      "step": 6007
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00027924735228347403,
+      "loss": 1.0967,
+      "step": 6008
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00027920881673125476,
+      "loss": 1.0563,
+      "step": 6009
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00027917027769107437,
+      "loss": 1.1119,
+      "step": 6010
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00027913173516462976,
+      "loss": 1.0993,
+      "step": 6011
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00027909318915361823,
+      "loss": 1.0845,
+      "step": 6012
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00027905463965973717,
+      "loss": 1.0746,
+      "step": 6013
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.000279016086684684,
+      "loss": 1.0462,
+      "step": 6014
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002789775302301565,
+      "loss": 1.0608,
+      "step": 6015
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002789389702978525,
+      "loss": 1.1163,
+      "step": 6016
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00027890040688946993,
+      "loss": 1.0203,
+      "step": 6017
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00027886184000670697,
+      "loss": 1.0589,
+      "step": 6018
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002788232696512619,
+      "loss": 1.1059,
+      "step": 6019
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002787846958248332,
+      "loss": 1.0035,
+      "step": 6020
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002787461185291194,
+      "loss": 1.086,
+      "step": 6021
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00027870753776581934,
+      "loss": 1.0051,
+      "step": 6022
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00027866895353663185,
+      "loss": 1.1487,
+      "step": 6023
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00027863036584325606,
+      "loss": 1.0275,
+      "step": 6024
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002785917746873911,
+      "loss": 1.0704,
+      "step": 6025
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00027855318007073633,
+      "loss": 1.1213,
+      "step": 6026
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00027851458199499137,
+      "loss": 1.0261,
+      "step": 6027
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002784759804618557,
+      "loss": 1.0383,
+      "step": 6028
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002784373754730294,
+      "loss": 1.0909,
+      "step": 6029
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002783987670302122,
+      "loss": 1.0422,
+      "step": 6030
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002783601551351044,
+      "loss": 1.0973,
+      "step": 6031
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00027832153978940615,
+      "loss": 1.0706,
+      "step": 6032
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00027828292099481793,
+      "loss": 1.0946,
+      "step": 6033
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002782442987530403,
+      "loss": 1.1207,
+      "step": 6034
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.000278205673065774,
+      "loss": 1.0225,
+      "step": 6035
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002781670439347199,
+      "loss": 1.0984,
+      "step": 6036
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002781284113615791,
+      "loss": 1.0565,
+      "step": 6037
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002780897753480527,
+      "loss": 1.0446,
+      "step": 6038
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00027805113589584213,
+      "loss": 1.0551,
+      "step": 6039
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00027801249300664875,
+      "loss": 1.0197,
+      "step": 6040
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00027797384668217434,
+      "loss": 1.0593,
+      "step": 6041
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00027793519692412055,
+      "loss": 1.0603,
+      "step": 6042
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002778965437341895,
+      "loss": 1.0422,
+      "step": 6043
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00027785788711408305,
+      "loss": 1.0919,
+      "step": 6044
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00027781922706550367,
+      "loss": 1.0678,
+      "step": 6045
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00027778056359015365,
+      "loss": 1.0144,
+      "step": 6046
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00027774189668973545,
+      "loss": 1.1602,
+      "step": 6047
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00027770322636595194,
+      "loss": 1.1024,
+      "step": 6048
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00027766455262050595,
+      "loss": 1.0518,
+      "step": 6049
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002776258754551003,
+      "loss": 1.0628,
+      "step": 6050
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00027758719487143836,
+      "loss": 1.0778,
+      "step": 6051
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002775485108712233,
+      "loss": 1.0334,
+      "step": 6052
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002775098234561586,
+      "loss": 0.9906,
+      "step": 6053
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002774711326279478,
+      "loss": 1.0371,
+      "step": 6054
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00027743243838829483,
+      "loss": 0.9052,
+      "step": 6055
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00027739374073890345,
+      "loss": 1.0461,
+      "step": 6056
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002773550396814777,
+      "loss": 1.0913,
+      "step": 6057
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002773163352177218,
+      "loss": 1.1048,
+      "step": 6058
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00027727762734934016,
+      "loss": 1.0723,
+      "step": 6059
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002772389160780372,
+      "loss": 1.0704,
+      "step": 6060
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00027720020140551766,
+      "loss": 1.1665,
+      "step": 6061
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00027716148333348624,
+      "loss": 1.1533,
+      "step": 6062
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00027712276186364794,
+      "loss": 1.0165,
+      "step": 6063
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002770840369977079,
+      "loss": 1.1151,
+      "step": 6064
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00027704530873737135,
+      "loss": 1.0035,
+      "step": 6065
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00027700657708434363,
+      "loss": 1.061,
+      "step": 6066
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00027696784204033035,
+      "loss": 1.0504,
+      "step": 6067
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002769291036070372,
+      "loss": 1.091,
+      "step": 6068
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00027689036178617005,
+      "loss": 1.0922,
+      "step": 6069
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002768516165794348,
+      "loss": 1.1005,
+      "step": 6070
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00027681286798853774,
+      "loss": 1.0464,
+      "step": 6071
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.000276774116015185,
+      "loss": 1.0108,
+      "step": 6072
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002767353606610831,
+      "loss": 1.0548,
+      "step": 6073
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002766966019279387,
+      "loss": 1.1104,
+      "step": 6074
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002766578398174585,
+      "loss": 1.0648,
+      "step": 6075
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002766190743313493,
+      "loss": 1.092,
+      "step": 6076
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002765803054713183,
+      "loss": 1.0909,
+      "step": 6077
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00027654153323907247,
+      "loss": 1.0953,
+      "step": 6078
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002765027576363194,
+      "loss": 1.0048,
+      "step": 6079
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00027646397866476637,
+      "loss": 1.0613,
+      "step": 6080
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00027642519632612117,
+      "loss": 1.045,
+      "step": 6081
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002763864106220915,
+      "loss": 1.0518,
+      "step": 6082
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00027634762155438514,
+      "loss": 1.0142,
+      "step": 6083
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002763088291247104,
+      "loss": 1.1187,
+      "step": 6084
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00027627003333477543,
+      "loss": 1.1301,
+      "step": 6085
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002762312341862886,
+      "loss": 1.101,
+      "step": 6086
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002761924316809585,
+      "loss": 1.0611,
+      "step": 6087
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002761536258204937,
+      "loss": 1.0451,
+      "step": 6088
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00027611481660660297,
+      "loss": 1.0563,
+      "step": 6089
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00027607600404099534,
+      "loss": 1.0749,
+      "step": 6090
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00027603718812538,
+      "loss": 1.0236,
+      "step": 6091
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00027599836886146607,
+      "loss": 1.094,
+      "step": 6092
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002759595462509631,
+      "loss": 1.0535,
+      "step": 6093
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00027592072029558053,
+      "loss": 1.0793,
+      "step": 6094
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.000275881890997028,
+      "loss": 1.0692,
+      "step": 6095
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002758430583570156,
+      "loss": 1.0368,
+      "step": 6096
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00027580422237725316,
+      "loss": 1.0615,
+      "step": 6097
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002757653830594508,
+      "loss": 1.1023,
+      "step": 6098
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00027572654040531886,
+      "loss": 0.9991,
+      "step": 6099
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002756876944165678,
+      "loss": 1.039,
+      "step": 6100
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002756488450949082,
+      "loss": 1.0847,
+      "step": 6101
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00027560999244205075,
+      "loss": 0.9713,
+      "step": 6102
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002755711364597063,
+      "loss": 1.0378,
+      "step": 6103
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002755322771495859,
+      "loss": 1.0344,
+      "step": 6104
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00027549341451340084,
+      "loss": 1.0874,
+      "step": 6105
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002754545485528623,
+      "loss": 1.077,
+      "step": 6106
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002754156792696817,
+      "loss": 1.0444,
+      "step": 6107
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00027537680666557085,
+      "loss": 1.0409,
+      "step": 6108
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00027533793074224126,
+      "loss": 1.1092,
+      "step": 6109
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002752990515014051,
+      "loss": 1.0792,
+      "step": 6110
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002752601689447741,
+      "loss": 1.1944,
+      "step": 6111
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00027522128307406075,
+      "loss": 1.1341,
+      "step": 6112
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00027518239389097724,
+      "loss": 1.0995,
+      "step": 6113
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002751435013972361,
+      "loss": 1.1266,
+      "step": 6114
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00027510460559454985,
+      "loss": 1.0549,
+      "step": 6115
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00027506570648463144,
+      "loss": 1.0356,
+      "step": 6116
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002750268040691937,
+      "loss": 1.021,
+      "step": 6117
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00027498789834994974,
+      "loss": 1.0795,
+      "step": 6118
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00027494898932861275,
+      "loss": 1.0533,
+      "step": 6119
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00027491007700689604,
+      "loss": 1.0048,
+      "step": 6120
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00027487116138651314,
+      "loss": 1.0478,
+      "step": 6121
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002748322424691778,
+      "loss": 1.0377,
+      "step": 6122
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00027479332025660364,
+      "loss": 1.1196,
+      "step": 6123
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002747543947505048,
+      "loss": 1.0824,
+      "step": 6124
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00027471546595259517,
+      "loss": 1.1576,
+      "step": 6125
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002746765338645891,
+      "loss": 1.0218,
+      "step": 6126
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00027463759848820094,
+      "loss": 1.1178,
+      "step": 6127
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002745986598251452,
+      "loss": 1.0289,
+      "step": 6128
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002745597178771365,
+      "loss": 1.0402,
+      "step": 6129
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00027452077264588975,
+      "loss": 0.983,
+      "step": 6130
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002744818241331199,
+      "loss": 1.0416,
+      "step": 6131
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00027444287234054194,
+      "loss": 1.0993,
+      "step": 6132
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00027440391726987117,
+      "loss": 1.0762,
+      "step": 6133
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.000274364958922823,
+      "loss": 1.1356,
+      "step": 6134
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00027432599730111285,
+      "loss": 1.0695,
+      "step": 6135
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00027428703240645665,
+      "loss": 1.059,
+      "step": 6136
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00027424806424057,
+      "loss": 1.0384,
+      "step": 6137
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00027420909280516885,
+      "loss": 1.0736,
+      "step": 6138
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00027417011810196946,
+      "loss": 1.0541,
+      "step": 6139
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002741311401326879,
+      "loss": 1.047,
+      "step": 6140
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00027409215889904077,
+      "loss": 1.0463,
+      "step": 6141
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00027405317440274443,
+      "loss": 1.0799,
+      "step": 6142
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00027401418664551566,
+      "loss": 1.084,
+      "step": 6143
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002739751956290713,
+      "loss": 0.9813,
+      "step": 6144
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002739362013551282,
+      "loss": 1.1164,
+      "step": 6145
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002738972038254036,
+      "loss": 1.0524,
+      "step": 6146
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002738582030416147,
+      "loss": 1.0875,
+      "step": 6147
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00027381919900547894,
+      "loss": 1.1234,
+      "step": 6148
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002737801917187139,
+      "loss": 1.0698,
+      "step": 6149
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002737411811830371,
+      "loss": 1.1314,
+      "step": 6150
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002737021674001665,
+      "loss": 1.0999,
+      "step": 6151
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00027366315037182003,
+      "loss": 1.0131,
+      "step": 6152
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002736241300997158,
+      "loss": 1.0168,
+      "step": 6153
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00027358510658557215,
+      "loss": 1.0818,
+      "step": 6154
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002735460798311074,
+      "loss": 1.0638,
+      "step": 6155
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002735070498380401,
+      "loss": 1.0218,
+      "step": 6156
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002734680166080889,
+      "loss": 1.0978,
+      "step": 6157
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00027342898014297274,
+      "loss": 1.0649,
+      "step": 6158
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002733899404444105,
+      "loss": 0.9441,
+      "step": 6159
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00027335089751412135,
+      "loss": 1.0608,
+      "step": 6160
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002733118513538245,
+      "loss": 1.1304,
+      "step": 6161
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002732728019652394,
+      "loss": 1.0365,
+      "step": 6162
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002732337493500855,
+      "loss": 1.0871,
+      "step": 6163
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00027319469351008257,
+      "loss": 1.0331,
+      "step": 6164
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00027315563444695035,
+      "loss": 1.1121,
+      "step": 6165
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00027311657216240893,
+      "loss": 1.0867,
+      "step": 6166
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00027307750665817834,
+      "loss": 1.0739,
+      "step": 6167
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002730384379359788,
+      "loss": 1.0664,
+      "step": 6168
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002729993659975308,
+      "loss": 1.1087,
+      "step": 6169
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002729602908445547,
+      "loss": 1.0882,
+      "step": 6170
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002729212124787714,
+      "loss": 1.0538,
+      "step": 6171
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002728821309019016,
+      "loss": 1.1076,
+      "step": 6172
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002728430461156662,
+      "loss": 1.1711,
+      "step": 6173
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00027280395812178635,
+      "loss": 1.0281,
+      "step": 6174
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00027276486692198336,
+      "loss": 1.0419,
+      "step": 6175
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002727257725179785,
+      "loss": 1.1027,
+      "step": 6176
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002726866749114934,
+      "loss": 1.0879,
+      "step": 6177
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002726475741042496,
+      "loss": 1.1138,
+      "step": 6178
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.000272608470097969,
+      "loss": 1.0365,
+      "step": 6179
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002725693628943736,
+      "loss": 1.0738,
+      "step": 6180
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002725302524951853,
+      "loss": 1.0502,
+      "step": 6181
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002724911389021265,
+      "loss": 1.0752,
+      "step": 6182
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002724520221169194,
+      "loss": 1.0873,
+      "step": 6183
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002724129021412867,
+      "loss": 1.0159,
+      "step": 6184
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002723737789769509,
+      "loss": 1.0659,
+      "step": 6185
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002723346526256349,
+      "loss": 1.0779,
+      "step": 6186
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002722955230890615,
+      "loss": 1.0513,
+      "step": 6187
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002722563903689539,
+      "loss": 1.0558,
+      "step": 6188
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002722172544670353,
+      "loss": 1.0651,
+      "step": 6189
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00027217811538502893,
+      "loss": 1.0129,
+      "step": 6190
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002721389731246584,
+      "loss": 1.1102,
+      "step": 6191
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002720998276876472,
+      "loss": 1.1068,
+      "step": 6192
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002720606790757193,
+      "loss": 1.1166,
+      "step": 6193
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00027202152729059845,
+      "loss": 1.065,
+      "step": 6194
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002719823723340087,
+      "loss": 1.0446,
+      "step": 6195
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00027194321420767435,
+      "loss": 1.0831,
+      "step": 6196
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002719040529133197,
+      "loss": 1.0585,
+      "step": 6197
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00027186488845266915,
+      "loss": 1.0373,
+      "step": 6198
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002718257208274474,
+      "loss": 1.0034,
+      "step": 6199
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002717865500393791,
+      "loss": 1.0517,
+      "step": 6200
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00027174737609018913,
+      "loss": 1.099,
+      "step": 6201
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00027170819898160263,
+      "loss": 1.0363,
+      "step": 6202
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00027166901871534464,
+      "loss": 1.0256,
+      "step": 6203
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00027162983529314055,
+      "loss": 1.0895,
+      "step": 6204
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002715906487167158,
+      "loss": 1.0695,
+      "step": 6205
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.000271551458987796,
+      "loss": 1.069,
+      "step": 6206
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00027151226610810665,
+      "loss": 1.0926,
+      "step": 6207
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002714730700793739,
+      "loss": 1.0883,
+      "step": 6208
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002714338709033236,
+      "loss": 1.1416,
+      "step": 6209
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002713946685816819,
+      "loss": 1.0186,
+      "step": 6210
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00027135546311617505,
+      "loss": 1.1659,
+      "step": 6211
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00027131625450852956,
+      "loss": 0.9847,
+      "step": 6212
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002712770427604718,
+      "loss": 1.0743,
+      "step": 6213
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00027123782787372874,
+      "loss": 1.0964,
+      "step": 6214
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002711986098500269,
+      "loss": 1.0662,
+      "step": 6215
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002711593886910935,
+      "loss": 1.0467,
+      "step": 6216
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00027112016439865546,
+      "loss": 1.0671,
+      "step": 6217
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002710809369744402,
+      "loss": 1.0527,
+      "step": 6218
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00027104170642017493,
+      "loss": 1.0901,
+      "step": 6219
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00027100247273758725,
+      "loss": 1.0908,
+      "step": 6220
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00027096323592840473,
+      "loss": 1.1004,
+      "step": 6221
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002709239959943553,
+      "loss": 0.9995,
+      "step": 6222
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002708847529371668,
+      "loss": 1.0509,
+      "step": 6223
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002708455067585674,
+      "loss": 1.1365,
+      "step": 6224
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00027080625746028514,
+      "loss": 1.0731,
+      "step": 6225
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002707670050440485,
+      "loss": 1.0882,
+      "step": 6226
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00027072774951158584,
+      "loss": 0.9682,
+      "step": 6227
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00027068849086462596,
+      "loss": 1.1671,
+      "step": 6228
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002706492291048974,
+      "loss": 1.0556,
+      "step": 6229
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00027060996423412933,
+      "loss": 1.0491,
+      "step": 6230
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002705706962540505,
+      "loss": 1.0934,
+      "step": 6231
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002705314251663901,
+      "loss": 1.0763,
+      "step": 6232
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002704921509728777,
+      "loss": 1.0906,
+      "step": 6233
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002704528736752424,
+      "loss": 1.0643,
+      "step": 6234
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00027041359327521395,
+      "loss": 1.0639,
+      "step": 6235
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00027037430977452215,
+      "loss": 1.0499,
+      "step": 6236
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002703350231748967,
+      "loss": 1.0833,
+      "step": 6237
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002702957334780676,
+      "loss": 1.1548,
+      "step": 6238
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00027025644068576503,
+      "loss": 1.0667,
+      "step": 6239
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002702171447997192,
+      "loss": 1.0202,
+      "step": 6240
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002701778458216605,
+      "loss": 1.1091,
+      "step": 6241
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00027013854375331956,
+      "loss": 1.0644,
+      "step": 6242
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.000270099238596427,
+      "loss": 1.0316,
+      "step": 6243
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00027005993035271355,
+      "loss": 1.0465,
+      "step": 6244
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00027002061902391014,
+      "loss": 1.0501,
+      "step": 6245
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026998130461174796,
+      "loss": 1.0324,
+      "step": 6246
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026994198711795817,
+      "loss": 1.0168,
+      "step": 6247
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002699026665442721,
+      "loss": 1.0482,
+      "step": 6248
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026986334289242123,
+      "loss": 1.146,
+      "step": 6249
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002698240161641372,
+      "loss": 0.9672,
+      "step": 6250
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002697846863611518,
+      "loss": 1.0293,
+      "step": 6251
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026974535348519673,
+      "loss": 1.0252,
+      "step": 6252
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002697060175380043,
+      "loss": 1.1535,
+      "step": 6253
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002696666785213064,
+      "loss": 1.0264,
+      "step": 6254
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026962733643683557,
+      "loss": 1.074,
+      "step": 6255
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002695879912863241,
+      "loss": 1.1246,
+      "step": 6256
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002695486430715045,
+      "loss": 1.0543,
+      "step": 6257
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026950929179410957,
+      "loss": 1.0008,
+      "step": 6258
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002694699374558721,
+      "loss": 1.0523,
+      "step": 6259
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026943058005852506,
+      "loss": 1.0494,
+      "step": 6260
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026939121960380157,
+      "loss": 1.1021,
+      "step": 6261
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002693518560934349,
+      "loss": 1.0709,
+      "step": 6262
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002693124895291584,
+      "loss": 1.0547,
+      "step": 6263
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002692731199127055,
+      "loss": 0.9982,
+      "step": 6264
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002692337472458099,
+      "loss": 1.0277,
+      "step": 6265
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002691943715302054,
+      "loss": 1.045,
+      "step": 6266
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.000269154992767626,
+      "loss": 1.0216,
+      "step": 6267
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026911561095980553,
+      "loss": 0.9825,
+      "step": 6268
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002690762261084783,
+      "loss": 1.0566,
+      "step": 6269
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026903683821537857,
+      "loss": 1.037,
+      "step": 6270
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002689974472822408,
+      "loss": 1.0864,
+      "step": 6271
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002689580533107996,
+      "loss": 1.0311,
+      "step": 6272
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026891865630278966,
+      "loss": 1.0587,
+      "step": 6273
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026887925625994585,
+      "loss": 1.052,
+      "step": 6274
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002688398531840032,
+      "loss": 1.0067,
+      "step": 6275
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002688004470766966,
+      "loss": 1.076,
+      "step": 6276
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002687610379397616,
+      "loss": 1.065,
+      "step": 6277
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026872162577493337,
+      "loss": 1.0162,
+      "step": 6278
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026868221058394754,
+      "loss": 1.0298,
+      "step": 6279
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002686427923685397,
+      "loss": 1.0586,
+      "step": 6280
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002686033711304457,
+      "loss": 1.022,
+      "step": 6281
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026856394687140126,
+      "loss": 1.0486,
+      "step": 6282
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002685245195931427,
+      "loss": 1.0278,
+      "step": 6283
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.000268485089297406,
+      "loss": 1.1041,
+      "step": 6284
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026844565598592755,
+      "loss": 1.1063,
+      "step": 6285
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002684062196604438,
+      "loss": 1.0681,
+      "step": 6286
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026836678032269136,
+      "loss": 1.0812,
+      "step": 6287
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002683273379744068,
+      "loss": 1.0254,
+      "step": 6288
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002682878926173272,
+      "loss": 1.0985,
+      "step": 6289
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026824844425318926,
+      "loss": 1.0825,
+      "step": 6290
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002682089928837303,
+      "loss": 1.0972,
+      "step": 6291
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002681695385106875,
+      "loss": 0.9924,
+      "step": 6292
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026813008113579827,
+      "loss": 1.093,
+      "step": 6293
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002680906207608,
+      "loss": 1.0124,
+      "step": 6294
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002680511573874304,
+      "loss": 1.1106,
+      "step": 6295
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026801169101742727,
+      "loss": 0.9723,
+      "step": 6296
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002679722216525285,
+      "loss": 1.0674,
+      "step": 6297
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026793274929447203,
+      "loss": 1.0701,
+      "step": 6298
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026789327394499614,
+      "loss": 1.0401,
+      "step": 6299
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002678537956058391,
+      "loss": 0.959,
+      "step": 6300
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002678143142787393,
+      "loss": 1.0894,
+      "step": 6301
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002677748299654354,
+      "loss": 1.0647,
+      "step": 6302
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026773534266766595,
+      "loss": 0.9981,
+      "step": 6303
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00026769585238716983,
+      "loss": 1.0236,
+      "step": 6304
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002676563591256861,
+      "loss": 1.0714,
+      "step": 6305
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00026761686288495374,
+      "loss": 1.0282,
+      "step": 6306
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002675773636667119,
+      "loss": 1.1181,
+      "step": 6307
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002675378614727,
+      "loss": 1.0625,
+      "step": 6308
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00026749835630465767,
+      "loss": 1.0781,
+      "step": 6309
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002674588481643243,
+      "loss": 1.0649,
+      "step": 6310
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00026741933705343976,
+      "loss": 1.013,
+      "step": 6311
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00026737982297374383,
+      "loss": 1.0922,
+      "step": 6312
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00026734030592697654,
+      "loss": 0.9928,
+      "step": 6313
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002673007859148781,
+      "loss": 1.0494,
+      "step": 6314
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002672612629391886,
+      "loss": 1.0517,
+      "step": 6315
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00026722173700164867,
+      "loss": 1.0824,
+      "step": 6316
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002671822081039987,
+      "loss": 1.0718,
+      "step": 6317
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002671426762479793,
+      "loss": 1.0101,
+      "step": 6318
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002671031414353314,
+      "loss": 1.0849,
+      "step": 6319
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002670636036677958,
+      "loss": 1.0627,
+      "step": 6320
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002670240629471136,
+      "loss": 1.0479,
+      "step": 6321
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002669845192750259,
+      "loss": 1.1009,
+      "step": 6322
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002669449726532741,
+      "loss": 1.0924,
+      "step": 6323
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00026690542308359963,
+      "loss": 1.0647,
+      "step": 6324
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.000266865870567744,
+      "loss": 1.079,
+      "step": 6325
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00026682631510744893,
+      "loss": 1.0341,
+      "step": 6326
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002667867567044562,
+      "loss": 1.0074,
+      "step": 6327
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00026674719536050784,
+      "loss": 1.0771,
+      "step": 6328
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00026670763107734596,
+      "loss": 1.0565,
+      "step": 6329
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002666680638567126,
+      "loss": 1.0369,
+      "step": 6330
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00026662849370035033,
+      "loss": 1.0583,
+      "step": 6331
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00026658892061000143,
+      "loss": 1.0934,
+      "step": 6332
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002665493445874086,
+      "loss": 1.005,
+      "step": 6333
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00026650976563431463,
+      "loss": 1.0174,
+      "step": 6334
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00026647018375246224,
+      "loss": 1.1123,
+      "step": 6335
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00026643059894359455,
+      "loss": 1.079,
+      "step": 6336
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002663910112094546,
+      "loss": 1.0812,
+      "step": 6337
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002663514205517856,
+      "loss": 1.0728,
+      "step": 6338
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002663118269723309,
+      "loss": 1.0973,
+      "step": 6339
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002662722304728342,
+      "loss": 1.0694,
+      "step": 6340
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00026623263105503893,
+      "loss": 0.9776,
+      "step": 6341
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.000266193028720689,
+      "loss": 1.0903,
+      "step": 6342
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00026615342347152823,
+      "loss": 1.0816,
+      "step": 6343
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00026611381530930055,
+      "loss": 1.0933,
+      "step": 6344
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00026607420423575026,
+      "loss": 0.9688,
+      "step": 6345
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002660345902526215,
+      "loss": 1.0162,
+      "step": 6346
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00026599497336165877,
+      "loss": 1.0418,
+      "step": 6347
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00026595535356460664,
+      "loss": 1.1108,
+      "step": 6348
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002659157308632096,
+      "loss": 1.0362,
+      "step": 6349
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00026587610525921266,
+      "loss": 1.0558,
+      "step": 6350
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00026583647675436047,
+      "loss": 1.0439,
+      "step": 6351
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00026579684535039824,
+      "loss": 1.0481,
+      "step": 6352
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00026575721104907116,
+      "loss": 1.0009,
+      "step": 6353
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00026571757385212446,
+      "loss": 1.043,
+      "step": 6354
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00026567793376130355,
+      "loss": 1.0413,
+      "step": 6355
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002656382907783541,
+      "loss": 1.0917,
+      "step": 6356
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00026559864490502156,
+      "loss": 1.0705,
+      "step": 6357
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00026555899614305194,
+      "loss": 1.0428,
+      "step": 6358
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002655193444941912,
+      "loss": 1.0043,
+      "step": 6359
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002654796899601852,
+      "loss": 1.0795,
+      "step": 6360
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002654400325427802,
+      "loss": 1.0608,
+      "step": 6361
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00026540037224372266,
+      "loss": 1.0113,
+      "step": 6362
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002653607090647589,
+      "loss": 1.0658,
+      "step": 6363
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00026532104300763554,
+      "loss": 1.1186,
+      "step": 6364
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00026528137407409914,
+      "loss": 1.0351,
+      "step": 6365
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00026524170226589663,
+      "loss": 1.0441,
+      "step": 6366
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00026520202758477504,
+      "loss": 1.1175,
+      "step": 6367
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002651623500324813,
+      "loss": 1.1382,
+      "step": 6368
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002651226696107627,
+      "loss": 1.1275,
+      "step": 6369
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002650829863213665,
+      "loss": 1.0486,
+      "step": 6370
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00026504330016604015,
+      "loss": 1.0645,
+      "step": 6371
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002650036111465313,
+      "loss": 1.1017,
+      "step": 6372
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002649639192645877,
+      "loss": 1.0865,
+      "step": 6373
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.000264924224521957,
+      "loss": 0.9826,
+      "step": 6374
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002648845269203874,
+      "loss": 1.0155,
+      "step": 6375
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00026484482646162676,
+      "loss": 1.0402,
+      "step": 6376
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00026480512314742337,
+      "loss": 1.0336,
+      "step": 6377
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00026476541697952565,
+      "loss": 1.124,
+      "step": 6378
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002647257079596819,
+      "loss": 1.0623,
+      "step": 6379
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00026468599608964086,
+      "loss": 0.9962,
+      "step": 6380
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00026464628137115115,
+      "loss": 1.0083,
+      "step": 6381
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002646065638059617,
+      "loss": 1.0347,
+      "step": 6382
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002645668433958213,
+      "loss": 1.0878,
+      "step": 6383
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00026452712014247927,
+      "loss": 0.9991,
+      "step": 6384
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002644873940476846,
+      "loss": 1.0391,
+      "step": 6385
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00026444766511318676,
+      "loss": 1.0451,
+      "step": 6386
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002644079333407352,
+      "loss": 1.0735,
+      "step": 6387
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002643681987320795,
+      "loss": 1.0291,
+      "step": 6388
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00026432846128896935,
+      "loss": 1.1038,
+      "step": 6389
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00026428872101315467,
+      "loss": 1.0218,
+      "step": 6390
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002642489779063853,
+      "loss": 1.0245,
+      "step": 6391
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00026420923197041147,
+      "loss": 1.0852,
+      "step": 6392
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002641694832069832,
+      "loss": 1.079,
+      "step": 6393
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00026412973161785104,
+      "loss": 1.0708,
+      "step": 6394
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002640899772047653,
+      "loss": 1.0811,
+      "step": 6395
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00026405021996947666,
+      "loss": 1.0903,
+      "step": 6396
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002640104599137357,
+      "loss": 1.096,
+      "step": 6397
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002639706970392934,
+      "loss": 1.1043,
+      "step": 6398
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00026393093134790066,
+      "loss": 1.0268,
+      "step": 6399
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002638911628413086,
+      "loss": 1.0561,
+      "step": 6400
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002638513915212684,
+      "loss": 1.0095,
+      "step": 6401
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002638116173895313,
+      "loss": 1.0422,
+      "step": 6402
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00026377184044784884,
+      "loss": 1.0498,
+      "step": 6403
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00026373206069797266,
+      "loss": 1.0134,
+      "step": 6404
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00026369227814165434,
+      "loss": 1.0982,
+      "step": 6405
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002636524927806458,
+      "loss": 1.056,
+      "step": 6406
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00026361270461669895,
+      "loss": 1.0359,
+      "step": 6407
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00026357291365156587,
+      "loss": 1.0579,
+      "step": 6408
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002635331198869987,
+      "loss": 1.0466,
+      "step": 6409
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002634933233247498,
+      "loss": 1.0423,
+      "step": 6410
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00026345352396657165,
+      "loss": 1.0838,
+      "step": 6411
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00026341372181421684,
+      "loss": 1.0443,
+      "step": 6412
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00026337391686943797,
+      "loss": 1.0428,
+      "step": 6413
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002633341091339878,
+      "loss": 1.0302,
+      "step": 6414
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00026329429860961947,
+      "loss": 1.1196,
+      "step": 6415
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002632544852980858,
+      "loss": 1.1865,
+      "step": 6416
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002632146692011402,
+      "loss": 1.0491,
+      "step": 6417
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002631748503205358,
+      "loss": 1.0618,
+      "step": 6418
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002631350286580261,
+      "loss": 1.0444,
+      "step": 6419
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00026309520421536464,
+      "loss": 1.0695,
+      "step": 6420
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.000263055376994305,
+      "loss": 1.1016,
+      "step": 6421
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002630155469966011,
+      "loss": 1.0437,
+      "step": 6422
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00026297571422400685,
+      "loss": 1.0507,
+      "step": 6423
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00026293587867827614,
+      "loss": 1.0149,
+      "step": 6424
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00026289604036116334,
+      "loss": 1.0229,
+      "step": 6425
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002628561992744226,
+      "loss": 1.0308,
+      "step": 6426
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002628163554198083,
+      "loss": 0.9914,
+      "step": 6427
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.000262776508799075,
+      "loss": 1.0397,
+      "step": 6428
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00026273665941397743,
+      "loss": 1.1048,
+      "step": 6429
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00026269680726627025,
+      "loss": 0.9929,
+      "step": 6430
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002626569523577084,
+      "loss": 1.0971,
+      "step": 6431
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00026261709469004686,
+      "loss": 1.1253,
+      "step": 6432
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00026257723426504075,
+      "loss": 1.121,
+      "step": 6433
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00026253737108444537,
+      "loss": 1.0944,
+      "step": 6434
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002624975051500161,
+      "loss": 1.0621,
+      "step": 6435
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002624576364635083,
+      "loss": 1.0273,
+      "step": 6436
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002624177650266779,
+      "loss": 0.9526,
+      "step": 6437
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002623778908412803,
+      "loss": 1.0471,
+      "step": 6438
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002623380139090715,
+      "loss": 1.0711,
+      "step": 6439
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002622981342318076,
+      "loss": 1.0699,
+      "step": 6440
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002622582518112445,
+      "loss": 1.0533,
+      "step": 6441
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00026221836664913845,
+      "loss": 1.0028,
+      "step": 6442
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00026217847874724596,
+      "loss": 1.0724,
+      "step": 6443
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002621385881073234,
+      "loss": 1.0565,
+      "step": 6444
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002620986947311273,
+      "loss": 1.0672,
+      "step": 6445
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002620587986204144,
+      "loss": 1.1055,
+      "step": 6446
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00026201889977694147,
+      "loss": 1.0627,
+      "step": 6447
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002619789982024656,
+      "loss": 1.0283,
+      "step": 6448
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002619390938987438,
+      "loss": 1.0049,
+      "step": 6449
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002618991868675332,
+      "loss": 0.9798,
+      "step": 6450
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002618592771105911,
+      "loss": 1.0202,
+      "step": 6451
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.000261819364629675,
+      "loss": 1.0516,
+      "step": 6452
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002617794494265424,
+      "loss": 0.9929,
+      "step": 6453
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.000261739531502951,
+      "loss": 1.0064,
+      "step": 6454
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002616996108606585,
+      "loss": 1.02,
+      "step": 6455
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00026165968750142295,
+      "loss": 1.1126,
+      "step": 6456
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002616197614270023,
+      "loss": 1.0145,
+      "step": 6457
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002615798326391546,
+      "loss": 0.993,
+      "step": 6458
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00026153990113963826,
+      "loss": 1.0523,
+      "step": 6459
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002614999669302116,
+      "loss": 1.0711,
+      "step": 6460
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00026146003001263315,
+      "loss": 1.0034,
+      "step": 6461
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002614200903886615,
+      "loss": 1.0667,
+      "step": 6462
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002613801480600554,
+      "loss": 0.9958,
+      "step": 6463
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00026134020302857376,
+      "loss": 1.0597,
+      "step": 6464
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002613002552959755,
+      "loss": 1.0479,
+      "step": 6465
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002612603048640197,
+      "loss": 1.0735,
+      "step": 6466
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002612203517344656,
+      "loss": 1.1004,
+      "step": 6467
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00026118039590907263,
+      "loss": 1.1338,
+      "step": 6468
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00026114043738960016,
+      "loss": 1.0315,
+      "step": 6469
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00026110047617780775,
+      "loss": 1.0674,
+      "step": 6470
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00026106051227545504,
+      "loss": 1.0895,
+      "step": 6471
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.000261020545684302,
+      "loss": 0.9794,
+      "step": 6472
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002609805764061085,
+      "loss": 1.0648,
+      "step": 6473
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002609406044426345,
+      "loss": 1.0714,
+      "step": 6474
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002609006297956402,
+      "loss": 1.0552,
+      "step": 6475
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.000260860652466886,
+      "loss": 1.0687,
+      "step": 6476
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00026082067245813213,
+      "loss": 1.0636,
+      "step": 6477
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002607806897711392,
+      "loss": 1.1179,
+      "step": 6478
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002607407044076679,
+      "loss": 1.0982,
+      "step": 6479
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00026070071636947887,
+      "loss": 1.0789,
+      "step": 6480
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002606607256583331,
+      "loss": 1.0618,
+      "step": 6481
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002606207322759915,
+      "loss": 1.1109,
+      "step": 6482
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002605807362242152,
+      "loss": 1.1348,
+      "step": 6483
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00026054073750476545,
+      "loss": 1.075,
+      "step": 6484
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00026050073611940354,
+      "loss": 1.0101,
+      "step": 6485
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002604607320698909,
+      "loss": 1.0328,
+      "step": 6486
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002604207253579893,
+      "loss": 0.9824,
+      "step": 6487
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002603807159854602,
+      "loss": 1.0126,
+      "step": 6488
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002603407039540656,
+      "loss": 1.0439,
+      "step": 6489
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00026030068926556735,
+      "loss": 1.0868,
+      "step": 6490
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00026026067192172746,
+      "loss": 1.1261,
+      "step": 6491
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00026022065192430813,
+      "loss": 1.0493,
+      "step": 6492
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002601806292750717,
+      "loss": 1.0549,
+      "step": 6493
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002601406039757805,
+      "loss": 1.1407,
+      "step": 6494
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.000260100576028197,
+      "loss": 1.094,
+      "step": 6495
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00026006054543408397,
+      "loss": 1.0709,
+      "step": 6496
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.000260020512195204,
+      "loss": 1.0283,
+      "step": 6497
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002599804763133201,
+      "loss": 1.0654,
+      "step": 6498
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002599404377901951,
+      "loss": 1.017,
+      "step": 6499
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00025990039662759224,
+      "loss": 1.0736,
+      "step": 6500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00025986035282727466,
+      "loss": 1.0307,
+      "step": 6501
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002598203063910057,
+      "loss": 1.0416,
+      "step": 6502
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00025978025732054875,
+      "loss": 1.0741,
+      "step": 6503
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00025974020561766754,
+      "loss": 1.0824,
+      "step": 6504
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002597001512841255,
+      "loss": 0.9913,
+      "step": 6505
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00025966009432168663,
+      "loss": 1.0079,
+      "step": 6506
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002596200347321148,
+      "loss": 1.0422,
+      "step": 6507
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00025957997251717396,
+      "loss": 1.0329,
+      "step": 6508
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00025953990767862826,
+      "loss": 1.0717,
+      "step": 6509
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025949984021824204,
+      "loss": 1.1039,
+      "step": 6510
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002594597701377796,
+      "loss": 1.0278,
+      "step": 6511
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002594196974390054,
+      "loss": 1.0699,
+      "step": 6512
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025937962212368423,
+      "loss": 1.055,
+      "step": 6513
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025933954419358053,
+      "loss": 1.0141,
+      "step": 6514
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025929946365045933,
+      "loss": 1.0544,
+      "step": 6515
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025925938049608545,
+      "loss": 1.0879,
+      "step": 6516
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.000259219294732224,
+      "loss": 1.0156,
+      "step": 6517
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002591792063606402,
+      "loss": 1.0839,
+      "step": 6518
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002591391153830994,
+      "loss": 1.027,
+      "step": 6519
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002590990218013668,
+      "loss": 1.0383,
+      "step": 6520
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025905892561720815,
+      "loss": 1.1351,
+      "step": 6521
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025901882683238887,
+      "loss": 1.0018,
+      "step": 6522
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025897872544867486,
+      "loss": 0.9882,
+      "step": 6523
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025893862146783196,
+      "loss": 1.0588,
+      "step": 6524
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002588985148916261,
+      "loss": 0.9618,
+      "step": 6525
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002588584057218235,
+      "loss": 1.0192,
+      "step": 6526
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025881829396019013,
+      "loss": 1.0463,
+      "step": 6527
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025877817960849256,
+      "loss": 1.0209,
+      "step": 6528
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002587380626684971,
+      "loss": 1.1268,
+      "step": 6529
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002586979431419703,
+      "loss": 1.0196,
+      "step": 6530
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025865782103067886,
+      "loss": 0.9738,
+      "step": 6531
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025861769633638953,
+      "loss": 1.0059,
+      "step": 6532
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002585775690608692,
+      "loss": 1.0621,
+      "step": 6533
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002585374392058849,
+      "loss": 1.0438,
+      "step": 6534
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025849730677320374,
+      "loss": 0.9895,
+      "step": 6535
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025845717176459297,
+      "loss": 1.0312,
+      "step": 6536
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025841703418181996,
+      "loss": 0.9928,
+      "step": 6537
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002583768940266521,
+      "loss": 1.0311,
+      "step": 6538
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002583367513008569,
+      "loss": 1.024,
+      "step": 6539
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002582966060062022,
+      "loss": 1.0392,
+      "step": 6540
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002582564581444557,
+      "loss": 1.061,
+      "step": 6541
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025821630771738534,
+      "loss": 0.995,
+      "step": 6542
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002581761547267592,
+      "loss": 0.9747,
+      "step": 6543
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025813599917434536,
+      "loss": 0.9973,
+      "step": 6544
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.000258095841061912,
+      "loss": 1.0081,
+      "step": 6545
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025805568039122766,
+      "loss": 1.1131,
+      "step": 6546
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002580155171640606,
+      "loss": 0.9875,
+      "step": 6547
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002579753513821796,
+      "loss": 1.0056,
+      "step": 6548
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025793518304735325,
+      "loss": 1.0494,
+      "step": 6549
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025789501216135044,
+      "loss": 1.0463,
+      "step": 6550
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025785483872594006,
+      "loss": 1.0167,
+      "step": 6551
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002578146627428911,
+      "loss": 1.0168,
+      "step": 6552
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002577744842139727,
+      "loss": 1.0273,
+      "step": 6553
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025773430314095426,
+      "loss": 1.1217,
+      "step": 6554
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.000257694119525605,
+      "loss": 1.0795,
+      "step": 6555
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002576539333696945,
+      "loss": 1.0876,
+      "step": 6556
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025761374467499237,
+      "loss": 1.0493,
+      "step": 6557
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002575735534432682,
+      "loss": 1.0592,
+      "step": 6558
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025753335967629195,
+      "loss": 0.9615,
+      "step": 6559
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025749316337583345,
+      "loss": 0.9885,
+      "step": 6560
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002574529645436628,
+      "loss": 1.0796,
+      "step": 6561
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002574127631815501,
+      "loss": 1.04,
+      "step": 6562
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025737255929126577,
+      "loss": 1.0421,
+      "step": 6563
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025733235287458,
+      "loss": 1.1095,
+      "step": 6564
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002572921439332634,
+      "loss": 1.0793,
+      "step": 6565
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002572519324690865,
+      "loss": 1.0377,
+      "step": 6566
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025721171848382,
+      "loss": 0.9982,
+      "step": 6567
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025717150197923487,
+      "loss": 1.1048,
+      "step": 6568
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002571312829571019,
+      "loss": 0.9703,
+      "step": 6569
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002570910614191921,
+      "loss": 1.1008,
+      "step": 6570
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002570508373672768,
+      "loss": 1.0575,
+      "step": 6571
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002570106108031271,
+      "loss": 1.0524,
+      "step": 6572
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002569703817285145,
+      "loss": 1.0542,
+      "step": 6573
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025693015014521044,
+      "loss": 0.9726,
+      "step": 6574
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002568899160549865,
+      "loss": 1.0496,
+      "step": 6575
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025684967945961435,
+      "loss": 1.0876,
+      "step": 6576
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025680944036086587,
+      "loss": 0.9658,
+      "step": 6577
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.000256769198760513,
+      "loss": 0.9692,
+      "step": 6578
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002567289546603278,
+      "loss": 1.0485,
+      "step": 6579
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025668870806208236,
+      "loss": 1.0275,
+      "step": 6580
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002566484589675489,
+      "loss": 1.0369,
+      "step": 6581
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002566082073784999,
+      "loss": 1.0884,
+      "step": 6582
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002565679532967078,
+      "loss": 1.073,
+      "step": 6583
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025652769672394513,
+      "loss": 1.0843,
+      "step": 6584
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025648743766198463,
+      "loss": 1.0239,
+      "step": 6585
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002564471761125991,
+      "loss": 0.9754,
+      "step": 6586
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002564069120775615,
+      "loss": 1.1142,
+      "step": 6587
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025636664555864485,
+      "loss": 1.1219,
+      "step": 6588
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025632637655762224,
+      "loss": 1.052,
+      "step": 6589
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002562861050762669,
+      "loss": 1.0697,
+      "step": 6590
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002562458311163523,
+      "loss": 1.0126,
+      "step": 6591
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025620555467965173,
+      "loss": 1.0024,
+      "step": 6592
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025616527576793896,
+      "loss": 1.0046,
+      "step": 6593
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025612499438298756,
+      "loss": 1.0583,
+      "step": 6594
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002560847105265713,
+      "loss": 0.9824,
+      "step": 6595
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002560444242004642,
+      "loss": 1.0378,
+      "step": 6596
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025600413540644007,
+      "loss": 1.004,
+      "step": 6597
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002559638441462731,
+      "loss": 1.0673,
+      "step": 6598
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002559235504217377,
+      "loss": 1.0908,
+      "step": 6599
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.000255883254234608,
+      "loss": 0.9883,
+      "step": 6600
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002558429555866585,
+      "loss": 1.0718,
+      "step": 6601
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002558026544796638,
+      "loss": 1.1569,
+      "step": 6602
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002557623509153985,
+      "loss": 1.056,
+      "step": 6603
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025572204489563734,
+      "loss": 1.0161,
+      "step": 6604
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002556817364221553,
+      "loss": 1.0348,
+      "step": 6605
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002556414254967273,
+      "loss": 1.0833,
+      "step": 6606
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025560111212112846,
+      "loss": 1.1023,
+      "step": 6607
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025556079629713386,
+      "loss": 1.0962,
+      "step": 6608
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025552047802651894,
+      "loss": 1.0234,
+      "step": 6609
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025548015731105913,
+      "loss": 1.0646,
+      "step": 6610
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025543983415252986,
+      "loss": 0.9986,
+      "step": 6611
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025539950855270685,
+      "loss": 1.0773,
+      "step": 6612
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025535918051336574,
+      "loss": 1.1683,
+      "step": 6613
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025531885003628245,
+      "loss": 1.0264,
+      "step": 6614
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002552785171232329,
+      "loss": 1.0685,
+      "step": 6615
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002552381817759932,
+      "loss": 1.1188,
+      "step": 6616
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002551978439963395,
+      "loss": 1.0973,
+      "step": 6617
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.000255157503786048,
+      "loss": 1.0615,
+      "step": 6618
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025511716114689516,
+      "loss": 1.0359,
+      "step": 6619
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002550768160806574,
+      "loss": 0.9901,
+      "step": 6620
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025503646858911146,
+      "loss": 1.0833,
+      "step": 6621
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002549961186740339,
+      "loss": 1.0243,
+      "step": 6622
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025495576633720145,
+      "loss": 1.1255,
+      "step": 6623
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002549154115803913,
+      "loss": 1.0238,
+      "step": 6624
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025487505440538027,
+      "loss": 0.9963,
+      "step": 6625
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002548346948139455,
+      "loss": 1.071,
+      "step": 6626
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025479433280786436,
+      "loss": 1.1217,
+      "step": 6627
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025475396838891396,
+      "loss": 1.0178,
+      "step": 6628
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.000254713601558872,
+      "loss": 1.0769,
+      "step": 6629
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025467323231951586,
+      "loss": 1.0565,
+      "step": 6630
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002546328606726233,
+      "loss": 1.0102,
+      "step": 6631
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.000254592486619972,
+      "loss": 1.0961,
+      "step": 6632
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002545521101633399,
+      "loss": 1.0797,
+      "step": 6633
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025451173130450487,
+      "loss": 0.9769,
+      "step": 6634
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025447135004524516,
+      "loss": 1.0225,
+      "step": 6635
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025443096638733885,
+      "loss": 1.0602,
+      "step": 6636
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002543905803325643,
+      "loss": 1.0236,
+      "step": 6637
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025435019188269986,
+      "loss": 1.0672,
+      "step": 6638
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025430980103952397,
+      "loss": 1.0036,
+      "step": 6639
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025426940780481536,
+      "loss": 1.0558,
+      "step": 6640
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002542290121803528,
+      "loss": 1.1395,
+      "step": 6641
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002541886141679149,
+      "loss": 1.0796,
+      "step": 6642
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025414821376928075,
+      "loss": 1.0071,
+      "step": 6643
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025410781098622934,
+      "loss": 1.1017,
+      "step": 6644
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025406740582053974,
+      "loss": 1.0144,
+      "step": 6645
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002540269982739913,
+      "loss": 1.0832,
+      "step": 6646
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002539865883483633,
+      "loss": 1.0131,
+      "step": 6647
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025394617604543524,
+      "loss": 1.1153,
+      "step": 6648
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002539057613669866,
+      "loss": 1.1063,
+      "step": 6649
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002538653443147972,
+      "loss": 1.0226,
+      "step": 6650
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002538249248906466,
+      "loss": 1.0505,
+      "step": 6651
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002537845030963148,
+      "loss": 1.0545,
+      "step": 6652
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025374407893358165,
+      "loss": 1.0795,
+      "step": 6653
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025370365240422736,
+      "loss": 1.0479,
+      "step": 6654
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025366322351003215,
+      "loss": 1.0823,
+      "step": 6655
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025362279225277614,
+      "loss": 1.0443,
+      "step": 6656
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025358235863423984,
+      "loss": 1.1302,
+      "step": 6657
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002535419226562038,
+      "loss": 1.091,
+      "step": 6658
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002535014843204484,
+      "loss": 1.0797,
+      "step": 6659
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002534610436287545,
+      "loss": 1.0606,
+      "step": 6660
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025342060058290293,
+      "loss": 1.0735,
+      "step": 6661
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002533801551846745,
+      "loss": 1.0292,
+      "step": 6662
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025333970743585035,
+      "loss": 0.9993,
+      "step": 6663
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025329925733821147,
+      "loss": 1.0595,
+      "step": 6664
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025325880489353905,
+      "loss": 1.0649,
+      "step": 6665
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002532183501036146,
+      "loss": 1.094,
+      "step": 6666
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002531778929702194,
+      "loss": 1.0156,
+      "step": 6667
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002531374334951351,
+      "loss": 1.0415,
+      "step": 6668
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002530969716801432,
+      "loss": 1.0935,
+      "step": 6669
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025305650752702544,
+      "loss": 1.117,
+      "step": 6670
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002530160410375638,
+      "loss": 1.0584,
+      "step": 6671
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025297557221354007,
+      "loss": 1.0586,
+      "step": 6672
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025293510105673636,
+      "loss": 1.0562,
+      "step": 6673
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002528946275689349,
+      "loss": 1.0589,
+      "step": 6674
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002528541517519178,
+      "loss": 1.01,
+      "step": 6675
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025281367360746745,
+      "loss": 0.9971,
+      "step": 6676
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025277319313736637,
+      "loss": 1.01,
+      "step": 6677
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025273271034339704,
+      "loss": 1.0426,
+      "step": 6678
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025269222522734215,
+      "loss": 1.0521,
+      "step": 6679
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002526517377909845,
+      "loss": 1.1369,
+      "step": 6680
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.000252611248036107,
+      "loss": 1.1156,
+      "step": 6681
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002525707559644924,
+      "loss": 1.0413,
+      "step": 6682
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.000252530261577924,
+      "loss": 1.1117,
+      "step": 6683
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002524897648781848,
+      "loss": 0.9838,
+      "step": 6684
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002524492658670582,
+      "loss": 0.9876,
+      "step": 6685
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025240876454632753,
+      "loss": 1.05,
+      "step": 6686
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025236826091777626,
+      "loss": 1.044,
+      "step": 6687
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.000252327754983188,
+      "loss": 1.0591,
+      "step": 6688
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025228724674434633,
+      "loss": 1.0616,
+      "step": 6689
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002522467362030351,
+      "loss": 1.0453,
+      "step": 6690
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025220622336103826,
+      "loss": 1.1038,
+      "step": 6691
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025216570822013965,
+      "loss": 1.1023,
+      "step": 6692
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002521251907821235,
+      "loss": 1.0394,
+      "step": 6693
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025208467104877397,
+      "loss": 1.0511,
+      "step": 6694
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002520441490218752,
+      "loss": 1.1415,
+      "step": 6695
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002520036247032117,
+      "loss": 0.9697,
+      "step": 6696
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025196309809456796,
+      "loss": 1.0381,
+      "step": 6697
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002519225691977285,
+      "loss": 0.9931,
+      "step": 6698
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025188203801447816,
+      "loss": 0.9988,
+      "step": 6699
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002518415045466016,
+      "loss": 1.1256,
+      "step": 6700
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025180096879588373,
+      "loss": 1.0949,
+      "step": 6701
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002517604307641095,
+      "loss": 1.0717,
+      "step": 6702
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025171989045306417,
+      "loss": 0.9864,
+      "step": 6703
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025167934786453274,
+      "loss": 1.0251,
+      "step": 6704
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025163880300030063,
+      "loss": 1.0189,
+      "step": 6705
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025159825586215315,
+      "loss": 1.0231,
+      "step": 6706
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002515577064518759,
+      "loss": 0.9641,
+      "step": 6707
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025151715477125434,
+      "loss": 1.0064,
+      "step": 6708
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002514766008220742,
+      "loss": 1.0232,
+      "step": 6709
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002514360446061214,
+      "loss": 1.0977,
+      "step": 6710
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002513954861251817,
+      "loss": 0.9935,
+      "step": 6711
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025135492538104113,
+      "loss": 1.0741,
+      "step": 6712
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002513143623754858,
+      "loss": 1.062,
+      "step": 6713
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002512737971103018,
+      "loss": 1.002,
+      "step": 6714
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002512332295872756,
+      "loss": 1.0382,
+      "step": 6715
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025119265980819345,
+      "loss": 1.0181,
+      "step": 6716
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002511520877748419,
+      "loss": 1.0938,
+      "step": 6717
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025111151348900754,
+      "loss": 1.105,
+      "step": 6718
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002510709369524771,
+      "loss": 1.0405,
+      "step": 6719
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025103035816703717,
+      "loss": 1.0606,
+      "step": 6720
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002509897771344749,
+      "loss": 1.0658,
+      "step": 6721
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025094919385657715,
+      "loss": 1.0475,
+      "step": 6722
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.000250908608335131,
+      "loss": 1.0872,
+      "step": 6723
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002508680205719237,
+      "loss": 1.0626,
+      "step": 6724
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025082743056874246,
+      "loss": 1.0431,
+      "step": 6725
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025078683832737466,
+      "loss": 0.9849,
+      "step": 6726
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002507462438496078,
+      "loss": 1.0979,
+      "step": 6727
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002507056471372295,
+      "loss": 1.1874,
+      "step": 6728
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002506650481920273,
+      "loss": 1.159,
+      "step": 6729
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025062444701578915,
+      "loss": 1.0904,
+      "step": 6730
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002505838436103029,
+      "loss": 1.047,
+      "step": 6731
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002505432379773564,
+      "loss": 0.9964,
+      "step": 6732
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002505026301187378,
+      "loss": 1.0128,
+      "step": 6733
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002504620200362352,
+      "loss": 0.9944,
+      "step": 6734
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025042140773163694,
+      "loss": 1.0987,
+      "step": 6735
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025038079320673144,
+      "loss": 1.1082,
+      "step": 6736
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002503401764633071,
+      "loss": 1.0947,
+      "step": 6737
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002502995575031524,
+      "loss": 1.027,
+      "step": 6738
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002502589363280561,
+      "loss": 1.0236,
+      "step": 6739
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025021831293980685,
+      "loss": 1.0468,
+      "step": 6740
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025017768734019363,
+      "loss": 1.0437,
+      "step": 6741
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002501370595310053,
+      "loss": 1.1031,
+      "step": 6742
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002500964295140309,
+      "loss": 1.0348,
+      "step": 6743
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002500557972910596,
+      "loss": 1.0604,
+      "step": 6744
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002500151628638807,
+      "loss": 1.0552,
+      "step": 6745
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00024997452623428345,
+      "loss": 1.0754,
+      "step": 6746
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002499338874040574,
+      "loss": 1.0746,
+      "step": 6747
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00024989324637499186,
+      "loss": 1.0079,
+      "step": 6748
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00024985260314887666,
+      "loss": 1.1786,
+      "step": 6749
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00024981195772750154,
+      "loss": 1.0349,
+      "step": 6750
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002497713101126562,
+      "loss": 1.0508,
+      "step": 6751
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002497306603061305,
+      "loss": 1.0407,
+      "step": 6752
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00024969000830971463,
+      "loss": 1.0963,
+      "step": 6753
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00024964935412519863,
+      "loss": 1.0704,
+      "step": 6754
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002496086977543727,
+      "loss": 1.0414,
+      "step": 6755
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00024956803919902714,
+      "loss": 1.0611,
+      "step": 6756
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002495273784609524,
+      "loss": 1.0843,
+      "step": 6757
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002494867155419389,
+      "loss": 0.9991,
+      "step": 6758
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00024944605044377726,
+      "loss": 1.0603,
+      "step": 6759
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00024940538316825815,
+      "loss": 0.932,
+      "step": 6760
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00024936471371717246,
+      "loss": 1.0402,
+      "step": 6761
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00024932404209231096,
+      "loss": 1.0181,
+      "step": 6762
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002492833682954646,
+      "loss": 1.0147,
+      "step": 6763
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00024924269232842457,
+      "loss": 0.9731,
+      "step": 6764
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00024920201419298193,
+      "loss": 0.9922,
+      "step": 6765
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.000249161333890928,
+      "loss": 1.0842,
+      "step": 6766
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002491206514240541,
+      "loss": 1.1349,
+      "step": 6767
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002490799667941517,
+      "loss": 1.1077,
+      "step": 6768
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002490392800030124,
+      "loss": 1.0117,
+      "step": 6769
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002489985910524278,
+      "loss": 1.1193,
+      "step": 6770
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002489578999441896,
+      "loss": 1.0916,
+      "step": 6771
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002489172066800897,
+      "loss": 1.0265,
+      "step": 6772
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00024887651126191993,
+      "loss": 1.0646,
+      "step": 6773
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002488358136914725,
+      "loss": 1.0355,
+      "step": 6774
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00024879511397053934,
+      "loss": 1.1077,
+      "step": 6775
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002487544121009127,
+      "loss": 1.0039,
+      "step": 6776
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002487137080843849,
+      "loss": 1.053,
+      "step": 6777
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002486730019227484,
+      "loss": 1.0175,
+      "step": 6778
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002486322936177957,
+      "loss": 1.0493,
+      "step": 6779
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002485915831713193,
+      "loss": 0.992,
+      "step": 6780
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.000248550870585112,
+      "loss": 1.0183,
+      "step": 6781
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00024851015586096635,
+      "loss": 1.099,
+      "step": 6782
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00024846943900067557,
+      "loss": 1.0664,
+      "step": 6783
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002484287200060323,
+      "loss": 1.0127,
+      "step": 6784
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002483879988788298,
+      "loss": 1.0681,
+      "step": 6785
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00024834727562086115,
+      "loss": 0.9822,
+      "step": 6786
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002483065502339197,
+      "loss": 0.9409,
+      "step": 6787
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002482658227197986,
+      "loss": 1.0404,
+      "step": 6788
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002482250930802915,
+      "loss": 1.0529,
+      "step": 6789
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00024818436131719175,
+      "loss": 1.0115,
+      "step": 6790
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.000248143627432293,
+      "loss": 1.0107,
+      "step": 6791
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00024810289142738914,
+      "loss": 1.0719,
+      "step": 6792
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002480621533042738,
+      "loss": 1.0508,
+      "step": 6793
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.000248021413064741,
+      "loss": 1.0662,
+      "step": 6794
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00024798067071058464,
+      "loss": 1.0584,
+      "step": 6795
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00024793992624359883,
+      "loss": 1.0893,
+      "step": 6796
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002478991796655778,
+      "loss": 1.1399,
+      "step": 6797
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00024785843097831575,
+      "loss": 1.031,
+      "step": 6798
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002478176801836072,
+      "loss": 0.9712,
+      "step": 6799
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002477769272832464,
+      "loss": 1.1198,
+      "step": 6800
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002477361722790281,
+      "loss": 1.0287,
+      "step": 6801
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00024769541517274675,
+      "loss": 1.0667,
+      "step": 6802
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002476546559661973,
+      "loss": 1.0309,
+      "step": 6803
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00024761389466117445,
+      "loss": 1.1251,
+      "step": 6804
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002475731312594732,
+      "loss": 1.015,
+      "step": 6805
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00024753236576288847,
+      "loss": 1.0683,
+      "step": 6806
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00024749159817321546,
+      "loss": 1.0595,
+      "step": 6807
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002474508284922493,
+      "loss": 1.1327,
+      "step": 6808
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002474100567217853,
+      "loss": 1.2069,
+      "step": 6809
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00024736928286361886,
+      "loss": 1.0992,
+      "step": 6810
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00024732850691954557,
+      "loss": 1.0475,
+      "step": 6811
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00024728772889136085,
+      "loss": 1.0571,
+      "step": 6812
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002472469487808603,
+      "loss": 1.0871,
+      "step": 6813
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002472061665898399,
+      "loss": 1.0769,
+      "step": 6814
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00024716538232009527,
+      "loss": 1.0461,
+      "step": 6815
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00024712459597342246,
+      "loss": 1.1059,
+      "step": 6816
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00024708380755161753,
+      "loss": 1.0466,
+      "step": 6817
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002470430170564765,
+      "loss": 1.0235,
+      "step": 6818
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00024700222448979575,
+      "loss": 1.0273,
+      "step": 6819
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00024696142985337135,
+      "loss": 1.0929,
+      "step": 6820
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002469206331489998,
+      "loss": 1.1232,
+      "step": 6821
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00024687983437847764,
+      "loss": 1.0629,
+      "step": 6822
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002468390335436013,
+      "loss": 0.9904,
+      "step": 6823
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002467982306461677,
+      "loss": 0.9978,
+      "step": 6824
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002467574256879733,
+      "loss": 1.0684,
+      "step": 6825
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002467166186708151,
+      "loss": 1.0623,
+      "step": 6826
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00024667580959649,
+      "loss": 1.0507,
+      "step": 6827
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002466349984667951,
+      "loss": 1.0542,
+      "step": 6828
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002465941852835274,
+      "loss": 1.0333,
+      "step": 6829
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00024655337004848426,
+      "loss": 0.9822,
+      "step": 6830
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002465125527634629,
+      "loss": 1.0024,
+      "step": 6831
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00024647173343026054,
+      "loss": 1.1005,
+      "step": 6832
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00024643091205067496,
+      "loss": 1.0527,
+      "step": 6833
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002463900886265035,
+      "loss": 1.0128,
+      "step": 6834
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.000246349263159544,
+      "loss": 1.0546,
+      "step": 6835
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002463084356515941,
+      "loss": 0.9912,
+      "step": 6836
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00024626760610445167,
+      "loss": 1.0308,
+      "step": 6837
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002462267745199146,
+      "loss": 1.0477,
+      "step": 6838
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.000246185940899781,
+      "loss": 1.0574,
+      "step": 6839
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00024614510524584884,
+      "loss": 1.0662,
+      "step": 6840
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002461042675599164,
+      "loss": 1.0991,
+      "step": 6841
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00024606342784378203,
+      "loss": 1.0611,
+      "step": 6842
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00024602258609924404,
+      "loss": 1.0516,
+      "step": 6843
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00024598174232810083,
+      "loss": 1.1273,
+      "step": 6844
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002459408965321511,
+      "loss": 1.0617,
+      "step": 6845
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002459000487131934,
+      "loss": 1.0869,
+      "step": 6846
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00024585919887302646,
+      "loss": 1.0883,
+      "step": 6847
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002458183470134491,
+      "loss": 1.1016,
+      "step": 6848
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002457774931362603,
+      "loss": 1.093,
+      "step": 6849
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00024573663724325906,
+      "loss": 1.078,
+      "step": 6850
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002456957793362444,
+      "loss": 0.9431,
+      "step": 6851
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002456549194170155,
+      "loss": 1.0449,
+      "step": 6852
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002456140574873717,
+      "loss": 1.0361,
+      "step": 6853
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00024557319354911225,
+      "loss": 1.0443,
+      "step": 6854
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002455323276040367,
+      "loss": 1.0442,
+      "step": 6855
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002454914596539446,
+      "loss": 1.1201,
+      "step": 6856
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002454505897006354,
+      "loss": 1.0203,
+      "step": 6857
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00024540971774590897,
+      "loss": 1.0908,
+      "step": 6858
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002453688437915651,
+      "loss": 1.031,
+      "step": 6859
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00024532796783940355,
+      "loss": 1.0468,
+      "step": 6860
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002452870898912244,
+      "loss": 1.0972,
+      "step": 6861
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002452462099488277,
+      "loss": 1.0428,
+      "step": 6862
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002452053280140136,
+      "loss": 1.0364,
+      "step": 6863
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002451644440885823,
+      "loss": 1.037,
+      "step": 6864
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002451235581743341,
+      "loss": 1.0657,
+      "step": 6865
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002450826702730695,
+      "loss": 1.0417,
+      "step": 6866
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00024504178038658894,
+      "loss": 1.0507,
+      "step": 6867
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00024500088851669305,
+      "loss": 1.0174,
+      "step": 6868
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00024495999466518246,
+      "loss": 1.0654,
+      "step": 6869
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.000244919098833858,
+      "loss": 1.0713,
+      "step": 6870
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00024487820102452035,
+      "loss": 1.0375,
+      "step": 6871
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00024483730123897064,
+      "loss": 0.9761,
+      "step": 6872
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00024479639947900975,
+      "loss": 1.1267,
+      "step": 6873
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002447554957464389,
+      "loss": 1.0874,
+      "step": 6874
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00024471459004305927,
+      "loss": 1.0465,
+      "step": 6875
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00024467368237067206,
+      "loss": 1.0941,
+      "step": 6876
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002446327727310787,
+      "loss": 1.0561,
+      "step": 6877
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00024459186112608066,
+      "loss": 1.0617,
+      "step": 6878
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002445509475574795,
+      "loss": 1.0507,
+      "step": 6879
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002445100320270767,
+      "loss": 1.0687,
+      "step": 6880
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002444691145366742,
+      "loss": 1.0036,
+      "step": 6881
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002444281950880736,
+      "loss": 0.9882,
+      "step": 6882
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002443872736830769,
+      "loss": 1.1329,
+      "step": 6883
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00024434635032348614,
+      "loss": 1.0039,
+      "step": 6884
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002443054250111032,
+      "loss": 0.9889,
+      "step": 6885
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002442644977477303,
+      "loss": 1.0102,
+      "step": 6886
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002442235685351698,
+      "loss": 0.941,
+      "step": 6887
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00024418263737522385,
+      "loss": 0.9429,
+      "step": 6888
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00024414170426969494,
+      "loss": 1.0165,
+      "step": 6889
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00024410076922038546,
+      "loss": 1.1507,
+      "step": 6890
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00024405983222909808,
+      "loss": 1.0691,
+      "step": 6891
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002440188932976355,
+      "loss": 1.1373,
+      "step": 6892
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00024397795242780035,
+      "loss": 1.0619,
+      "step": 6893
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00024393700962139554,
+      "loss": 1.0609,
+      "step": 6894
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00024389606488022403,
+      "loss": 1.0519,
+      "step": 6895
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00024385511820608868,
+      "loss": 1.0482,
+      "step": 6896
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00024381416960079263,
+      "loss": 1.0826,
+      "step": 6897
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00024377321906613912,
+      "loss": 1.0804,
+      "step": 6898
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00024373226660393134,
+      "loss": 1.0419,
+      "step": 6899
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00024369131221597278,
+      "loss": 1.0591,
+      "step": 6900
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00024365035590406665,
+      "loss": 1.0756,
+      "step": 6901
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00024360939767001648,
+      "loss": 1.0367,
+      "step": 6902
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00024356843751562605,
+      "loss": 1.0342,
+      "step": 6903
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002435274754426989,
+      "loss": 1.0413,
+      "step": 6904
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00024348651145303884,
+      "loss": 1.0219,
+      "step": 6905
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00024344554554844966,
+      "loss": 0.9997,
+      "step": 6906
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00024340457773073545,
+      "loss": 1.0375,
+      "step": 6907
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002433636080017,
+      "loss": 1.0998,
+      "step": 6908
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00024332263636314761,
+      "loss": 0.9972,
+      "step": 6909
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00024328166281688236,
+      "loss": 1.036,
+      "step": 6910
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00024324068736470854,
+      "loss": 1.0327,
+      "step": 6911
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00024319971000843054,
+      "loss": 1.0743,
+      "step": 6912
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00024315873074985276,
+      "loss": 1.1026,
+      "step": 6913
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002431177495907798,
+      "loss": 0.9771,
+      "step": 6914
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00024307676653301613,
+      "loss": 1.0923,
+      "step": 6915
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00024303578157836646,
+      "loss": 1.0628,
+      "step": 6916
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00024299479472863572,
+      "loss": 1.0704,
+      "step": 6917
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00024295380598562857,
+      "loss": 1.0483,
+      "step": 6918
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00024291281535115014,
+      "loss": 0.9992,
+      "step": 6919
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00024287182282700532,
+      "loss": 1.0353,
+      "step": 6920
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002428308284149992,
+      "loss": 1.1401,
+      "step": 6921
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00024278983211693702,
+      "loss": 1.0648,
+      "step": 6922
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00024274883393462414,
+      "loss": 1.0805,
+      "step": 6923
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00024270783386986575,
+      "loss": 1.0405,
+      "step": 6924
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00024266683192446738,
+      "loss": 1.0264,
+      "step": 6925
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00024262582810023457,
+      "loss": 0.9973,
+      "step": 6926
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00024258482239897285,
+      "loss": 1.0274,
+      "step": 6927
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00024254381482248798,
+      "loss": 1.0025,
+      "step": 6928
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00024250280537258564,
+      "loss": 1.0307,
+      "step": 6929
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00024246179405107174,
+      "loss": 1.0625,
+      "step": 6930
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002424207808597523,
+      "loss": 0.966,
+      "step": 6931
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002423797658004332,
+      "loss": 1.0405,
+      "step": 6932
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00024233874887492055,
+      "loss": 1.0847,
+      "step": 6933
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00024229773008502067,
+      "loss": 1.0597,
+      "step": 6934
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002422567094325396,
+      "loss": 1.0288,
+      "step": 6935
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00024221568691928385,
+      "loss": 0.9967,
+      "step": 6936
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00024217466254705983,
+      "loss": 1.0575,
+      "step": 6937
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00024213363631767402,
+      "loss": 0.9978,
+      "step": 6938
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00024209260823293297,
+      "loss": 1.0283,
+      "step": 6939
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002420515782946434,
+      "loss": 1.0348,
+      "step": 6940
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00024201054650461207,
+      "loss": 0.9985,
+      "step": 6941
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00024196951286464585,
+      "loss": 1.0518,
+      "step": 6942
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00024192847737655152,
+      "loss": 1.0737,
+      "step": 6943
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00024188744004213626,
+      "loss": 1.0806,
+      "step": 6944
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.000241846400863207,
+      "loss": 1.0161,
+      "step": 6945
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00024180535984157095,
+      "loss": 1.0052,
+      "step": 6946
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00024176431697903538,
+      "loss": 1.0721,
+      "step": 6947
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002417232722774076,
+      "loss": 1.0791,
+      "step": 6948
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00024168222573849502,
+      "loss": 1.0241,
+      "step": 6949
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002416411773641051,
+      "loss": 1.0469,
+      "step": 6950
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00024160012715604542,
+      "loss": 1.0721,
+      "step": 6951
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00024155907511612363,
+      "loss": 1.0381,
+      "step": 6952
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00024151802124614747,
+      "loss": 1.0646,
+      "step": 6953
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00024147696554792467,
+      "loss": 1.0251,
+      "step": 6954
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00024143590802326316,
+      "loss": 1.081,
+      "step": 6955
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.000241394848673971,
+      "loss": 1.0728,
+      "step": 6956
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00024135378750185616,
+      "loss": 1.0079,
+      "step": 6957
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00024131272450872673,
+      "loss": 1.0277,
+      "step": 6958
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.000241271659696391,
+      "loss": 1.0354,
+      "step": 6959
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002412305930666572,
+      "loss": 1.1056,
+      "step": 6960
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00024118952462133373,
+      "loss": 1.074,
+      "step": 6961
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00024114845436222902,
+      "loss": 1.0559,
+      "step": 6962
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00024110738229115165,
+      "loss": 1.0392,
+      "step": 6963
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00024106630840991014,
+      "loss": 1.1427,
+      "step": 6964
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002410252327203133,
+      "loss": 1.0545,
+      "step": 6965
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00024098415522416975,
+      "loss": 1.0001,
+      "step": 6966
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00024094307592328846,
+      "loss": 1.0621,
+      "step": 6967
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002409019948194783,
+      "loss": 1.0531,
+      "step": 6968
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00024086091191454832,
+      "loss": 1.0587,
+      "step": 6969
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00024081982721030758,
+      "loss": 1.0285,
+      "step": 6970
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00024077874070856518,
+      "loss": 1.0827,
+      "step": 6971
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00024073765241113045,
+      "loss": 1.0872,
+      "step": 6972
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002406965623198127,
+      "loss": 1.0792,
+      "step": 6973
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002406554704364213,
+      "loss": 0.9534,
+      "step": 6974
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002406143767627658,
+      "loss": 1.1017,
+      "step": 6975
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00024057328130065573,
+      "loss": 1.0486,
+      "step": 6976
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00024053218405190063,
+      "loss": 1.0489,
+      "step": 6977
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00024049108501831033,
+      "loss": 1.0128,
+      "step": 6978
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00024044998420169464,
+      "loss": 1.0417,
+      "step": 6979
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00024040888160386335,
+      "loss": 1.0145,
+      "step": 6980
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00024036777722662652,
+      "loss": 1.0175,
+      "step": 6981
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002403266710717941,
+      "loss": 1.022,
+      "step": 6982
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00024028556314117618,
+      "loss": 0.9958,
+      "step": 6983
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00024024445343658298,
+      "loss": 1.0712,
+      "step": 6984
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002402033419598248,
+      "loss": 0.9791,
+      "step": 6985
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00024016222871271192,
+      "loss": 1.0917,
+      "step": 6986
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00024012111369705488,
+      "loss": 1.0364,
+      "step": 6987
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002400799969146641,
+      "loss": 1.0448,
+      "step": 6988
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00024003887836735008,
+      "loss": 1.0355,
+      "step": 6989
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00023999775805692362,
+      "loss": 1.083,
+      "step": 6990
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00023995663598519533,
+      "loss": 1.0525,
+      "step": 6991
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00023991551215397604,
+      "loss": 1.0408,
+      "step": 6992
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00023987438656507677,
+      "loss": 1.0047,
+      "step": 6993
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002398332592203084,
+      "loss": 1.0706,
+      "step": 6994
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00023979213012148196,
+      "loss": 1.0159,
+      "step": 6995
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00023975099927040853,
+      "loss": 1.1155,
+      "step": 6996
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00023970986666889936,
+      "loss": 1.0176,
+      "step": 6997
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00023966873231876572,
+      "loss": 1.0947,
+      "step": 6998
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00023962759622181897,
+      "loss": 1.0213,
+      "step": 6999
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00023958645837987054,
+      "loss": 1.1,
+      "step": 7000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00023954531879473193,
+      "loss": 0.9696,
+      "step": 7001
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002395041774682147,
+      "loss": 1.0045,
+      "step": 7002
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00023946303440213046,
+      "loss": 1.1035,
+      "step": 7003
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002394218895982911,
+      "loss": 1.0377,
+      "step": 7004
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002393807430585083,
+      "loss": 1.0711,
+      "step": 7005
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00023933959478459403,
+      "loss": 1.1064,
+      "step": 7006
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002392984447783602,
+      "loss": 1.1157,
+      "step": 7007
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00023925729304161885,
+      "loss": 1.0242,
+      "step": 7008
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00023921613957618215,
+      "loss": 1.0624,
+      "step": 7009
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00023917498438386222,
+      "loss": 1.0567,
+      "step": 7010
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00023913382746647136,
+      "loss": 1.0365,
+      "step": 7011
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00023909266882582192,
+      "loss": 1.1208,
+      "step": 7012
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00023905150846372636,
+      "loss": 1.029,
+      "step": 7013
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00023901034638199712,
+      "loss": 1.0026,
+      "step": 7014
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002389691825824468,
+      "loss": 1.1259,
+      "step": 7015
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00023892801706688804,
+      "loss": 1.0948,
+      "step": 7016
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00023888684983713353,
+      "loss": 1.0654,
+      "step": 7017
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00023884568089499616,
+      "loss": 1.0099,
+      "step": 7018
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00023880451024228875,
+      "loss": 0.9811,
+      "step": 7019
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00023876333788082422,
+      "loss": 1.0374,
+      "step": 7020
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00023872216381241567,
+      "loss": 1.0074,
+      "step": 7021
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00023868098803887607,
+      "loss": 1.0706,
+      "step": 7022
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00023863981056201873,
+      "loss": 1.0596,
+      "step": 7023
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023859863138365683,
+      "loss": 1.0238,
+      "step": 7024
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023855745050560377,
+      "loss": 1.0857,
+      "step": 7025
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002385162679296729,
+      "loss": 0.9959,
+      "step": 7026
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023847508365767764,
+      "loss": 1.0997,
+      "step": 7027
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023843389769143158,
+      "loss": 1.0516,
+      "step": 7028
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023839271003274842,
+      "loss": 1.0823,
+      "step": 7029
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023835152068344174,
+      "loss": 1.1125,
+      "step": 7030
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023831032964532543,
+      "loss": 0.9898,
+      "step": 7031
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023826913692021325,
+      "loss": 1.0232,
+      "step": 7032
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023822794250991916,
+      "loss": 1.0984,
+      "step": 7033
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023818674641625713,
+      "loss": 1.0457,
+      "step": 7034
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023814554864104123,
+      "loss": 0.9878,
+      "step": 7035
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023810434918608561,
+      "loss": 1.0881,
+      "step": 7036
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002380631480532046,
+      "loss": 1.0329,
+      "step": 7037
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023802194524421232,
+      "loss": 1.0762,
+      "step": 7038
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023798074076092322,
+      "loss": 1.0178,
+      "step": 7039
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023793953460515176,
+      "loss": 1.0464,
+      "step": 7040
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023789832677871236,
+      "loss": 1.0206,
+      "step": 7041
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002378571172834197,
+      "loss": 1.1128,
+      "step": 7042
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023781590612108843,
+      "loss": 1.0332,
+      "step": 7043
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002377746932935333,
+      "loss": 0.9856,
+      "step": 7044
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023773347880256895,
+      "loss": 1.0168,
+      "step": 7045
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023769226265001053,
+      "loss": 1.0565,
+      "step": 7046
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023765104483767276,
+      "loss": 1.0518,
+      "step": 7047
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002376098253673708,
+      "loss": 1.0751,
+      "step": 7048
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002375686042409197,
+      "loss": 1.0152,
+      "step": 7049
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023752738146013468,
+      "loss": 1.0945,
+      "step": 7050
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023748615702683096,
+      "loss": 1.0208,
+      "step": 7051
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023744493094282382,
+      "loss": 1.0397,
+      "step": 7052
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023740370320992865,
+      "loss": 1.0189,
+      "step": 7053
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.000237362473829961,
+      "loss": 1.0168,
+      "step": 7054
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002373212428047363,
+      "loss": 1.0275,
+      "step": 7055
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002372800101360703,
+      "loss": 1.0145,
+      "step": 7056
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023723877582577853,
+      "loss": 1.06,
+      "step": 7057
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023719753987567677,
+      "loss": 1.0226,
+      "step": 7058
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002371563022875809,
+      "loss": 1.046,
+      "step": 7059
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023711506306330686,
+      "loss": 1.0436,
+      "step": 7060
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002370738222046705,
+      "loss": 1.057,
+      "step": 7061
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023703257971348794,
+      "loss": 1.0712,
+      "step": 7062
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023699133559157523,
+      "loss": 1.0471,
+      "step": 7063
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023695008984074868,
+      "loss": 1.1165,
+      "step": 7064
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023690884246282439,
+      "loss": 1.0645,
+      "step": 7065
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023686759345961878,
+      "loss": 1.0282,
+      "step": 7066
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023682634283294824,
+      "loss": 1.0385,
+      "step": 7067
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023678509058462925,
+      "loss": 1.0155,
+      "step": 7068
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023674383671647836,
+      "loss": 0.9583,
+      "step": 7069
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023670258123031207,
+      "loss": 1.0023,
+      "step": 7070
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002366613241279473,
+      "loss": 1.0734,
+      "step": 7071
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023662006541120057,
+      "loss": 1.0343,
+      "step": 7072
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002365788050818888,
+      "loss": 1.0765,
+      "step": 7073
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023653754314182895,
+      "loss": 1.1494,
+      "step": 7074
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023649627959283792,
+      "loss": 1.0469,
+      "step": 7075
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023645501443673276,
+      "loss": 1.0273,
+      "step": 7076
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023641374767533057,
+      "loss": 1.0777,
+      "step": 7077
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023637247931044853,
+      "loss": 1.0023,
+      "step": 7078
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023633120934390394,
+      "loss": 1.0853,
+      "step": 7079
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023628993777751408,
+      "loss": 1.0681,
+      "step": 7080
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002362486646130964,
+      "loss": 1.0618,
+      "step": 7081
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023620738985246834,
+      "loss": 1.0688,
+      "step": 7082
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023616611349744734,
+      "loss": 1.0834,
+      "step": 7083
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002361248355498511,
+      "loss": 1.0392,
+      "step": 7084
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002360835560114973,
+      "loss": 0.9888,
+      "step": 7085
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023604227488420367,
+      "loss": 0.99,
+      "step": 7086
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023600099216978802,
+      "loss": 1.0448,
+      "step": 7087
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023595970787006825,
+      "loss": 1.1042,
+      "step": 7088
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023591842198686226,
+      "loss": 0.9975,
+      "step": 7089
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023587713452198812,
+      "loss": 1.108,
+      "step": 7090
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002358358454772639,
+      "loss": 1.0893,
+      "step": 7091
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002357945548545078,
+      "loss": 1.0992,
+      "step": 7092
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023575326265553806,
+      "loss": 1.0811,
+      "step": 7093
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023571196888217294,
+      "loss": 1.0459,
+      "step": 7094
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002356706735362308,
+      "loss": 0.9837,
+      "step": 7095
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002356293766195301,
+      "loss": 1.0321,
+      "step": 7096
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023558807813388937,
+      "loss": 1.0492,
+      "step": 7097
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023554677808112717,
+      "loss": 1.0577,
+      "step": 7098
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002355054764630622,
+      "loss": 0.9519,
+      "step": 7099
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023546417328151314,
+      "loss": 1.0627,
+      "step": 7100
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023542286853829873,
+      "loss": 1.0697,
+      "step": 7101
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023538156223523786,
+      "loss": 1.1009,
+      "step": 7102
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023534025437414946,
+      "loss": 0.9884,
+      "step": 7103
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023529894495685256,
+      "loss": 1.0977,
+      "step": 7104
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023525763398516612,
+      "loss": 0.9872,
+      "step": 7105
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023521632146090938,
+      "loss": 1.0305,
+      "step": 7106
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002351750073859015,
+      "loss": 1.0159,
+      "step": 7107
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023513369176196168,
+      "loss": 1.0051,
+      "step": 7108
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023509237459090927,
+      "loss": 1.0186,
+      "step": 7109
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023505105587456383,
+      "loss": 1.1047,
+      "step": 7110
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023500973561474463,
+      "loss": 1.0863,
+      "step": 7111
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023496841381327133,
+      "loss": 1.0171,
+      "step": 7112
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023492709047196345,
+      "loss": 1.0749,
+      "step": 7113
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023488576559264071,
+      "loss": 1.0818,
+      "step": 7114
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023484443917712292,
+      "loss": 1.0423,
+      "step": 7115
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023480311122722972,
+      "loss": 1.0397,
+      "step": 7116
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002347617817447811,
+      "loss": 1.0061,
+      "step": 7117
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023472045073159704,
+      "loss": 1.0353,
+      "step": 7118
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023467911818949753,
+      "loss": 1.0361,
+      "step": 7119
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002346377841203025,
+      "loss": 1.0652,
+      "step": 7120
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023459644852583233,
+      "loss": 0.9983,
+      "step": 7121
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023455511140790708,
+      "loss": 1.0991,
+      "step": 7122
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023451377276834703,
+      "loss": 1.0468,
+      "step": 7123
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023447243260897263,
+      "loss": 1.0265,
+      "step": 7124
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002344310909316042,
+      "loss": 1.0358,
+      "step": 7125
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023438974773806227,
+      "loss": 1.1167,
+      "step": 7126
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00023434840303016737,
+      "loss": 1.0398,
+      "step": 7127
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00023430705680974008,
+      "loss": 1.0613,
+      "step": 7128
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00023426570907860117,
+      "loss": 1.0484,
+      "step": 7129
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002342243598385713,
+      "loss": 1.0021,
+      "step": 7130
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00023418300909147134,
+      "loss": 1.0939,
+      "step": 7131
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00023414165683912215,
+      "loss": 1.0475,
+      "step": 7132
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00023410030308334466,
+      "loss": 0.9942,
+      "step": 7133
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00023405894782595988,
+      "loss": 1.1333,
+      "step": 7134
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00023401759106878896,
+      "loss": 1.0463,
+      "step": 7135
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00023397623281365296,
+      "loss": 1.051,
+      "step": 7136
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00023393487306237317,
+      "loss": 1.1148,
+      "step": 7137
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002338935118167708,
+      "loss": 1.037,
+      "step": 7138
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00023385214907866719,
+      "loss": 0.9771,
+      "step": 7139
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00023381078484988383,
+      "loss": 1.0445,
+      "step": 7140
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00023376941913224206,
+      "loss": 1.0589,
+      "step": 7141
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00023372805192756354,
+      "loss": 1.121,
+      "step": 7142
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00023368668323766985,
+      "loss": 1.0368,
+      "step": 7143
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00023364531306438268,
+      "loss": 1.0629,
+      "step": 7144
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00023360394140952364,
+      "loss": 0.9541,
+      "step": 7145
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00023356256827491469,
+      "loss": 1.0105,
+      "step": 7146
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002335211936623776,
+      "loss": 1.1209,
+      "step": 7147
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002334798175737343,
+      "loss": 1.1124,
+      "step": 7148
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002334384400108069,
+      "loss": 1.0541,
+      "step": 7149
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00023339706097541738,
+      "loss": 1.0336,
+      "step": 7150
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00023335568046938776,
+      "loss": 1.0157,
+      "step": 7151
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00023331429849454046,
+      "loss": 1.0554,
+      "step": 7152
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002332729150526975,
+      "loss": 1.0636,
+      "step": 7153
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00023323153014568137,
+      "loss": 1.0735,
+      "step": 7154
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002331901437753144,
+      "loss": 1.003,
+      "step": 7155
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00023314875594341908,
+      "loss": 1.0692,
+      "step": 7156
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002331073666518178,
+      "loss": 1.0559,
+      "step": 7157
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00023306597590233325,
+      "loss": 0.9737,
+      "step": 7158
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.000233024583696788,
+      "loss": 1.0835,
+      "step": 7159
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00023298319003700488,
+      "loss": 1.023,
+      "step": 7160
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002329417949248065,
+      "loss": 1.0644,
+      "step": 7161
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00023290039836201584,
+      "loss": 1.0257,
+      "step": 7162
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002328590003504557,
+      "loss": 0.9643,
+      "step": 7163
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00023281760089194906,
+      "loss": 1.0766,
+      "step": 7164
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00023277619998831895,
+      "loss": 1.0272,
+      "step": 7165
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002327347976413885,
+      "loss": 1.0772,
+      "step": 7166
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00023269339385298076,
+      "loss": 1.0071,
+      "step": 7167
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00023265198862491913,
+      "loss": 0.9819,
+      "step": 7168
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002326105819590267,
+      "loss": 1.1396,
+      "step": 7169
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002325691738571269,
+      "loss": 1.0477,
+      "step": 7170
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00023252776432104315,
+      "loss": 1.0502,
+      "step": 7171
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00023248635335259885,
+      "loss": 1.0226,
+      "step": 7172
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00023244494095361757,
+      "loss": 1.0848,
+      "step": 7173
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00023240352712592297,
+      "loss": 1.0677,
+      "step": 7174
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00023236211187133862,
+      "loss": 1.0051,
+      "step": 7175
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00023232069519168826,
+      "loss": 1.0164,
+      "step": 7176
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00023227927708879573,
+      "loss": 1.022,
+      "step": 7177
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002322378575644848,
+      "loss": 1.0125,
+      "step": 7178
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00023219643662057937,
+      "loss": 1.0452,
+      "step": 7179
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00023215501425890352,
+      "loss": 1.0348,
+      "step": 7180
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00023211359048128124,
+      "loss": 1.0709,
+      "step": 7181
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00023207216528953656,
+      "loss": 1.0396,
+      "step": 7182
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00023203073868549362,
+      "loss": 1.09,
+      "step": 7183
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00023198931067097678,
+      "loss": 1.0479,
+      "step": 7184
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00023194788124781027,
+      "loss": 1.0058,
+      "step": 7185
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00023190645041781834,
+      "loss": 1.0476,
+      "step": 7186
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00023186501818282554,
+      "loss": 1.065,
+      "step": 7187
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00023182358454465626,
+      "loss": 1.0221,
+      "step": 7188
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00023178214950513498,
+      "loss": 1.068,
+      "step": 7189
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00023174071306608638,
+      "loss": 1.008,
+      "step": 7190
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00023169927522933512,
+      "loss": 1.0289,
+      "step": 7191
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00023165783599670583,
+      "loss": 1.0605,
+      "step": 7192
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002316163953700234,
+      "loss": 1.05,
+      "step": 7193
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002315749533511126,
+      "loss": 1.0341,
+      "step": 7194
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002315335099417983,
+      "loss": 1.0842,
+      "step": 7195
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00023149206514390551,
+      "loss": 1.0522,
+      "step": 7196
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00023145061895925921,
+      "loss": 1.0205,
+      "step": 7197
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00023140917138968453,
+      "loss": 0.9963,
+      "step": 7198
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00023136772243700663,
+      "loss": 1.0969,
+      "step": 7199
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002313262721030507,
+      "loss": 0.9965,
+      "step": 7200
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002312848203896419,
+      "loss": 0.9371,
+      "step": 7201
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00023124336729860576,
+      "loss": 0.9968,
+      "step": 7202
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00023120191283176747,
+      "loss": 1.0374,
+      "step": 7203
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00023116045699095254,
+      "loss": 1.0576,
+      "step": 7204
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00023111899977798657,
+      "loss": 0.9751,
+      "step": 7205
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002310775411946951,
+      "loss": 1.0678,
+      "step": 7206
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00023103608124290363,
+      "loss": 0.978,
+      "step": 7207
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00023099461992443795,
+      "loss": 1.0195,
+      "step": 7208
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002309531572411238,
+      "loss": 1.0289,
+      "step": 7209
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.000230911693194787,
+      "loss": 1.0631,
+      "step": 7210
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00023087022778725338,
+      "loss": 1.1019,
+      "step": 7211
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002308287610203489,
+      "loss": 1.0195,
+      "step": 7212
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00023078729289589963,
+      "loss": 1.0169,
+      "step": 7213
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002307458234157314,
+      "loss": 1.1178,
+      "step": 7214
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00023070435258167052,
+      "loss": 0.9959,
+      "step": 7215
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002306628803955431,
+      "loss": 0.9729,
+      "step": 7216
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00023062140685917534,
+      "loss": 1.1153,
+      "step": 7217
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002305799319743936,
+      "loss": 1.1249,
+      "step": 7218
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00023053845574302418,
+      "loss": 1.0232,
+      "step": 7219
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002304969781668934,
+      "loss": 1.0045,
+      "step": 7220
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00023045549924782788,
+      "loss": 0.9695,
+      "step": 7221
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00023041401898765403,
+      "loss": 1.0927,
+      "step": 7222
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00023037253738819846,
+      "loss": 0.9669,
+      "step": 7223
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00023033105445128787,
+      "loss": 0.9993,
+      "step": 7224
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00023028957017874895,
+      "loss": 0.9514,
+      "step": 7225
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00023024808457240833,
+      "loss": 1.0269,
+      "step": 7226
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00023020659763409304,
+      "loss": 1.067,
+      "step": 7227
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00023016510936562976,
+      "loss": 0.9772,
+      "step": 7228
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00023012361976884553,
+      "loss": 1.076,
+      "step": 7229
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002300821288455674,
+      "loss": 1.1142,
+      "step": 7230
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00023004063659762234,
+      "loss": 1.0074,
+      "step": 7231
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00022999914302683744,
+      "loss": 1.0864,
+      "step": 7232
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002299576481350399,
+      "loss": 1.0591,
+      "step": 7233
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002299161519240569,
+      "loss": 1.0404,
+      "step": 7234
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002298746543957159,
+      "loss": 1.0448,
+      "step": 7235
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00022983315555184404,
+      "loss": 1.0086,
+      "step": 7236
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002297916553942688,
+      "loss": 1.017,
+      "step": 7237
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00022975015392481774,
+      "loss": 1.0433,
+      "step": 7238
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00022970865114531818,
+      "loss": 1.0491,
+      "step": 7239
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002296671470575978,
+      "loss": 1.028,
+      "step": 7240
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002296256416634843,
+      "loss": 1.0378,
+      "step": 7241
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00022958413496480522,
+      "loss": 0.9358,
+      "step": 7242
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00022954262696338847,
+      "loss": 1.034,
+      "step": 7243
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00022950111766106176,
+      "loss": 1.0613,
+      "step": 7244
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002294596070596529,
+      "loss": 0.9987,
+      "step": 7245
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00022941809516098993,
+      "loss": 1.099,
+      "step": 7246
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002293765819669008,
+      "loss": 1.0386,
+      "step": 7247
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00022933506747921346,
+      "loss": 0.9984,
+      "step": 7248
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00022929355169975613,
+      "loss": 1.1072,
+      "step": 7249
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00022925203463035682,
+      "loss": 1.0037,
+      "step": 7250
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00022921051627284382,
+      "loss": 1.1514,
+      "step": 7251
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00022916899662904538,
+      "loss": 1.0428,
+      "step": 7252
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00022912747570078983,
+      "loss": 0.9922,
+      "step": 7253
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00022908595348990546,
+      "loss": 1.0009,
+      "step": 7254
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00022904442999822084,
+      "loss": 1.0435,
+      "step": 7255
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002290029052275644,
+      "loss": 1.0759,
+      "step": 7256
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002289613791797646,
+      "loss": 1.0215,
+      "step": 7257
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00022891985185665017,
+      "loss": 1.0428,
+      "step": 7258
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00022887832326004963,
+      "loss": 1.0152,
+      "step": 7259
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00022883679339179183,
+      "loss": 1.0997,
+      "step": 7260
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002287952622537054,
+      "loss": 1.0717,
+      "step": 7261
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00022875372984761935,
+      "loss": 1.0945,
+      "step": 7262
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002287121961753624,
+      "loss": 1.0215,
+      "step": 7263
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002286706612387635,
+      "loss": 0.9822,
+      "step": 7264
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00022862912503965167,
+      "loss": 1.0615,
+      "step": 7265
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00022858758757985604,
+      "loss": 0.9839,
+      "step": 7266
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00022854604886120554,
+      "loss": 1.0896,
+      "step": 7267
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002285045088855295,
+      "loss": 1.0568,
+      "step": 7268
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00022846296765465705,
+      "loss": 1.0267,
+      "step": 7269
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00022842142517041746,
+      "loss": 1.108,
+      "step": 7270
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00022837988143464003,
+      "loss": 1.031,
+      "step": 7271
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002283383364491542,
+      "loss": 1.0522,
+      "step": 7272
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00022829679021578935,
+      "loss": 1.06,
+      "step": 7273
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00022825524273637502,
+      "loss": 0.992,
+      "step": 7274
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00022821369401274075,
+      "loss": 1.0938,
+      "step": 7275
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00022817214404671603,
+      "loss": 1.0607,
+      "step": 7276
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00022813059284013066,
+      "loss": 1.0091,
+      "step": 7277
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00022808904039481423,
+      "loss": 1.0509,
+      "step": 7278
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00022804748671259657,
+      "loss": 1.0166,
+      "step": 7279
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00022800593179530756,
+      "loss": 1.0547,
+      "step": 7280
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00022796437564477696,
+      "loss": 1.0257,
+      "step": 7281
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002279228182628347,
+      "loss": 0.9708,
+      "step": 7282
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00022788125965131082,
+      "loss": 1.012,
+      "step": 7283
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00022783969981203525,
+      "loss": 1.1379,
+      "step": 7284
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00022779813874683824,
+      "loss": 1.0659,
+      "step": 7285
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00022775657645754982,
+      "loss": 0.9818,
+      "step": 7286
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00022771501294600025,
+      "loss": 1.0225,
+      "step": 7287
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002276734482140197,
+      "loss": 0.9494,
+      "step": 7288
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002276318822634385,
+      "loss": 1.0053,
+      "step": 7289
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.000227590315096087,
+      "loss": 0.9931,
+      "step": 7290
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002275487467137957,
+      "loss": 1.0318,
+      "step": 7291
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00022750717711839499,
+      "loss": 0.9973,
+      "step": 7292
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00022746560631171537,
+      "loss": 1.109,
+      "step": 7293
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00022742403429558742,
+      "loss": 1.0036,
+      "step": 7294
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002273824610718418,
+      "loss": 1.0429,
+      "step": 7295
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00022734088664230913,
+      "loss": 1.0938,
+      "step": 7296
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002272993110088202,
+      "loss": 0.9837,
+      "step": 7297
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002272577341732058,
+      "loss": 1.0544,
+      "step": 7298
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00022721615613729675,
+      "loss": 1.1077,
+      "step": 7299
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00022717457690292391,
+      "loss": 1.0544,
+      "step": 7300
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00022713299647191822,
+      "loss": 1.0239,
+      "step": 7301
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00022709141484611068,
+      "loss": 1.0518,
+      "step": 7302
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002270498320273324,
+      "loss": 1.0592,
+      "step": 7303
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00022700824801741436,
+      "loss": 1.0676,
+      "step": 7304
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002269666628181878,
+      "loss": 0.9339,
+      "step": 7305
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.000226925076431484,
+      "loss": 1.0172,
+      "step": 7306
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00022688348885913402,
+      "loss": 1.0873,
+      "step": 7307
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00022684190010296935,
+      "loss": 0.948,
+      "step": 7308
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002268003101648212,
+      "loss": 1.071,
+      "step": 7309
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00022675871904652106,
+      "loss": 1.1277,
+      "step": 7310
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00022671712674990046,
+      "loss": 0.9875,
+      "step": 7311
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00022667553327679083,
+      "loss": 1.0919,
+      "step": 7312
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00022663393862902375,
+      "loss": 1.0617,
+      "step": 7313
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00022659234280843088,
+      "loss": 0.9859,
+      "step": 7314
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00022655074581684383,
+      "loss": 1.0585,
+      "step": 7315
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00022650914765609436,
+      "loss": 1.0195,
+      "step": 7316
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00022646754832801426,
+      "loss": 1.0863,
+      "step": 7317
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00022642594783443536,
+      "loss": 1.0256,
+      "step": 7318
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002263843461771895,
+      "loss": 1.0603,
+      "step": 7319
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00022634274335810858,
+      "loss": 0.9788,
+      "step": 7320
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002263011393790247,
+      "loss": 0.9675,
+      "step": 7321
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002262595342417698,
+      "loss": 0.9838,
+      "step": 7322
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00022621792794817595,
+      "loss": 0.9963,
+      "step": 7323
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00022617632050007534,
+      "loss": 1.0083,
+      "step": 7324
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00022613471189930016,
+      "loss": 1.0121,
+      "step": 7325
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002260931021476826,
+      "loss": 1.0339,
+      "step": 7326
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00022605149124705495,
+      "loss": 1.0394,
+      "step": 7327
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00022600987919924953,
+      "loss": 0.9418,
+      "step": 7328
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002259682660060988,
+      "loss": 1.0293,
+      "step": 7329
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00022592665166943518,
+      "loss": 0.9991,
+      "step": 7330
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00022588503619109108,
+      "loss": 1.0514,
+      "step": 7331
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00022584341957289911,
+      "loss": 1.0344,
+      "step": 7332
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00022580180181669185,
+      "loss": 1.0353,
+      "step": 7333
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002257601829243019,
+      "loss": 1.1185,
+      "step": 7334
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00022571856289756195,
+      "loss": 1.09,
+      "step": 7335
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00022567694173830484,
+      "loss": 1.1015,
+      "step": 7336
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00022563531944836326,
+      "loss": 1.0432,
+      "step": 7337
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00022559369602957,
+      "loss": 1.0615,
+      "step": 7338
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00022555207148375813,
+      "loss": 1.0134,
+      "step": 7339
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00022551044581276038,
+      "loss": 0.9954,
+      "step": 7340
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002254688190184099,
+      "loss": 1.0638,
+      "step": 7341
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00022542719110253963,
+      "loss": 1.034,
+      "step": 7342
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002253855620669827,
+      "loss": 0.9565,
+      "step": 7343
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00022534393191357222,
+      "loss": 1.0331,
+      "step": 7344
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00022530230064414134,
+      "loss": 1.0108,
+      "step": 7345
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00022526066826052336,
+      "loss": 1.069,
+      "step": 7346
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00022521903476455156,
+      "loss": 1.0427,
+      "step": 7347
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00022517740015805924,
+      "loss": 1.0368,
+      "step": 7348
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002251357644428798,
+      "loss": 0.9895,
+      "step": 7349
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002250941276208467,
+      "loss": 0.9843,
+      "step": 7350
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002250524896937933,
+      "loss": 1.0179,
+      "step": 7351
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00022501085066355323,
+      "loss": 1.0863,
+      "step": 7352
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00022496921053196008,
+      "loss": 1.0523,
+      "step": 7353
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00022492756930084739,
+      "loss": 1.0289,
+      "step": 7354
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00022488592697204887,
+      "loss": 0.987,
+      "step": 7355
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002248442835473983,
+      "loss": 1.0985,
+      "step": 7356
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00022480263902872937,
+      "loss": 1.0592,
+      "step": 7357
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002247609934178759,
+      "loss": 1.0407,
+      "step": 7358
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00022471934671667177,
+      "loss": 1.0571,
+      "step": 7359
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00022467769892695093,
+      "loss": 1.1013,
+      "step": 7360
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002246360500505473,
+      "loss": 1.0554,
+      "step": 7361
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00022459440008929494,
+      "loss": 1.0432,
+      "step": 7362
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002245527490450278,
+      "loss": 1.0495,
+      "step": 7363
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002245110969195801,
+      "loss": 1.0461,
+      "step": 7364
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00022446944371478592,
+      "loss": 0.962,
+      "step": 7365
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00022442778943247945,
+      "loss": 1.0244,
+      "step": 7366
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00022438613407449505,
+      "loss": 1.081,
+      "step": 7367
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002243444776426669,
+      "loss": 1.0617,
+      "step": 7368
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002243028201388294,
+      "loss": 1.0431,
+      "step": 7369
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00022426116156481688,
+      "loss": 0.9358,
+      "step": 7370
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00022421950192246377,
+      "loss": 0.9955,
+      "step": 7371
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00022417784121360463,
+      "loss": 1.045,
+      "step": 7372
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00022413617944007393,
+      "loss": 1.0604,
+      "step": 7373
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00022409451660370633,
+      "loss": 1.0105,
+      "step": 7374
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002240528527063364,
+      "loss": 1.0288,
+      "step": 7375
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002240111877497987,
+      "loss": 1.0547,
+      "step": 7376
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002239695217359281,
+      "loss": 1.0779,
+      "step": 7377
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00022392785466655935,
+      "loss": 1.0552,
+      "step": 7378
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00022388618654352714,
+      "loss": 1.0518,
+      "step": 7379
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00022384451736866647,
+      "loss": 1.0291,
+      "step": 7380
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00022380284714381224,
+      "loss": 1.0055,
+      "step": 7381
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00022376117587079923,
+      "loss": 1.0354,
+      "step": 7382
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00022371950355146262,
+      "loss": 1.0386,
+      "step": 7383
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00022367783018763734,
+      "loss": 1.0398,
+      "step": 7384
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00022363615578115852,
+      "loss": 0.9961,
+      "step": 7385
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002235944803338613,
+      "loss": 1.0788,
+      "step": 7386
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00022355280384758083,
+      "loss": 1.054,
+      "step": 7387
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00022351112632415233,
+      "loss": 0.9468,
+      "step": 7388
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00022346944776541116,
+      "loss": 1.043,
+      "step": 7389
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00022342776817319248,
+      "loss": 1.0917,
+      "step": 7390
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00022338608754933175,
+      "loss": 0.9804,
+      "step": 7391
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002233444058956644,
+      "loss": 1.0603,
+      "step": 7392
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002233027232140259,
+      "loss": 1.1065,
+      "step": 7393
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00022326103950625164,
+      "loss": 1.0775,
+      "step": 7394
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00022321935477417716,
+      "loss": 0.9598,
+      "step": 7395
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00022317766901963816,
+      "loss": 1.0206,
+      "step": 7396
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00022313598224447026,
+      "loss": 1.0998,
+      "step": 7397
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.000223094294450509,
+      "loss": 1.1393,
+      "step": 7398
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00022305260563959021,
+      "loss": 1.0559,
+      "step": 7399
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00022301091581354975,
+      "loss": 1.0806,
+      "step": 7400
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00022296922497422322,
+      "loss": 1.0261,
+      "step": 7401
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002229275331234466,
+      "loss": 1.0766,
+      "step": 7402
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002228858402630558,
+      "loss": 1.0747,
+      "step": 7403
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00022284414639488676,
+      "loss": 1.1194,
+      "step": 7404
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00022280245152077538,
+      "loss": 1.0357,
+      "step": 7405
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002227607556425579,
+      "loss": 1.0856,
+      "step": 7406
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00022271905876207013,
+      "loss": 1.0549,
+      "step": 7407
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00022267736088114838,
+      "loss": 0.9803,
+      "step": 7408
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00022263566200162873,
+      "loss": 1.0553,
+      "step": 7409
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002225939621253474,
+      "loss": 1.1182,
+      "step": 7410
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00022255226125414074,
+      "loss": 1.0499,
+      "step": 7411
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00022251055938984497,
+      "loss": 1.0569,
+      "step": 7412
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00022246885653429637,
+      "loss": 1.0636,
+      "step": 7413
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00022242715268933148,
+      "loss": 1.0298,
+      "step": 7414
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00022238544785678656,
+      "loss": 0.9865,
+      "step": 7415
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00022234374203849817,
+      "loss": 1.034,
+      "step": 7416
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002223020352363029,
+      "loss": 1.0624,
+      "step": 7417
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00022226032745203717,
+      "loss": 1.0609,
+      "step": 7418
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002222186186875377,
+      "loss": 1.1011,
+      "step": 7419
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00022217690894464105,
+      "loss": 1.0079,
+      "step": 7420
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002221351982251839,
+      "loss": 1.0616,
+      "step": 7421
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00022209348653100307,
+      "loss": 1.0483,
+      "step": 7422
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00022205177386393525,
+      "loss": 1.0573,
+      "step": 7423
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00022201006022581728,
+      "loss": 1.0689,
+      "step": 7424
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00022196834561848613,
+      "loss": 0.9392,
+      "step": 7425
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00022192663004377857,
+      "loss": 0.9603,
+      "step": 7426
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00022188491350353152,
+      "loss": 1.0502,
+      "step": 7427
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002218431959995821,
+      "loss": 1.0349,
+      "step": 7428
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002218014775337672,
+      "loss": 1.0203,
+      "step": 7429
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.000221759758107924,
+      "loss": 0.9882,
+      "step": 7430
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00022171803772388967,
+      "loss": 1.0694,
+      "step": 7431
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00022167631638350118,
+      "loss": 1.0485,
+      "step": 7432
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00022163459408859582,
+      "loss": 1.0619,
+      "step": 7433
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00022159287084101088,
+      "loss": 0.9991,
+      "step": 7434
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002215511466425836,
+      "loss": 1.0476,
+      "step": 7435
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00022150942149515133,
+      "loss": 1.0201,
+      "step": 7436
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00022146769540055136,
+      "loss": 1.0459,
+      "step": 7437
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00022142596836062124,
+      "loss": 0.9897,
+      "step": 7438
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00022138424037719827,
+      "loss": 1.0914,
+      "step": 7439
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00022134251145212002,
+      "loss": 1.0236,
+      "step": 7440
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00022130078158722402,
+      "loss": 1.0754,
+      "step": 7441
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00022125905078434785,
+      "loss": 1.0567,
+      "step": 7442
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002212173190453291,
+      "loss": 1.0145,
+      "step": 7443
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00022117558637200544,
+      "loss": 1.093,
+      "step": 7444
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00022113385276621462,
+      "loss": 1.0428,
+      "step": 7445
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00022109211822979424,
+      "loss": 1.0904,
+      "step": 7446
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00022105038276458224,
+      "loss": 1.0592,
+      "step": 7447
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00022100864637241636,
+      "loss": 1.1426,
+      "step": 7448
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00022096690905513446,
+      "loss": 1.0167,
+      "step": 7449
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002209251708145745,
+      "loss": 1.0531,
+      "step": 7450
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00022088343165257434,
+      "loss": 1.0717,
+      "step": 7451
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00022084169157097203,
+      "loss": 0.9924,
+      "step": 7452
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00022079995057160557,
+      "loss": 1.0058,
+      "step": 7453
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.000220758208656313,
+      "loss": 1.0251,
+      "step": 7454
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00022071646582693245,
+      "loss": 1.0835,
+      "step": 7455
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00022067472208530214,
+      "loss": 1.0499,
+      "step": 7456
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00022063297743326018,
+      "loss": 1.0106,
+      "step": 7457
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00022059123187264477,
+      "loss": 0.9653,
+      "step": 7458
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002205494854052942,
+      "loss": 1.0301,
+      "step": 7459
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00022050773803304678,
+      "loss": 1.0427,
+      "step": 7460
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002204659897577409,
+      "loss": 1.1172,
+      "step": 7461
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00022042424058121488,
+      "loss": 1.0387,
+      "step": 7462
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00022038249050530717,
+      "loss": 1.1024,
+      "step": 7463
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00022034073953185627,
+      "loss": 1.0415,
+      "step": 7464
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00022029898766270058,
+      "loss": 1.0774,
+      "step": 7465
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00022025723489967874,
+      "loss": 0.9395,
+      "step": 7466
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00022021548124462936,
+      "loss": 1.01,
+      "step": 7467
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00022017372669939092,
+      "loss": 1.0376,
+      "step": 7468
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002201319712658023,
+      "loss": 1.023,
+      "step": 7469
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00022009021494570204,
+      "loss": 1.1269,
+      "step": 7470
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00022004845774092882,
+      "loss": 1.0171,
+      "step": 7471
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002200066996533216,
+      "loss": 1.1582,
+      "step": 7472
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00021996494068471912,
+      "loss": 1.0549,
+      "step": 7473
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00021992318083696017,
+      "loss": 1.0319,
+      "step": 7474
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00021988142011188376,
+      "loss": 1.0788,
+      "step": 7475
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002198396585113288,
+      "loss": 1.0196,
+      "step": 7476
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00021979789603713412,
+      "loss": 1.0858,
+      "step": 7477
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00021975613269113896,
+      "loss": 1.0538,
+      "step": 7478
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00021971436847518217,
+      "loss": 1.0655,
+      "step": 7479
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00021967260339110286,
+      "loss": 1.072,
+      "step": 7480
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00021963083744074035,
+      "loss": 1.0031,
+      "step": 7481
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00021958907062593364,
+      "loss": 0.9948,
+      "step": 7482
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002195473029485219,
+      "loss": 1.0608,
+      "step": 7483
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002195055344103445,
+      "loss": 1.0334,
+      "step": 7484
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002194637650132406,
+      "loss": 1.0294,
+      "step": 7485
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002194219947590496,
+      "loss": 0.999,
+      "step": 7486
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002193802236496108,
+      "loss": 1.1257,
+      "step": 7487
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00021933845168676364,
+      "loss": 1.0563,
+      "step": 7488
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00021929667887234753,
+      "loss": 0.9505,
+      "step": 7489
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00021925490520820188,
+      "loss": 1.0233,
+      "step": 7490
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00021921313069616621,
+      "loss": 0.9506,
+      "step": 7491
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002191713553380802,
+      "loss": 1.0447,
+      "step": 7492
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002191295791357832,
+      "loss": 1.002,
+      "step": 7493
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00021908780209111504,
+      "loss": 1.1306,
+      "step": 7494
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00021904602420591527,
+      "loss": 0.9576,
+      "step": 7495
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002190042454820235,
+      "loss": 1.0822,
+      "step": 7496
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002189624659212796,
+      "loss": 1.0309,
+      "step": 7497
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002189206855255233,
+      "loss": 0.9933,
+      "step": 7498
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00021887890429659433,
+      "loss": 1.0053,
+      "step": 7499
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00021883712223633262,
+      "loss": 0.984,
+      "step": 7500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00021879533934657803,
+      "loss": 1.1164,
+      "step": 7501
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00021875355562917036,
+      "loss": 1.0296,
+      "step": 7502
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002187117710859497,
+      "loss": 1.0435,
+      "step": 7503
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002186699857187559,
+      "loss": 1.0073,
+      "step": 7504
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00021862819952942904,
+      "loss": 1.0666,
+      "step": 7505
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00021858641251980924,
+      "loss": 1.0302,
+      "step": 7506
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00021854462469173656,
+      "loss": 1.0228,
+      "step": 7507
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00021850283604705104,
+      "loss": 1.0272,
+      "step": 7508
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00021846104658759292,
+      "loss": 1.0213,
+      "step": 7509
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00021841925631520238,
+      "loss": 1.0057,
+      "step": 7510
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00021837746523171964,
+      "loss": 1.032,
+      "step": 7511
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00021833567333898505,
+      "loss": 1.007,
+      "step": 7512
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00021829388063883883,
+      "loss": 1.0568,
+      "step": 7513
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00021825208713312133,
+      "loss": 1.105,
+      "step": 7514
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00021821029282367298,
+      "loss": 0.9994,
+      "step": 7515
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00021816849771233413,
+      "loss": 1.1055,
+      "step": 7516
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00021812670180094528,
+      "loss": 1.0821,
+      "step": 7517
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00021808490509134685,
+      "loss": 1.0447,
+      "step": 7518
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00021804310758537945,
+      "loss": 0.9911,
+      "step": 7519
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002180013092848836,
+      "loss": 0.9916,
+      "step": 7520
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002179595101916998,
+      "loss": 1.0411,
+      "step": 7521
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00021791771030766876,
+      "loss": 1.0117,
+      "step": 7522
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00021787590963463118,
+      "loss": 1.0252,
+      "step": 7523
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00021783410817442764,
+      "loss": 1.0523,
+      "step": 7524
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00021779230592889896,
+      "loss": 0.9742,
+      "step": 7525
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00021775050289988586,
+      "loss": 1.0044,
+      "step": 7526
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002177086990892291,
+      "loss": 1.0346,
+      "step": 7527
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00021766689449876964,
+      "loss": 0.9547,
+      "step": 7528
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002176250891303482,
+      "loss": 1.0067,
+      "step": 7529
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00021758328298580572,
+      "loss": 1.0008,
+      "step": 7530
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00021754147606698325,
+      "loss": 1.03,
+      "step": 7531
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002174996683757216,
+      "loss": 1.0713,
+      "step": 7532
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00021745785991386183,
+      "loss": 1.1431,
+      "step": 7533
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00021741605068324504,
+      "loss": 1.1047,
+      "step": 7534
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00021737424068571216,
+      "loss": 1.0733,
+      "step": 7535
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00021733242992310437,
+      "loss": 1.0543,
+      "step": 7536
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002172906183972629,
+      "loss": 0.971,
+      "step": 7537
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002172488061100288,
+      "loss": 1.0314,
+      "step": 7538
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002172069930632433,
+      "loss": 0.9859,
+      "step": 7539
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00021716517925874764,
+      "loss": 1.0177,
+      "step": 7540
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002171233646983831,
+      "loss": 0.989,
+      "step": 7541
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.000217081549383991,
+      "loss": 1.0705,
+      "step": 7542
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002170397333174126,
+      "loss": 1.1434,
+      "step": 7543
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00021699791650048943,
+      "loss": 1.0333,
+      "step": 7544
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00021695609893506278,
+      "loss": 1.1065,
+      "step": 7545
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00021691428062297404,
+      "loss": 1.0685,
+      "step": 7546
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00021687246156606475,
+      "loss": 1.0072,
+      "step": 7547
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00021683064176617647,
+      "loss": 1.0308,
+      "step": 7548
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00021678882122515058,
+      "loss": 1.0424,
+      "step": 7549
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00021674699994482884,
+      "loss": 1.0101,
+      "step": 7550
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002167051779270527,
+      "loss": 1.0221,
+      "step": 7551
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00021666335517366383,
+      "loss": 0.9997,
+      "step": 7552
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002166215316865039,
+      "loss": 1.0523,
+      "step": 7553
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00021657970746741465,
+      "loss": 1.0608,
+      "step": 7554
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00021653788251823773,
+      "loss": 1.0212,
+      "step": 7555
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00021649605684081498,
+      "loss": 1.0603,
+      "step": 7556
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002164542304369882,
+      "loss": 1.0782,
+      "step": 7557
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002164124033085991,
+      "loss": 1.0949,
+      "step": 7558
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002163705754574897,
+      "loss": 1.0653,
+      "step": 7559
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00021632874688550175,
+      "loss": 1.0608,
+      "step": 7560
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00021628691759447718,
+      "loss": 0.9945,
+      "step": 7561
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002162450875862581,
+      "loss": 1.0619,
+      "step": 7562
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00021620325686268633,
+      "loss": 1.0473,
+      "step": 7563
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00021616142542560396,
+      "loss": 1.1577,
+      "step": 7564
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00021611959327685301,
+      "loss": 1.007,
+      "step": 7565
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002160777604182756,
+      "loss": 0.973,
+      "step": 7566
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00021603592685171373,
+      "loss": 1.0317,
+      "step": 7567
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00021599409257900968,
+      "loss": 1.0285,
+      "step": 7568
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002159522576020056,
+      "loss": 0.9673,
+      "step": 7569
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00021591042192254363,
+      "loss": 1.0935,
+      "step": 7570
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00021586858554246602,
+      "loss": 1.0275,
+      "step": 7571
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00021582674846361502,
+      "loss": 1.0093,
+      "step": 7572
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00021578491068783298,
+      "loss": 1.047,
+      "step": 7573
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002157430722169622,
+      "loss": 1.0673,
+      "step": 7574
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002157012330528451,
+      "loss": 1.0328,
+      "step": 7575
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00021565939319732395,
+      "loss": 1.0793,
+      "step": 7576
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00021561755265224122,
+      "loss": 1.0839,
+      "step": 7577
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00021557571141943936,
+      "loss": 1.1179,
+      "step": 7578
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002155338695007609,
+      "loss": 1.0675,
+      "step": 7579
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002154920268980482,
+      "loss": 1.0116,
+      "step": 7580
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00021545018361314404,
+      "loss": 1.0275,
+      "step": 7581
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002154083396478908,
+      "loss": 1.0289,
+      "step": 7582
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00021536649500413112,
+      "loss": 1.0259,
+      "step": 7583
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00021532464968370766,
+      "loss": 1.0601,
+      "step": 7584
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00021528280368846302,
+      "loss": 1.0389,
+      "step": 7585
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00021524095702023992,
+      "loss": 0.9925,
+      "step": 7586
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00021519910968088114,
+      "loss": 1.019,
+      "step": 7587
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002151572616722294,
+      "loss": 1.001,
+      "step": 7588
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00021511541299612738,
+      "loss": 1.1189,
+      "step": 7589
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.000215073563654418,
+      "loss": 0.9825,
+      "step": 7590
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00021503171364894405,
+      "loss": 1.0696,
+      "step": 7591
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00021498986298154838,
+      "loss": 1.014,
+      "step": 7592
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00021494801165407394,
+      "loss": 1.0782,
+      "step": 7593
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00021490615966836366,
+      "loss": 1.0563,
+      "step": 7594
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00021486430702626042,
+      "loss": 1.0542,
+      "step": 7595
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002148224537296072,
+      "loss": 1.0428,
+      "step": 7596
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00021478059978024706,
+      "loss": 1.0788,
+      "step": 7597
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00021473874518002304,
+      "loss": 0.9767,
+      "step": 7598
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002146968899307782,
+      "loss": 1.0628,
+      "step": 7599
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002146550340343556,
+      "loss": 1.0212,
+      "step": 7600
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00021461317749259845,
+      "loss": 0.9901,
+      "step": 7601
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00021457132030734978,
+      "loss": 1.0624,
+      "step": 7602
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00021452946248045284,
+      "loss": 0.9811,
+      "step": 7603
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002144876040137509,
+      "loss": 1.056,
+      "step": 7604
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002144457449090871,
+      "loss": 0.9651,
+      "step": 7605
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00021440388516830478,
+      "loss": 0.9465,
+      "step": 7606
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00021436202479324719,
+      "loss": 0.9845,
+      "step": 7607
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002143201637857576,
+      "loss": 1.0392,
+      "step": 7608
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00021427830214767947,
+      "loss": 0.9856,
+      "step": 7609
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002142364398808561,
+      "loss": 1.095,
+      "step": 7610
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00021419457698713091,
+      "loss": 1.0004,
+      "step": 7611
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00021415271346834742,
+      "loss": 1.0243,
+      "step": 7612
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00021411084932634896,
+      "loss": 0.9895,
+      "step": 7613
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002140689845629791,
+      "loss": 1.0949,
+      "step": 7614
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00021402711918008135,
+      "loss": 1.1097,
+      "step": 7615
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00021398525317949915,
+      "loss": 1.0559,
+      "step": 7616
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00021394338656307622,
+      "loss": 1.0452,
+      "step": 7617
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00021390151933265607,
+      "loss": 1.03,
+      "step": 7618
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00021385965149008237,
+      "loss": 1.0961,
+      "step": 7619
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002138177830371987,
+      "loss": 1.0153,
+      "step": 7620
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00021377591397584884,
+      "loss": 0.9998,
+      "step": 7621
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00021373404430787636,
+      "loss": 1.1343,
+      "step": 7622
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00021369217403512512,
+      "loss": 1.1225,
+      "step": 7623
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00021365030315943879,
+      "loss": 1.0491,
+      "step": 7624
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00021360843168266125,
+      "loss": 0.991,
+      "step": 7625
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00021356655960663625,
+      "loss": 1.0511,
+      "step": 7626
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00021352468693320757,
+      "loss": 1.0122,
+      "step": 7627
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00021348281366421912,
+      "loss": 0.9408,
+      "step": 7628
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00021344093980151487,
+      "loss": 1.0353,
+      "step": 7629
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00021339906534693866,
+      "loss": 0.9841,
+      "step": 7630
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00021335719030233442,
+      "loss": 1.0287,
+      "step": 7631
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00021331531466954617,
+      "loss": 1.0481,
+      "step": 7632
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002132734384504178,
+      "loss": 1.0375,
+      "step": 7633
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00021323156164679342,
+      "loss": 1.0211,
+      "step": 7634
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00021318968426051714,
+      "loss": 1.1509,
+      "step": 7635
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00021314780629343286,
+      "loss": 1.1213,
+      "step": 7636
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00021310592774738486,
+      "loss": 1.0413,
+      "step": 7637
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00021306404862421713,
+      "loss": 1.0556,
+      "step": 7638
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00021302216892577378,
+      "loss": 1.0614,
+      "step": 7639
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002129802886538992,
+      "loss": 1.0153,
+      "step": 7640
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00021293840781043733,
+      "loss": 1.0291,
+      "step": 7641
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002128965263972325,
+      "loss": 1.0689,
+      "step": 7642
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00021285464441612905,
+      "loss": 1.0314,
+      "step": 7643
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00021281276186897117,
+      "loss": 0.976,
+      "step": 7644
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00021277087875760313,
+      "loss": 1.0419,
+      "step": 7645
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002127289950838693,
+      "loss": 0.983,
+      "step": 7646
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00021268711084961397,
+      "loss": 1.0411,
+      "step": 7647
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002126452260566816,
+      "loss": 1.0553,
+      "step": 7648
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00021260334070691652,
+      "loss": 1.0552,
+      "step": 7649
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00021256145480216324,
+      "loss": 1.0686,
+      "step": 7650
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002125195683442661,
+      "loss": 1.0563,
+      "step": 7651
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00021247768133506956,
+      "loss": 1.0446,
+      "step": 7652
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00021243579377641823,
+      "loss": 0.9982,
+      "step": 7653
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00021239390567015653,
+      "loss": 1.1591,
+      "step": 7654
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00021235201701812906,
+      "loss": 1.1653,
+      "step": 7655
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00021231012782218042,
+      "loss": 1.0559,
+      "step": 7656
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00021226823808415506,
+      "loss": 0.995,
+      "step": 7657
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00021222634780589773,
+      "loss": 1.0072,
+      "step": 7658
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00021218445698925294,
+      "loss": 1.0921,
+      "step": 7659
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00021214256563606556,
+      "loss": 1.0758,
+      "step": 7660
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00021210067374818006,
+      "loss": 0.9725,
+      "step": 7661
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00021205878132744133,
+      "loss": 1.0237,
+      "step": 7662
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00021201688837569398,
+      "loss": 0.9689,
+      "step": 7663
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00021197499489478277,
+      "loss": 1.0293,
+      "step": 7664
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002119331008865525,
+      "loss": 0.9623,
+      "step": 7665
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00021189120635284803,
+      "loss": 1.0458,
+      "step": 7666
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002118493112955141,
+      "loss": 1.0896,
+      "step": 7667
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002118074157163956,
+      "loss": 1.0296,
+      "step": 7668
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00021176551961733746,
+      "loss": 1.1075,
+      "step": 7669
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002117236230001845,
+      "loss": 1.0214,
+      "step": 7670
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00021168172586678163,
+      "loss": 1.0902,
+      "step": 7671
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002116398282189738,
+      "loss": 1.0207,
+      "step": 7672
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.000211597930058606,
+      "loss": 1.0649,
+      "step": 7673
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00021155603138752326,
+      "loss": 1.0168,
+      "step": 7674
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002115141322075705,
+      "loss": 0.9937,
+      "step": 7675
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002114722325205928,
+      "loss": 1.0623,
+      "step": 7676
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00021143033232843514,
+      "loss": 1.1157,
+      "step": 7677
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00021138843163294267,
+      "loss": 1.0398,
+      "step": 7678
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00021134653043596054,
+      "loss": 1.0341,
+      "step": 7679
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00021130462873933373,
+      "loss": 1.0614,
+      "step": 7680
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002112627265449075,
+      "loss": 1.0118,
+      "step": 7681
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00021122082385452697,
+      "loss": 1.1207,
+      "step": 7682
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002111789206700373,
+      "loss": 1.0431,
+      "step": 7683
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00021113701699328367,
+      "loss": 1.0644,
+      "step": 7684
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00021109511282611142,
+      "loss": 1.078,
+      "step": 7685
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00021105320817036568,
+      "loss": 1.0452,
+      "step": 7686
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00021101130302789186,
+      "loss": 1.001,
+      "step": 7687
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00021096939740053517,
+      "loss": 1.0424,
+      "step": 7688
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002109274912901409,
+      "loss": 1.0016,
+      "step": 7689
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00021088558469855442,
+      "loss": 1.0635,
+      "step": 7690
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00021084367762762106,
+      "loss": 1.0235,
+      "step": 7691
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00021080177007918623,
+      "loss": 1.0071,
+      "step": 7692
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00021075986205509534,
+      "loss": 0.9977,
+      "step": 7693
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00021071795355719379,
+      "loss": 1.0545,
+      "step": 7694
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.000210676044587327,
+      "loss": 1.0085,
+      "step": 7695
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00021063413514734047,
+      "loss": 1.0583,
+      "step": 7696
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00021059222523907964,
+      "loss": 0.9943,
+      "step": 7697
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00021055031486439006,
+      "loss": 0.9532,
+      "step": 7698
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00021050840402511725,
+      "loss": 0.9602,
+      "step": 7699
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00021046649272310675,
+      "loss": 1.0412,
+      "step": 7700
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002104245809602041,
+      "loss": 1.0253,
+      "step": 7701
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00021038266873825486,
+      "loss": 1.0947,
+      "step": 7702
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002103407560591047,
+      "loss": 1.0253,
+      "step": 7703
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00021029884292459925,
+      "loss": 1.0403,
+      "step": 7704
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00021025692933658406,
+      "loss": 1.0259,
+      "step": 7705
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00021021501529690492,
+      "loss": 0.9631,
+      "step": 7706
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00021017310080740745,
+      "loss": 1.0591,
+      "step": 7707
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002101311858699373,
+      "loss": 1.0038,
+      "step": 7708
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00021008927048634028,
+      "loss": 1.0362,
+      "step": 7709
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00021004735465846212,
+      "loss": 1.0731,
+      "step": 7710
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00021000543838814854,
+      "loss": 0.9974,
+      "step": 7711
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00020996352167724545,
+      "loss": 0.997,
+      "step": 7712
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002099216045275985,
+      "loss": 1.0143,
+      "step": 7713
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00020987968694105352,
+      "loss": 1.0489,
+      "step": 7714
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00020983776891945644,
+      "loss": 1.0963,
+      "step": 7715
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00020979585046465316,
+      "loss": 0.9493,
+      "step": 7716
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00020975393157848937,
+      "loss": 1.0994,
+      "step": 7717
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00020971201226281116,
+      "loss": 0.9922,
+      "step": 7718
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00020967009251946436,
+      "loss": 1.039,
+      "step": 7719
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00020962817235029485,
+      "loss": 0.9907,
+      "step": 7720
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00020958625175714873,
+      "loss": 0.9712,
+      "step": 7721
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00020954433074187186,
+      "loss": 1.0871,
+      "step": 7722
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00020950240930631023,
+      "loss": 0.9738,
+      "step": 7723
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00020946048745230998,
+      "loss": 1.0276,
+      "step": 7724
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.000209418565181717,
+      "loss": 0.9879,
+      "step": 7725
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00020937664249637736,
+      "loss": 1.0693,
+      "step": 7726
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002093347193981372,
+      "loss": 1.0715,
+      "step": 7727
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00020929279588884252,
+      "loss": 1.0097,
+      "step": 7728
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00020925087197033948,
+      "loss": 1.036,
+      "step": 7729
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00020920894764447413,
+      "loss": 1.0217,
+      "step": 7730
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00020916702291309269,
+      "loss": 0.9698,
+      "step": 7731
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00020912509777804133,
+      "loss": 0.9815,
+      "step": 7732
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002090831722411661,
+      "loss": 1.0319,
+      "step": 7733
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00020904124630431324,
+      "loss": 1.0158,
+      "step": 7734
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002089993199693291,
+      "loss": 1.0653,
+      "step": 7735
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002089573932380597,
+      "loss": 1.0603,
+      "step": 7736
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00020891546611235144,
+      "loss": 0.9724,
+      "step": 7737
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00020887353859405053,
+      "loss": 1.0265,
+      "step": 7738
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002088316106850032,
+      "loss": 1.066,
+      "step": 7739
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00020878968238705574,
+      "loss": 0.9954,
+      "step": 7740
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00020874775370205458,
+      "loss": 0.9954,
+      "step": 7741
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00020870582463184588,
+      "loss": 1.0211,
+      "step": 7742
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00020866389517827622,
+      "loss": 0.9729,
+      "step": 7743
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00020862196534319178,
+      "loss": 1.0398,
+      "step": 7744
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00020858003512843896,
+      "loss": 1.0019,
+      "step": 7745
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00020853810453586423,
+      "loss": 1.0906,
+      "step": 7746
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002084961735673139,
+      "loss": 1.0314,
+      "step": 7747
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00020845424222463445,
+      "loss": 1.006,
+      "step": 7748
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002084123105096724,
+      "loss": 1.12,
+      "step": 7749
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00020837037842427413,
+      "loss": 1.0546,
+      "step": 7750
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00020832844597028615,
+      "loss": 1.0519,
+      "step": 7751
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00020828651314955496,
+      "loss": 1.033,
+      "step": 7752
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.000208244579963927,
+      "loss": 1.0653,
+      "step": 7753
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00020820264641524888,
+      "loss": 1.082,
+      "step": 7754
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00020816071250536714,
+      "loss": 1.0856,
+      "step": 7755
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00020811877823612837,
+      "loss": 1.0688,
+      "step": 7756
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002080768436093791,
+      "loss": 1.0862,
+      "step": 7757
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00020803490862696586,
+      "loss": 1.0052,
+      "step": 7758
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002079929732907353,
+      "loss": 1.0685,
+      "step": 7759
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00020795103760253413,
+      "loss": 0.9899,
+      "step": 7760
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00020790910156420886,
+      "loss": 1.0871,
+      "step": 7761
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00020786716517760625,
+      "loss": 1.0087,
+      "step": 7762
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00020782522844457293,
+      "loss": 1.0363,
+      "step": 7763
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00020778329136695556,
+      "loss": 0.9954,
+      "step": 7764
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00020774135394660084,
+      "loss": 1.037,
+      "step": 7765
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002076994161853555,
+      "loss": 1.0225,
+      "step": 7766
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00020765747808506628,
+      "loss": 1.0243,
+      "step": 7767
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00020761553964757993,
+      "loss": 0.9353,
+      "step": 7768
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00020757360087474323,
+      "loss": 0.9997,
+      "step": 7769
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00020753166176840288,
+      "loss": 0.9927,
+      "step": 7770
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002074897223304057,
+      "loss": 1.0384,
+      "step": 7771
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00020744778256259852,
+      "loss": 0.9584,
+      "step": 7772
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00020740584246682815,
+      "loss": 1.0706,
+      "step": 7773
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00020736390204494141,
+      "loss": 0.966,
+      "step": 7774
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00020732196129878515,
+      "loss": 0.9505,
+      "step": 7775
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00020728002023020622,
+      "loss": 1.119,
+      "step": 7776
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00020723807884105155,
+      "loss": 0.9991,
+      "step": 7777
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00020719613713316798,
+      "loss": 1.0303,
+      "step": 7778
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00020715419510840234,
+      "loss": 1.0129,
+      "step": 7779
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00020711225276860172,
+      "loss": 1.0615,
+      "step": 7780
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00020707031011561297,
+      "loss": 1.0108,
+      "step": 7781
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00020702836715128298,
+      "loss": 0.9936,
+      "step": 7782
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00020698642387745878,
+      "loss": 1.075,
+      "step": 7783
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00020694448029598728,
+      "loss": 1.0161,
+      "step": 7784
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00020690253640871556,
+      "loss": 1.0243,
+      "step": 7785
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00020686059221749047,
+      "loss": 0.9589,
+      "step": 7786
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002068186477241592,
+      "loss": 0.9353,
+      "step": 7787
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00020677670293056872,
+      "loss": 0.9229,
+      "step": 7788
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00020673475783856597,
+      "loss": 1.0143,
+      "step": 7789
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00020669281244999806,
+      "loss": 0.9307,
+      "step": 7790
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00020665086676671215,
+      "loss": 0.9333,
+      "step": 7791
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00020660892079055518,
+      "loss": 1.0726,
+      "step": 7792
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00020656697452337434,
+      "loss": 1.0435,
+      "step": 7793
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002065250279670167,
+      "loss": 1.0222,
+      "step": 7794
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00020648308112332934,
+      "loss": 1.0199,
+      "step": 7795
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00020644113399415943,
+      "loss": 0.9894,
+      "step": 7796
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00020639918658135406,
+      "loss": 1.0652,
+      "step": 7797
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00020635723888676047,
+      "loss": 1.0369,
+      "step": 7798
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002063152909122258,
+      "loss": 1.0895,
+      "step": 7799
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002062733426595972,
+      "loss": 1.0293,
+      "step": 7800
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002062313941307219,
+      "loss": 1.0063,
+      "step": 7801
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00020618944532744705,
+      "loss": 1.0622,
+      "step": 7802
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00020614749625161993,
+      "loss": 1.1054,
+      "step": 7803
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002061055469050877,
+      "loss": 1.0433,
+      "step": 7804
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002060635972896977,
+      "loss": 1.121,
+      "step": 7805
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00020602164740729707,
+      "loss": 1.008,
+      "step": 7806
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00020597969725973312,
+      "loss": 1.0203,
+      "step": 7807
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002059377468488531,
+      "loss": 1.0143,
+      "step": 7808
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00020589579617650436,
+      "loss": 0.9968,
+      "step": 7809
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002058538452445342,
+      "loss": 1.0415,
+      "step": 7810
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00020581189405478984,
+      "loss": 1.0988,
+      "step": 7811
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00020576994260911868,
+      "loss": 1.0144,
+      "step": 7812
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00020572799090936802,
+      "loss": 1.1311,
+      "step": 7813
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00020568603895738518,
+      "loss": 1.0436,
+      "step": 7814
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00020564408675501756,
+      "loss": 1.0537,
+      "step": 7815
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00020560213430411252,
+      "loss": 0.9859,
+      "step": 7816
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002055601816065174,
+      "loss": 1.037,
+      "step": 7817
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00020551822866407963,
+      "loss": 1.0559,
+      "step": 7818
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00020547627547864666,
+      "loss": 1.0106,
+      "step": 7819
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00020543432205206573,
+      "loss": 0.9734,
+      "step": 7820
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002053923683861844,
+      "loss": 1.0087,
+      "step": 7821
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00020535041448285006,
+      "loss": 0.9878,
+      "step": 7822
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00020530846034391012,
+      "loss": 1.0163,
+      "step": 7823
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00020526650597121212,
+      "loss": 1.0402,
+      "step": 7824
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00020522455136660347,
+      "loss": 1.0318,
+      "step": 7825
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00020518259653193156,
+      "loss": 0.9732,
+      "step": 7826
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00020514064146904404,
+      "loss": 0.9632,
+      "step": 7827
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00020509868617978825,
+      "loss": 1.0833,
+      "step": 7828
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002050567306660117,
+      "loss": 1.0621,
+      "step": 7829
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00020501477492956206,
+      "loss": 1.0078,
+      "step": 7830
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00020497281897228673,
+      "loss": 1.0392,
+      "step": 7831
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00020493086279603316,
+      "loss": 1.0689,
+      "step": 7832
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00020488890640264908,
+      "loss": 1.0431,
+      "step": 7833
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00020484694979398192,
+      "loss": 1.011,
+      "step": 7834
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002048049929718792,
+      "loss": 0.9497,
+      "step": 7835
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00020476303593818863,
+      "loss": 0.9814,
+      "step": 7836
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00020472107869475776,
+      "loss": 0.9683,
+      "step": 7837
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00020467912124343406,
+      "loss": 1.012,
+      "step": 7838
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00020463716358606518,
+      "loss": 1.0026,
+      "step": 7839
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00020459520572449879,
+      "loss": 1.0304,
+      "step": 7840
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00020455324766058245,
+      "loss": 1.0797,
+      "step": 7841
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00020451128939616377,
+      "loss": 1.0321,
+      "step": 7842
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00020446933093309043,
+      "loss": 0.9993,
+      "step": 7843
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00020442737227321006,
+      "loss": 1.0173,
+      "step": 7844
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00020438541341837025,
+      "loss": 0.9646,
+      "step": 7845
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002043434543704187,
+      "loss": 1.0451,
+      "step": 7846
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00020430149513120315,
+      "loss": 1.087,
+      "step": 7847
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00020425953570257113,
+      "loss": 1.0057,
+      "step": 7848
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00020421757608637045,
+      "loss": 1.0483,
+      "step": 7849
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00020417561628444876,
+      "loss": 1.0408,
+      "step": 7850
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00020413365629865372,
+      "loss": 1.0597,
+      "step": 7851
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00020409169613083315,
+      "loss": 1.0509,
+      "step": 7852
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002040497357828346,
+      "loss": 0.9952,
+      "step": 7853
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00020400777525650588,
+      "loss": 1.0369,
+      "step": 7854
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002039658145536948,
+      "loss": 1.051,
+      "step": 7855
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00020392385367624903,
+      "loss": 1.0139,
+      "step": 7856
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00020388189262601626,
+      "loss": 1.0201,
+      "step": 7857
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00020383993140484435,
+      "loss": 1.0517,
+      "step": 7858
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00020379797001458095,
+      "loss": 1.0211,
+      "step": 7859
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002037560084570739,
+      "loss": 1.1123,
+      "step": 7860
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00020371404673417103,
+      "loss": 0.9697,
+      "step": 7861
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00020367208484772003,
+      "loss": 1.0573,
+      "step": 7862
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00020363012279956877,
+      "loss": 0.9779,
+      "step": 7863
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002035881605915649,
+      "loss": 1.0304,
+      "step": 7864
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00020354619822555642,
+      "loss": 1.0224,
+      "step": 7865
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00020350423570339104,
+      "loss": 1.0185,
+      "step": 7866
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00020346227302691657,
+      "loss": 1.0354,
+      "step": 7867
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00020342031019798094,
+      "loss": 1.0728,
+      "step": 7868
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00020337834721843187,
+      "loss": 1.0204,
+      "step": 7869
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002033363840901172,
+      "loss": 1.043,
+      "step": 7870
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002032944208148848,
+      "loss": 1.0741,
+      "step": 7871
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00020325245739458262,
+      "loss": 0.9556,
+      "step": 7872
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002032104938310584,
+      "loss": 0.9931,
+      "step": 7873
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00020316853012616003,
+      "loss": 1.0397,
+      "step": 7874
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00020312656628173543,
+      "loss": 1.0125,
+      "step": 7875
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002030846022996324,
+      "loss": 0.9763,
+      "step": 7876
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002030426381816989,
+      "loss": 0.97,
+      "step": 7877
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00020300067392978278,
+      "loss": 1.0983,
+      "step": 7878
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00020295870954573192,
+      "loss": 0.9794,
+      "step": 7879
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00020291674503139433,
+      "loss": 1.0177,
+      "step": 7880
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002028747803886178,
+      "loss": 0.9917,
+      "step": 7881
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00020283281561925024,
+      "loss": 1.031,
+      "step": 7882
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00020279085072513967,
+      "loss": 1.0076,
+      "step": 7883
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002027488857081339,
+      "loss": 1.045,
+      "step": 7884
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002027069205700809,
+      "loss": 1.0036,
+      "step": 7885
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00020266495531282862,
+      "loss": 0.9563,
+      "step": 7886
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00020262298993822507,
+      "loss": 1.056,
+      "step": 7887
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00020258102444811805,
+      "loss": 0.9589,
+      "step": 7888
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00020253905884435564,
+      "loss": 1.0025,
+      "step": 7889
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00020249709312878574,
+      "loss": 1.024,
+      "step": 7890
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002024551273032563,
+      "loss": 1.0453,
+      "step": 7891
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002024131613696153,
+      "loss": 1.0216,
+      "step": 7892
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00020237119532971073,
+      "loss": 0.9976,
+      "step": 7893
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00020232922918539053,
+      "loss": 0.9866,
+      "step": 7894
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00020228726293850265,
+      "loss": 1.0532,
+      "step": 7895
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002022452965908951,
+      "loss": 1.0107,
+      "step": 7896
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00020220333014441598,
+      "loss": 1.0906,
+      "step": 7897
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00020216136360091312,
+      "loss": 0.9546,
+      "step": 7898
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00020211939696223463,
+      "loss": 1.0602,
+      "step": 7899
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002020774302302285,
+      "loss": 1.0083,
+      "step": 7900
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00020203546340674265,
+      "loss": 1.0127,
+      "step": 7901
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002019934964936251,
+      "loss": 1.0474,
+      "step": 7902
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.000201951529492724,
+      "loss": 1.0571,
+      "step": 7903
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002019095624058872,
+      "loss": 1.1068,
+      "step": 7904
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00020186759523496285,
+      "loss": 1.0128,
+      "step": 7905
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002018256279817989,
+      "loss": 1.0524,
+      "step": 7906
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00020178366064824336,
+      "loss": 0.9967,
+      "step": 7907
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002017416932361444,
+      "loss": 1.0509,
+      "step": 7908
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00020169972574734987,
+      "loss": 1.0839,
+      "step": 7909
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002016577581837079,
+      "loss": 0.9967,
+      "step": 7910
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00020161579054706654,
+      "loss": 1.0116,
+      "step": 7911
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00020157382283927388,
+      "loss": 1.049,
+      "step": 7912
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00020153185506217782,
+      "loss": 1.0294,
+      "step": 7913
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002014898872176266,
+      "loss": 0.9573,
+      "step": 7914
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002014479193074681,
+      "loss": 0.9936,
+      "step": 7915
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00020140595133355044,
+      "loss": 1.0037,
+      "step": 7916
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00020136398329772176,
+      "loss": 0.928,
+      "step": 7917
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00020132201520183006,
+      "loss": 1.0599,
+      "step": 7918
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002012800470477234,
+      "loss": 1.0099,
+      "step": 7919
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00020123807883724984,
+      "loss": 0.9393,
+      "step": 7920
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00020119611057225743,
+      "loss": 1.0656,
+      "step": 7921
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002011541422545943,
+      "loss": 1.012,
+      "step": 7922
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00020111217388610846,
+      "loss": 0.985,
+      "step": 7923
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00020107020546864806,
+      "loss": 1.1121,
+      "step": 7924
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00020102823700406122,
+      "loss": 1.0299,
+      "step": 7925
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00020098626849419587,
+      "loss": 0.9643,
+      "step": 7926
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00020094429994090018,
+      "loss": 0.9838,
+      "step": 7927
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00020090233134602227,
+      "loss": 1.0736,
+      "step": 7928
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00020086036271141017,
+      "loss": 1.0462,
+      "step": 7929
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00020081839403891197,
+      "loss": 1.0509,
+      "step": 7930
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002007764253303758,
+      "loss": 1.0202,
+      "step": 7931
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002007344565876497,
+      "loss": 1.0613,
+      "step": 7932
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00020069248781258184,
+      "loss": 1.0375,
+      "step": 7933
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00020065051900702025,
+      "loss": 0.9733,
+      "step": 7934
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.000200608550172813,
+      "loss": 1.0644,
+      "step": 7935
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00020056658131180832,
+      "loss": 1.0238,
+      "step": 7936
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002005246124258542,
+      "loss": 1.0352,
+      "step": 7937
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002004826435167987,
+      "loss": 1.0328,
+      "step": 7938
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002004406745864901,
+      "loss": 1.0164,
+      "step": 7939
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002003987056367763,
+      "loss": 1.0443,
+      "step": 7940
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00020035673666950548,
+      "loss": 1.018,
+      "step": 7941
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002003147676865258,
+      "loss": 1.0748,
+      "step": 7942
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00020027279868968528,
+      "loss": 1.053,
+      "step": 7943
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00020023082968083217,
+      "loss": 0.9936,
+      "step": 7944
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002001888606618144,
+      "loss": 0.9999,
+      "step": 7945
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00020014689163448015,
+      "loss": 0.9559,
+      "step": 7946
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00020010492260067752,
+      "loss": 0.9254,
+      "step": 7947
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00020006295356225459,
+      "loss": 1.0778,
+      "step": 7948
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00020002098452105954,
+      "loss": 1.0421,
+      "step": 7949
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001999790154789405,
+      "loss": 1.054,
+      "step": 7950
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00019993704643774546,
+      "loss": 1.0661,
+      "step": 7951
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00019989507739932253,
+      "loss": 0.9724,
+      "step": 7952
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00019985310836551993,
+      "loss": 0.9865,
+      "step": 7953
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001998111393381857,
+      "loss": 1.0916,
+      "step": 7954
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00019976917031916784,
+      "loss": 0.9873,
+      "step": 7955
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00019972720131031474,
+      "loss": 1.0319,
+      "step": 7956
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00019968523231347427,
+      "loss": 1.005,
+      "step": 7957
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00019964326333049453,
+      "loss": 1.0027,
+      "step": 7958
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00019960129436322378,
+      "loss": 0.9915,
+      "step": 7959
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00019955932541351,
+      "loss": 1.0172,
+      "step": 7960
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001995173564832013,
+      "loss": 1.045,
+      "step": 7961
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00019947538757414586,
+      "loss": 0.9248,
+      "step": 7962
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00019943341868819178,
+      "loss": 1.0091,
+      "step": 7963
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.000199391449827187,
+      "loss": 1.065,
+      "step": 7964
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001993494809929798,
+      "loss": 0.9887,
+      "step": 7965
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00019930751218741823,
+      "loss": 1.0687,
+      "step": 7966
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00019926554341235028,
+      "loss": 0.9442,
+      "step": 7967
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00019922357466962422,
+      "loss": 1.0383,
+      "step": 7968
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001991816059610881,
+      "loss": 1.0202,
+      "step": 7969
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00019913963728858985,
+      "loss": 1.0055,
+      "step": 7970
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00019909766865397778,
+      "loss": 1.0638,
+      "step": 7971
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00019905570005909987,
+      "loss": 0.9534,
+      "step": 7972
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00019901373150580412,
+      "loss": 1.0315,
+      "step": 7973
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001989717629959388,
+      "loss": 1.0145,
+      "step": 7974
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00019892979453135199,
+      "loss": 1.0677,
+      "step": 7975
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001988878261138915,
+      "loss": 1.0837,
+      "step": 7976
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00019884585774540575,
+      "loss": 0.9992,
+      "step": 7977
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00019880388942774264,
+      "loss": 1.0428,
+      "step": 7978
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001987619211627503,
+      "loss": 1.0569,
+      "step": 7979
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00019871995295227664,
+      "loss": 1.0406,
+      "step": 7980
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00019867798479816998,
+      "loss": 1.037,
+      "step": 7981
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00019863601670227831,
+      "loss": 0.9896,
+      "step": 7982
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00019859404866644955,
+      "loss": 0.9583,
+      "step": 7983
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00019855208069253195,
+      "loss": 1.0804,
+      "step": 7984
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001985101127823735,
+      "loss": 1.0356,
+      "step": 7985
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00019846814493782215,
+      "loss": 1.011,
+      "step": 7986
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00019842617716072617,
+      "loss": 1.0154,
+      "step": 7987
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00019838420945293353,
+      "loss": 1.0007,
+      "step": 7988
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001983422418162921,
+      "loss": 1.0255,
+      "step": 7989
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00019830027425265018,
+      "loss": 1.0682,
+      "step": 7990
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001982583067638557,
+      "loss": 1.0649,
+      "step": 7991
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001982163393517566,
+      "loss": 0.9876,
+      "step": 7992
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00019817437201820112,
+      "loss": 1.0144,
+      "step": 7993
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001981324047650372,
+      "loss": 1.0862,
+      "step": 7994
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001980904375941128,
+      "loss": 0.9467,
+      "step": 7995
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00019804847050727606,
+      "loss": 1.0031,
+      "step": 7996
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00019800650350637494,
+      "loss": 1.0656,
+      "step": 7997
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00019796453659325737,
+      "loss": 0.9672,
+      "step": 7998
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00019792256976977154,
+      "loss": 1.0746,
+      "step": 7999
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00019788060303776542,
+      "loss": 1.0196,
+      "step": 8000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00019783863639908687,
+      "loss": 1.1096,
+      "step": 8001
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00019779666985558406,
+      "loss": 1.0479,
+      "step": 8002
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001977547034091049,
+      "loss": 1.0429,
+      "step": 8003
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00019771273706149745,
+      "loss": 0.9864,
+      "step": 8004
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00019767077081460954,
+      "loss": 0.964,
+      "step": 8005
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00019762880467028934,
+      "loss": 1.0417,
+      "step": 8006
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001975868386303848,
+      "loss": 1.0432,
+      "step": 8007
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00019754487269674375,
+      "loss": 1.0491,
+      "step": 8008
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001975029068712143,
+      "loss": 1.0667,
+      "step": 8009
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001974609411556444,
+      "loss": 1.1217,
+      "step": 8010
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00019741897555188194,
+      "loss": 1.0296,
+      "step": 8011
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00019737701006177497,
+      "loss": 1.1122,
+      "step": 8012
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001973350446871714,
+      "loss": 1.0336,
+      "step": 8013
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00019729307942991913,
+      "loss": 1.0567,
+      "step": 8014
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00019725111429186619,
+      "loss": 1.0512,
+      "step": 8015
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00019720914927486046,
+      "loss": 1.1017,
+      "step": 8016
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00019716718438074978,
+      "loss": 1.0537,
+      "step": 8017
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00019712521961138226,
+      "loss": 1.016,
+      "step": 8018
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00019708325496860574,
+      "loss": 1.0303,
+      "step": 8019
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00019704129045426807,
+      "loss": 1.0375,
+      "step": 8020
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00019699932607021727,
+      "loss": 0.9362,
+      "step": 8021
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00019695736181830117,
+      "loss": 1.019,
+      "step": 8022
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001969153977003676,
+      "loss": 1.055,
+      "step": 8023
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00019687343371826461,
+      "loss": 0.987,
+      "step": 8024
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00019683146987384,
+      "loss": 1.0615,
+      "step": 8025
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00019678950616894162,
+      "loss": 1.0966,
+      "step": 8026
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00019674754260541743,
+      "loss": 1.016,
+      "step": 8027
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00019670557918511524,
+      "loss": 0.9623,
+      "step": 8028
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001966636159098828,
+      "loss": 1.0239,
+      "step": 8029
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00019662165278156818,
+      "loss": 1.0455,
+      "step": 8030
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00019657968980201914,
+      "loss": 1.0371,
+      "step": 8031
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00019653772697308345,
+      "loss": 0.9961,
+      "step": 8032
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00019649576429660897,
+      "loss": 1.0049,
+      "step": 8033
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00019645380177444363,
+      "loss": 0.9764,
+      "step": 8034
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00019641183940843516,
+      "loss": 1.0359,
+      "step": 8035
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001963698772004313,
+      "loss": 0.9946,
+      "step": 8036
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00019632791515228,
+      "loss": 0.968,
+      "step": 8037
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00019628595326582905,
+      "loss": 0.9643,
+      "step": 8038
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001962439915429261,
+      "loss": 0.9657,
+      "step": 8039
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001962020299854191,
+      "loss": 0.9549,
+      "step": 8040
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00019616006859515578,
+      "loss": 0.9164,
+      "step": 8041
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00019611810737398376,
+      "loss": 1.0168,
+      "step": 8042
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00019607614632375102,
+      "loss": 1.0811,
+      "step": 8043
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00019603418544630522,
+      "loss": 1.0187,
+      "step": 8044
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001959922247434941,
+      "loss": 1.123,
+      "step": 8045
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00019595026421716545,
+      "loss": 1.0051,
+      "step": 8046
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00019590830386916695,
+      "loss": 0.9914,
+      "step": 8047
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00019586634370134627,
+      "loss": 1.0522,
+      "step": 8048
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00019582438371555126,
+      "loss": 1.0394,
+      "step": 8049
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00019578242391362957,
+      "loss": 1.0606,
+      "step": 8050
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00019574046429742887,
+      "loss": 1.0365,
+      "step": 8051
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001956985048687969,
+      "loss": 1.0922,
+      "step": 8052
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00019565654562958134,
+      "loss": 1.0429,
+      "step": 8053
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00019561458658162975,
+      "loss": 1.0542,
+      "step": 8054
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00019557262772678999,
+      "loss": 1.0681,
+      "step": 8055
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00019553066906690962,
+      "loss": 0.9955,
+      "step": 8056
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00019548871060383628,
+      "loss": 0.9921,
+      "step": 8057
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001954467523394176,
+      "loss": 0.9869,
+      "step": 8058
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00019540479427550126,
+      "loss": 1.0238,
+      "step": 8059
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001953628364139349,
+      "loss": 1.0665,
+      "step": 8060
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00019532087875656596,
+      "loss": 0.9299,
+      "step": 8061
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001952789213052423,
+      "loss": 1.048,
+      "step": 8062
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001952369640618114,
+      "loss": 1.0398,
+      "step": 8063
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00019519500702812077,
+      "loss": 1.0503,
+      "step": 8064
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00019515305020601813,
+      "loss": 1.0323,
+      "step": 8065
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.000195111093597351,
+      "loss": 1.043,
+      "step": 8066
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00019506913720396683,
+      "loss": 1.0093,
+      "step": 8067
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00019502718102771334,
+      "loss": 1.1204,
+      "step": 8068
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00019498522507043799,
+      "loss": 1.0465,
+      "step": 8069
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00019494326933398827,
+      "loss": 1.0252,
+      "step": 8070
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00019490131382021182,
+      "loss": 1.0218,
+      "step": 8071
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00019485935853095606,
+      "loss": 1.0754,
+      "step": 8072
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00019481740346806843,
+      "loss": 0.9841,
+      "step": 8073
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00019477544863339658,
+      "loss": 0.9102,
+      "step": 8074
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001947334940287879,
+      "loss": 1.0923,
+      "step": 8075
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00019469153965608987,
+      "loss": 0.9396,
+      "step": 8076
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00019464958551714998,
+      "loss": 1.0294,
+      "step": 8077
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00019460763161381568,
+      "loss": 0.9586,
+      "step": 8078
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001945656779479343,
+      "loss": 1.0044,
+      "step": 8079
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001945237245213534,
+      "loss": 1.0386,
+      "step": 8080
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00019448177133592038,
+      "loss": 0.9899,
+      "step": 8081
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001944398183934826,
+      "loss": 1.0185,
+      "step": 8082
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00019439786569588752,
+      "loss": 1.0825,
+      "step": 8083
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001943559132449825,
+      "loss": 0.9844,
+      "step": 8084
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001943139610426149,
+      "loss": 1.1157,
+      "step": 8085
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.000194272009090632,
+      "loss": 1.0723,
+      "step": 8086
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00019423005739088137,
+      "loss": 1.0245,
+      "step": 8087
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001941881059452102,
+      "loss": 1.0284,
+      "step": 8088
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00019414615475546584,
+      "loss": 0.9581,
+      "step": 8089
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001941042038234957,
+      "loss": 1.038,
+      "step": 8090
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00019406225315114694,
+      "loss": 0.9802,
+      "step": 8091
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001940203027402669,
+      "loss": 1.0451,
+      "step": 8092
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00019397835259270298,
+      "loss": 0.9992,
+      "step": 8093
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00019393640271030238,
+      "loss": 0.9319,
+      "step": 8094
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001938944530949123,
+      "loss": 1.0036,
+      "step": 8095
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001938525037483801,
+      "loss": 1.0899,
+      "step": 8096
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.000193810554672553,
+      "loss": 0.9663,
+      "step": 8097
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001937686058692781,
+      "loss": 1.0534,
+      "step": 8098
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001937266573404028,
+      "loss": 1.0193,
+      "step": 8099
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00019368470908777424,
+      "loss": 0.9757,
+      "step": 8100
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00019364276111323955,
+      "loss": 0.9384,
+      "step": 8101
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.000193600813418646,
+      "loss": 0.9993,
+      "step": 8102
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00019355886600584067,
+      "loss": 1.0244,
+      "step": 8103
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00019351691887667068,
+      "loss": 1.0698,
+      "step": 8104
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00019347497203298336,
+      "loss": 1.0717,
+      "step": 8105
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001934330254766257,
+      "loss": 1.0724,
+      "step": 8106
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00019339107920944484,
+      "loss": 0.9706,
+      "step": 8107
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001933491332332879,
+      "loss": 1.0675,
+      "step": 8108
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00019330718755000196,
+      "loss": 0.963,
+      "step": 8109
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00019326524216143405,
+      "loss": 1.0244,
+      "step": 8110
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00019322329706943135,
+      "loss": 1.0496,
+      "step": 8111
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00019318135227584083,
+      "loss": 0.951,
+      "step": 8112
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00019313940778250955,
+      "loss": 1.0703,
+      "step": 8113
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001930974635912845,
+      "loss": 1.0426,
+      "step": 8114
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00019305551970401276,
+      "loss": 1.0567,
+      "step": 8115
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00019301357612254132,
+      "loss": 0.9755,
+      "step": 8116
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00019297163284871704,
+      "loss": 1.0222,
+      "step": 8117
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00019292968988438708,
+      "loss": 0.9346,
+      "step": 8118
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001928877472313983,
+      "loss": 0.9949,
+      "step": 8119
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00019284580489159765,
+      "loss": 1.0546,
+      "step": 8120
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001928038628668321,
+      "loss": 1.1113,
+      "step": 8121
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00019276192115894852,
+      "loss": 1.1011,
+      "step": 8122
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00019271997976979378,
+      "loss": 1.1147,
+      "step": 8123
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00019267803870121487,
+      "loss": 1.0345,
+      "step": 8124
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00019263609795505863,
+      "loss": 0.9647,
+      "step": 8125
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00019259415753317187,
+      "loss": 1.0853,
+      "step": 8126
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001925522174374015,
+      "loss": 1.0093,
+      "step": 8127
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00019251027766959436,
+      "loss": 0.9909,
+      "step": 8128
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00019246833823159714,
+      "loss": 0.9584,
+      "step": 8129
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001924263991252568,
+      "loss": 0.9728,
+      "step": 8130
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001923844603524201,
+      "loss": 1.0182,
+      "step": 8131
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001923425219149337,
+      "loss": 1.0137,
+      "step": 8132
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00019230058381464454,
+      "loss": 1.0197,
+      "step": 8133
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00019225864605339923,
+      "loss": 1.0714,
+      "step": 8134
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00019221670863304446,
+      "loss": 1.0884,
+      "step": 8135
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001921747715554271,
+      "loss": 1.0419,
+      "step": 8136
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001921328348223938,
+      "loss": 0.9528,
+      "step": 8137
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001920908984357912,
+      "loss": 1.0043,
+      "step": 8138
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00019204896239746592,
+      "loss": 1.0442,
+      "step": 8139
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00019200702670926474,
+      "loss": 0.9623,
+      "step": 8140
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00019196509137303424,
+      "loss": 1.1132,
+      "step": 8141
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00019192315639062095,
+      "loss": 1.069,
+      "step": 8142
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00019188122176387167,
+      "loss": 1.0758,
+      "step": 8143
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00019183928749463288,
+      "loss": 0.9245,
+      "step": 8144
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001917973535847511,
+      "loss": 1.0521,
+      "step": 8145
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00019175542003607303,
+      "loss": 1.0201,
+      "step": 8146
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00019171348685044514,
+      "loss": 1.0494,
+      "step": 8147
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00019167155402971387,
+      "loss": 0.9811,
+      "step": 8148
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001916296215757259,
+      "loss": 0.9951,
+      "step": 8149
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00019158768949032766,
+      "loss": 0.9917,
+      "step": 8150
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00019154575777536556,
+      "loss": 0.941,
+      "step": 8151
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00019150382643268616,
+      "loss": 1.0384,
+      "step": 8152
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001914618954641359,
+      "loss": 1.0264,
+      "step": 8153
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00019141996487156106,
+      "loss": 1.0684,
+      "step": 8154
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001913780346568083,
+      "loss": 0.9483,
+      "step": 8155
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00019133610482172385,
+      "loss": 1.1082,
+      "step": 8156
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00019129417536815408,
+      "loss": 1.0527,
+      "step": 8157
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00019125224629794546,
+      "loss": 0.9494,
+      "step": 8158
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001912103176129443,
+      "loss": 0.962,
+      "step": 8159
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00019116838931499682,
+      "loss": 1.027,
+      "step": 8160
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001911264614059495,
+      "loss": 0.9473,
+      "step": 8161
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001910845338876486,
+      "loss": 0.9985,
+      "step": 8162
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00019104260676194027,
+      "loss": 1.078,
+      "step": 8163
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00019100068003067094,
+      "loss": 1.0173,
+      "step": 8164
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00019095875369568678,
+      "loss": 0.9745,
+      "step": 8165
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00019091682775883398,
+      "loss": 1.0069,
+      "step": 8166
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001908749022219587,
+      "loss": 1.0381,
+      "step": 8167
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00019083297708690733,
+      "loss": 1.0682,
+      "step": 8168
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00019079105235552592,
+      "loss": 0.9755,
+      "step": 8169
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001907491280296606,
+      "loss": 1.0127,
+      "step": 8170
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00019070720411115756,
+      "loss": 1.0007,
+      "step": 8171
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00019066528060186292,
+      "loss": 0.9808,
+      "step": 8172
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00019062335750362263,
+      "loss": 1.0245,
+      "step": 8173
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00019058143481828305,
+      "loss": 0.9861,
+      "step": 8174
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001905395125476901,
+      "loss": 1.0413,
+      "step": 8175
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00019049759069368976,
+      "loss": 1.0654,
+      "step": 8176
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00019045566925812818,
+      "loss": 1.0304,
+      "step": 8177
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00019041374824285134,
+      "loss": 1.0384,
+      "step": 8178
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00019037182764970514,
+      "loss": 0.9683,
+      "step": 8179
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001903299074805357,
+      "loss": 1.039,
+      "step": 8180
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001902879877371889,
+      "loss": 0.9847,
+      "step": 8181
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00019024606842151063,
+      "loss": 0.9734,
+      "step": 8182
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001902041495353469,
+      "loss": 1.026,
+      "step": 8183
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00019016223108054358,
+      "loss": 0.989,
+      "step": 8184
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00019012031305894647,
+      "loss": 1.0712,
+      "step": 8185
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00019007839547240154,
+      "loss": 0.9736,
+      "step": 8186
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00019003647832275462,
+      "loss": 1.0542,
+      "step": 8187
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00018999456161185145,
+      "loss": 0.9337,
+      "step": 8188
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00018995264534153793,
+      "loss": 1.0213,
+      "step": 8189
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00018991072951365977,
+      "loss": 1.0053,
+      "step": 8190
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00018986881413006278,
+      "loss": 1.0935,
+      "step": 8191
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00018982689919259263,
+      "loss": 0.9852,
+      "step": 8192
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00018978498470309513,
+      "loss": 1.0027,
+      "step": 8193
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00018974307066341598,
+      "loss": 1.0275,
+      "step": 8194
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00018970115707540083,
+      "loss": 0.9348,
+      "step": 8195
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00018965924394089535,
+      "loss": 1.0453,
+      "step": 8196
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00018961733126174522,
+      "loss": 1.0599,
+      "step": 8197
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00018957541903979593,
+      "loss": 0.9786,
+      "step": 8198
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001895335072768933,
+      "loss": 1.0333,
+      "step": 8199
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001894915959748828,
+      "loss": 1.0365,
+      "step": 8200
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00018944968513560996,
+      "loss": 1.0128,
+      "step": 8201
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001894077747609204,
+      "loss": 0.9694,
+      "step": 8202
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001893658648526596,
+      "loss": 1.0146,
+      "step": 8203
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.000189323955412673,
+      "loss": 1.0407,
+      "step": 8204
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00018928204644280626,
+      "loss": 1.0306,
+      "step": 8205
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001892401379449047,
+      "loss": 1.0017,
+      "step": 8206
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001891982299208138,
+      "loss": 0.9583,
+      "step": 8207
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00018915632237237898,
+      "loss": 1.0835,
+      "step": 8208
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00018911441530144568,
+      "loss": 0.9961,
+      "step": 8209
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00018907250870985913,
+      "loss": 1.0543,
+      "step": 8210
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00018903060259946487,
+      "loss": 0.9305,
+      "step": 8211
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00018898869697210816,
+      "loss": 1.0811,
+      "step": 8212
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001889467918296343,
+      "loss": 1.0133,
+      "step": 8213
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00018890488717388863,
+      "loss": 1.1219,
+      "step": 8214
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001888629830067164,
+      "loss": 1.034,
+      "step": 8215
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00018882107932996272,
+      "loss": 1.002,
+      "step": 8216
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001887791761454731,
+      "loss": 1.039,
+      "step": 8217
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00018873727345509255,
+      "loss": 1.0744,
+      "step": 8218
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00018869537126066632,
+      "loss": 0.9624,
+      "step": 8219
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001886534695640395,
+      "loss": 0.9959,
+      "step": 8220
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00018861156836705734,
+      "loss": 1.0785,
+      "step": 8221
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00018856966767156493,
+      "loss": 0.9927,
+      "step": 8222
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00018852776747940725,
+      "loss": 1.0645,
+      "step": 8223
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00018848586779242953,
+      "loss": 1.0507,
+      "step": 8224
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001884439686124768,
+      "loss": 1.0522,
+      "step": 8225
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.000188402069941394,
+      "loss": 1.0113,
+      "step": 8226
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00018836017178102622,
+      "loss": 0.9648,
+      "step": 8227
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00018831827413321844,
+      "loss": 1.1383,
+      "step": 8228
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00018827637699981553,
+      "loss": 1.0608,
+      "step": 8229
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00018823448038266256,
+      "loss": 1.0261,
+      "step": 8230
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001881925842836044,
+      "loss": 1.0248,
+      "step": 8231
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001881506887044859,
+      "loss": 1.0609,
+      "step": 8232
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00018810879364715202,
+      "loss": 1.0319,
+      "step": 8233
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00018806689911344754,
+      "loss": 0.9724,
+      "step": 8234
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00018802500510521722,
+      "loss": 1.0156,
+      "step": 8235
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00018798311162430606,
+      "loss": 0.9888,
+      "step": 8236
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00018794121867255872,
+      "loss": 1.0237,
+      "step": 8237
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001878993262518199,
+      "loss": 1.1029,
+      "step": 8238
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00018785743436393448,
+      "loss": 1.0225,
+      "step": 8239
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00018781554301074707,
+      "loss": 1.0524,
+      "step": 8240
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001877736521941023,
+      "loss": 1.0236,
+      "step": 8241
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00018773176191584496,
+      "loss": 1.0033,
+      "step": 8242
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00018768987217781966,
+      "loss": 1.0021,
+      "step": 8243
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00018764798298187094,
+      "loss": 0.9186,
+      "step": 8244
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001876060943298435,
+      "loss": 1.0377,
+      "step": 8245
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00018756420622358184,
+      "loss": 1.0287,
+      "step": 8246
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00018752231866493051,
+      "loss": 1.0133,
+      "step": 8247
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00018748043165573396,
+      "loss": 0.9885,
+      "step": 8248
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00018743854519783683,
+      "loss": 0.9434,
+      "step": 8249
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00018739665929308353,
+      "loss": 1.0609,
+      "step": 8250
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001873547739433184,
+      "loss": 1.0836,
+      "step": 8251
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00018731288915038608,
+      "loss": 1.0835,
+      "step": 8252
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001872710049161308,
+      "loss": 1.0217,
+      "step": 8253
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001872291212423969,
+      "loss": 1.0289,
+      "step": 8254
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00018718723813102887,
+      "loss": 1.0287,
+      "step": 8255
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.000187145355583871,
+      "loss": 1.0811,
+      "step": 8256
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001871034736027675,
+      "loss": 1.0214,
+      "step": 8257
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00018706159218956274,
+      "loss": 1.0105,
+      "step": 8258
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001870197113461009,
+      "loss": 0.9856,
+      "step": 8259
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00018697783107422618,
+      "loss": 1.0407,
+      "step": 8260
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00018693595137578292,
+      "loss": 1.0217,
+      "step": 8261
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00018689407225261522,
+      "loss": 1.0581,
+      "step": 8262
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00018685219370656714,
+      "loss": 0.9661,
+      "step": 8263
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001868103157394829,
+      "loss": 0.9751,
+      "step": 8264
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001867684383532066,
+      "loss": 1.0111,
+      "step": 8265
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00018672656154958216,
+      "loss": 0.9976,
+      "step": 8266
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001866846853304539,
+      "loss": 0.9712,
+      "step": 8267
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00018664280969766563,
+      "loss": 1.0461,
+      "step": 8268
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00018660093465306136,
+      "loss": 1.0181,
+      "step": 8269
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00018655906019848518,
+      "loss": 0.9827,
+      "step": 8270
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001865171863357809,
+      "loss": 0.9449,
+      "step": 8271
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001864753130667925,
+      "loss": 0.9482,
+      "step": 8272
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00018643344039336382,
+      "loss": 1.0008,
+      "step": 8273
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001863915683173388,
+      "loss": 1.0072,
+      "step": 8274
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00018634969684056123,
+      "loss": 1.072,
+      "step": 8275
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001863078259648749,
+      "loss": 1.0364,
+      "step": 8276
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00018626595569212368,
+      "loss": 1.0245,
+      "step": 8277
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00018622408602415126,
+      "loss": 1.0521,
+      "step": 8278
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00018618221696280128,
+      "loss": 0.9814,
+      "step": 8279
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00018614034850991768,
+      "loss": 0.9385,
+      "step": 8280
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00018609848066734397,
+      "loss": 1.0398,
+      "step": 8281
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001860566134369238,
+      "loss": 1.0173,
+      "step": 8282
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001860147468205009,
+      "loss": 1.0591,
+      "step": 8283
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00018597288081991875,
+      "loss": 0.965,
+      "step": 8284
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001859310154370209,
+      "loss": 1.0597,
+      "step": 8285
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00018588915067365105,
+      "loss": 1.0637,
+      "step": 8286
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00018584728653165262,
+      "loss": 1.0255,
+      "step": 8287
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00018580542301286908,
+      "loss": 1.0735,
+      "step": 8288
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00018576356011914394,
+      "loss": 1.0357,
+      "step": 8289
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00018572169785232057,
+      "loss": 1.0404,
+      "step": 8290
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001856798362142424,
+      "loss": 1.0113,
+      "step": 8291
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00018563797520675286,
+      "loss": 1.0407,
+      "step": 8292
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001855961148316953,
+      "loss": 1.1031,
+      "step": 8293
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00018555425509091288,
+      "loss": 1.0317,
+      "step": 8294
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00018551239598624912,
+      "loss": 1.0262,
+      "step": 8295
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00018547053751954718,
+      "loss": 1.0459,
+      "step": 8296
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001854286796926502,
+      "loss": 0.9835,
+      "step": 8297
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00018538682250740157,
+      "loss": 1.0509,
+      "step": 8298
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001853449659656444,
+      "loss": 0.9727,
+      "step": 8299
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00018530311006922186,
+      "loss": 1.0644,
+      "step": 8300
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00018526125481997698,
+      "loss": 1.0192,
+      "step": 8301
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00018521940021975299,
+      "loss": 1.0619,
+      "step": 8302
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00018517754627039287,
+      "loss": 0.9735,
+      "step": 8303
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001851356929737396,
+      "loss": 1.0319,
+      "step": 8304
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001850938403316364,
+      "loss": 1.0594,
+      "step": 8305
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00018505198834592608,
+      "loss": 1.0526,
+      "step": 8306
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001850101370184516,
+      "loss": 1.0165,
+      "step": 8307
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.000184968286351056,
+      "loss": 1.0258,
+      "step": 8308
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00018492643634558206,
+      "loss": 1.0581,
+      "step": 8309
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001848845870038726,
+      "loss": 0.9884,
+      "step": 8310
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00018484273832777067,
+      "loss": 1.0211,
+      "step": 8311
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001848008903191189,
+      "loss": 1.049,
+      "step": 8312
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00018475904297976008,
+      "loss": 1.0312,
+      "step": 8313
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00018471719631153703,
+      "loss": 1.0548,
+      "step": 8314
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00018467535031629244,
+      "loss": 1.0549,
+      "step": 8315
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001846335049958689,
+      "loss": 1.0595,
+      "step": 8316
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00018459166035210924,
+      "loss": 1.0404,
+      "step": 8317
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.000184549816386856,
+      "loss": 0.9945,
+      "step": 8318
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00018450797310195176,
+      "loss": 0.9768,
+      "step": 8319
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00018446613049923917,
+      "loss": 1.0141,
+      "step": 8320
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001844242885805607,
+      "loss": 1.0483,
+      "step": 8321
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00018438244734775877,
+      "loss": 1.015,
+      "step": 8322
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00018434060680267607,
+      "loss": 1.0311,
+      "step": 8323
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00018429876694715498,
+      "loss": 1.0294,
+      "step": 8324
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00018425692778303783,
+      "loss": 1.0532,
+      "step": 8325
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00018421508931216703,
+      "loss": 1.0458,
+      "step": 8326
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00018417325153638502,
+      "loss": 0.9975,
+      "step": 8327
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00018413141445753408,
+      "loss": 0.9992,
+      "step": 8328
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001840895780774564,
+      "loss": 0.949,
+      "step": 8329
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00018404774239799446,
+      "loss": 1.0325,
+      "step": 8330
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00018400590742099034,
+      "loss": 1.1107,
+      "step": 8331
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00018396407314828629,
+      "loss": 1.0423,
+      "step": 8332
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001839222395817245,
+      "loss": 1.0037,
+      "step": 8333
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00018388040672314706,
+      "loss": 1.0816,
+      "step": 8334
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00018383857457439603,
+      "loss": 0.9529,
+      "step": 8335
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001837967431373137,
+      "loss": 1.0274,
+      "step": 8336
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00018375491241374195,
+      "loss": 1.0456,
+      "step": 8337
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00018371308240552282,
+      "loss": 1.0392,
+      "step": 8338
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001836712531144983,
+      "loss": 1.0694,
+      "step": 8339
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001836294245425104,
+      "loss": 1.08,
+      "step": 8340
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00018358759669140088,
+      "loss": 1.0715,
+      "step": 8341
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00018354576956301183,
+      "loss": 1.0205,
+      "step": 8342
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00018350394315918504,
+      "loss": 0.989,
+      "step": 8343
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001834621174817623,
+      "loss": 1.0175,
+      "step": 8344
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00018342029253258542,
+      "loss": 0.9769,
+      "step": 8345
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00018337846831349616,
+      "loss": 1.0762,
+      "step": 8346
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001833366448263362,
+      "loss": 1.0326,
+      "step": 8347
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00018329482207294734,
+      "loss": 1.1391,
+      "step": 8348
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00018325300005517123,
+      "loss": 1.04,
+      "step": 8349
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001832111787748494,
+      "loss": 1.0341,
+      "step": 8350
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001831693582338236,
+      "loss": 0.9895,
+      "step": 8351
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001831275384339353,
+      "loss": 0.9498,
+      "step": 8352
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00018308571937702606,
+      "loss": 1.0041,
+      "step": 8353
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00018304390106493727,
+      "loss": 1.0679,
+      "step": 8354
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001830020834995106,
+      "loss": 1.0614,
+      "step": 8355
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00018296026668258741,
+      "loss": 1.009,
+      "step": 8356
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00018291845061600904,
+      "loss": 1.0084,
+      "step": 8357
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00018287663530161695,
+      "loss": 1.0107,
+      "step": 8358
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00018283482074125243,
+      "loss": 1.0693,
+      "step": 8359
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00018279300693675672,
+      "loss": 1.0786,
+      "step": 8360
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00018275119388997123,
+      "loss": 0.9371,
+      "step": 8361
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00018270938160273715,
+      "loss": 0.9678,
+      "step": 8362
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00018266757007689562,
+      "loss": 1.0314,
+      "step": 8363
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00018262575931428789,
+      "loss": 1.0143,
+      "step": 8364
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00018258394931675506,
+      "loss": 1.0675,
+      "step": 8365
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00018254214008613817,
+      "loss": 1.0148,
+      "step": 8366
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00018250033162427843,
+      "loss": 1.0177,
+      "step": 8367
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00018245852393301682,
+      "loss": 1.0444,
+      "step": 8368
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00018241671701419427,
+      "loss": 0.9958,
+      "step": 8369
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00018237491086965183,
+      "loss": 1.1,
+      "step": 8370
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00018233310550123044,
+      "loss": 1.0424,
+      "step": 8371
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00018229130091077086,
+      "loss": 1.0207,
+      "step": 8372
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00018224949710011416,
+      "loss": 0.9968,
+      "step": 8373
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001822076940711011,
+      "loss": 0.9939,
+      "step": 8374
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00018216589182557238,
+      "loss": 1.084,
+      "step": 8375
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001821240903653689,
+      "loss": 1.0445,
+      "step": 8376
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00018208228969233126,
+      "loss": 1.0276,
+      "step": 8377
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001820404898083002,
+      "loss": 1.0359,
+      "step": 8378
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00018199869071511645,
+      "loss": 1.0462,
+      "step": 8379
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001819568924146206,
+      "loss": 1.0134,
+      "step": 8380
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001819150949086532,
+      "loss": 0.9417,
+      "step": 8381
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00018187329819905476,
+      "loss": 1.0389,
+      "step": 8382
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00018183150228766592,
+      "loss": 1.0293,
+      "step": 8383
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00018178970717632712,
+      "loss": 0.9744,
+      "step": 8384
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00018174791286687867,
+      "loss": 1.0557,
+      "step": 8385
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00018170611936116118,
+      "loss": 1.0388,
+      "step": 8386
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.000181664326661015,
+      "loss": 0.9657,
+      "step": 8387
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00018162253476828038,
+      "loss": 1.0003,
+      "step": 8388
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00018158074368479767,
+      "loss": 1.0512,
+      "step": 8389
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00018153895341240715,
+      "loss": 0.9363,
+      "step": 8390
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00018149716395294898,
+      "loss": 0.983,
+      "step": 8391
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00018145537530826352,
+      "loss": 1.0404,
+      "step": 8392
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001814135874801908,
+      "loss": 1.0403,
+      "step": 8393
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00018137180047057095,
+      "loss": 1.0275,
+      "step": 8394
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00018133001428124416,
+      "loss": 1.0626,
+      "step": 8395
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001812882289140504,
+      "loss": 1.0516,
+      "step": 8396
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00018124644437082964,
+      "loss": 0.9424,
+      "step": 8397
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00018120466065342202,
+      "loss": 0.9558,
+      "step": 8398
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001811628777636674,
+      "loss": 0.9605,
+      "step": 8399
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00018112109570340567,
+      "loss": 0.9889,
+      "step": 8400
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00018107931447447674,
+      "loss": 1.0534,
+      "step": 8401
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00018103753407872043,
+      "loss": 1.0132,
+      "step": 8402
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00018099575451797648,
+      "loss": 1.0627,
+      "step": 8403
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001809539757940848,
+      "loss": 0.9865,
+      "step": 8404
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.000180912197908885,
+      "loss": 0.9283,
+      "step": 8405
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00018087042086421684,
+      "loss": 0.9643,
+      "step": 8406
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00018082864466191986,
+      "loss": 1.0545,
+      "step": 8407
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001807868693038338,
+      "loss": 1.0832,
+      "step": 8408
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00018074509479179822,
+      "loss": 0.9997,
+      "step": 8409
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001807033211276525,
+      "loss": 0.9616,
+      "step": 8410
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00018066154831323638,
+      "loss": 1.0586,
+      "step": 8411
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00018061977635038922,
+      "loss": 1.0382,
+      "step": 8412
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00018057800524095042,
+      "loss": 0.932,
+      "step": 8413
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00018053623498675943,
+      "loss": 1.0168,
+      "step": 8414
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00018049446558965557,
+      "loss": 1.037,
+      "step": 8415
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00018045269705147806,
+      "loss": 1.0634,
+      "step": 8416
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001804109293740664,
+      "loss": 1.0077,
+      "step": 8417
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001803691625592597,
+      "loss": 0.9835,
+      "step": 8418
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001803273966088971,
+      "loss": 1.0073,
+      "step": 8419
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001802856315248179,
+      "loss": 1.0082,
+      "step": 8420
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00018024386730886117,
+      "loss": 1.0422,
+      "step": 8421
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00018020210396286587,
+      "loss": 1.0259,
+      "step": 8422
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00018016034148867128,
+      "loss": 1.0766,
+      "step": 8423
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00018011857988811626,
+      "loss": 0.9943,
+      "step": 8424
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00018007681916303982,
+      "loss": 0.9918,
+      "step": 8425
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00018003505931528092,
+      "loss": 1.0483,
+      "step": 8426
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017999330034667845,
+      "loss": 1.004,
+      "step": 8427
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017995154225907114,
+      "loss": 1.0058,
+      "step": 8428
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.000179909785054298,
+      "loss": 1.0162,
+      "step": 8429
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017986802873419774,
+      "loss": 1.101,
+      "step": 8430
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017982627330060904,
+      "loss": 1.0028,
+      "step": 8431
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001797845187553707,
+      "loss": 0.9942,
+      "step": 8432
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001797427651003213,
+      "loss": 0.9952,
+      "step": 8433
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001797010123372995,
+      "loss": 1.0942,
+      "step": 8434
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017965926046814378,
+      "loss": 0.9674,
+      "step": 8435
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017961750949469288,
+      "loss": 1.0135,
+      "step": 8436
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001795757594187852,
+      "loss": 1.1063,
+      "step": 8437
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017953401024225915,
+      "loss": 1.0923,
+      "step": 8438
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017949226196695327,
+      "loss": 1.0206,
+      "step": 8439
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017945051459470588,
+      "loss": 1.0806,
+      "step": 8440
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017940876812735525,
+      "loss": 1.0268,
+      "step": 8441
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017936702256673986,
+      "loss": 1.0771,
+      "step": 8442
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017932527791469788,
+      "loss": 1.0876,
+      "step": 8443
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001792835341730675,
+      "loss": 1.0947,
+      "step": 8444
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017924179134368702,
+      "loss": 0.9831,
+      "step": 8445
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001792000494283945,
+      "loss": 1.0264,
+      "step": 8446
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017915830842902796,
+      "loss": 1.0255,
+      "step": 8447
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017911656834742568,
+      "loss": 0.9626,
+      "step": 8448
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017907482918542555,
+      "loss": 0.9607,
+      "step": 8449
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017903309094486556,
+      "loss": 0.9578,
+      "step": 8450
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017899135362758369,
+      "loss": 1.0208,
+      "step": 8451
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001789496172354178,
+      "loss": 1.0197,
+      "step": 8452
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017890788177020575,
+      "loss": 1.0451,
+      "step": 8453
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017886614723378543,
+      "loss": 1.0483,
+      "step": 8454
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017882441362799457,
+      "loss": 0.9951,
+      "step": 8455
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017878268095467093,
+      "loss": 1.0404,
+      "step": 8456
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001787409492156522,
+      "loss": 1.019,
+      "step": 8457
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017869921841277603,
+      "loss": 1.1181,
+      "step": 8458
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017865748854788006,
+      "loss": 1.0061,
+      "step": 8459
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017861575962280175,
+      "loss": 1.0312,
+      "step": 8460
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017857403163937884,
+      "loss": 1.0432,
+      "step": 8461
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017853230459944868,
+      "loss": 0.9725,
+      "step": 8462
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017849057850484871,
+      "loss": 0.9901,
+      "step": 8463
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017844885335741647,
+      "loss": 1.0716,
+      "step": 8464
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001784071291589892,
+      "loss": 1.083,
+      "step": 8465
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017836540591140417,
+      "loss": 1.0663,
+      "step": 8466
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017832368361649886,
+      "loss": 1.0085,
+      "step": 8467
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017828196227611043,
+      "loss": 0.9549,
+      "step": 8468
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.000178240241892076,
+      "loss": 0.9623,
+      "step": 8469
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017819852246623284,
+      "loss": 1.037,
+      "step": 8470
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.000178156804000418,
+      "loss": 0.9777,
+      "step": 8471
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017811508649646847,
+      "loss": 1.0231,
+      "step": 8472
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001780733699562215,
+      "loss": 1.0428,
+      "step": 8473
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017803165438151395,
+      "loss": 1.0596,
+      "step": 8474
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017798993977418271,
+      "loss": 1.0542,
+      "step": 8475
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001779482261360648,
+      "loss": 1.0984,
+      "step": 8476
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.000177906513468997,
+      "loss": 0.97,
+      "step": 8477
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001778648017748161,
+      "loss": 1.0255,
+      "step": 8478
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.000177823091055359,
+      "loss": 1.0001,
+      "step": 8479
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017778138131246236,
+      "loss": 1.0239,
+      "step": 8480
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017773967254796285,
+      "loss": 0.9552,
+      "step": 8481
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017769796476369714,
+      "loss": 1.0239,
+      "step": 8482
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017765625796150184,
+      "loss": 1.047,
+      "step": 8483
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017761455214321343,
+      "loss": 1.0865,
+      "step": 8484
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017757284731066857,
+      "loss": 0.9528,
+      "step": 8485
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017753114346570365,
+      "loss": 1.0024,
+      "step": 8486
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001774894406101551,
+      "loss": 1.0945,
+      "step": 8487
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017744773874585928,
+      "loss": 0.9958,
+      "step": 8488
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017740603787465262,
+      "loss": 1.0963,
+      "step": 8489
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017736433799837135,
+      "loss": 0.9709,
+      "step": 8490
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017732263911885164,
+      "loss": 0.9882,
+      "step": 8491
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017728094123792992,
+      "loss": 1.0952,
+      "step": 8492
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001772392443574422,
+      "loss": 0.9739,
+      "step": 8493
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017719754847922463,
+      "loss": 0.9745,
+      "step": 8494
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017715585360511332,
+      "loss": 0.9742,
+      "step": 8495
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017711415973694427,
+      "loss": 0.9188,
+      "step": 8496
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017707246687655338,
+      "loss": 0.9703,
+      "step": 8497
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001770307750257768,
+      "loss": 0.9879,
+      "step": 8498
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017698908418645033,
+      "loss": 0.9711,
+      "step": 8499
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017694739436040978,
+      "loss": 0.9989,
+      "step": 8500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017690570554949105,
+      "loss": 1.0498,
+      "step": 8501
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017686401775552984,
+      "loss": 1.0268,
+      "step": 8502
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017682233098036183,
+      "loss": 0.9344,
+      "step": 8503
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017678064522582286,
+      "loss": 1.0496,
+      "step": 8504
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017673896049374846,
+      "loss": 1.0401,
+      "step": 8505
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017669727678597415,
+      "loss": 1.0582,
+      "step": 8506
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017665559410433561,
+      "loss": 1.0356,
+      "step": 8507
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017661391245066827,
+      "loss": 1.011,
+      "step": 8508
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001765722318268075,
+      "loss": 1.0354,
+      "step": 8509
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001765305522345889,
+      "loss": 0.9777,
+      "step": 8510
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001764888736758477,
+      "loss": 1.0111,
+      "step": 8511
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017644719615241924,
+      "loss": 0.9458,
+      "step": 8512
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017640551966613876,
+      "loss": 0.9914,
+      "step": 8513
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017636384421884155,
+      "loss": 0.9922,
+      "step": 8514
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017632216981236276,
+      "loss": 1.0718,
+      "step": 8515
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001762804964485374,
+      "loss": 1.0542,
+      "step": 8516
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00017623882412920078,
+      "loss": 1.0223,
+      "step": 8517
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00017619715285618787,
+      "loss": 0.9557,
+      "step": 8518
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00017615548263133355,
+      "loss": 0.998,
+      "step": 8519
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00017611381345647288,
+      "loss": 0.9848,
+      "step": 8520
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00017607214533344075,
+      "loss": 1.0611,
+      "step": 8521
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001760304782640719,
+      "loss": 1.003,
+      "step": 8522
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001759888122502013,
+      "loss": 0.9017,
+      "step": 8523
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001759471472936637,
+      "loss": 1.0591,
+      "step": 8524
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00017590548339629371,
+      "loss": 1.0187,
+      "step": 8525
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00017586382055992608,
+      "loss": 1.0097,
+      "step": 8526
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00017582215878639542,
+      "loss": 1.0193,
+      "step": 8527
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00017578049807753622,
+      "loss": 0.9784,
+      "step": 8528
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001757388384351832,
+      "loss": 0.9671,
+      "step": 8529
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00017569717986117068,
+      "loss": 1.0174,
+      "step": 8530
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00017565552235733313,
+      "loss": 1.07,
+      "step": 8531
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.000175613865925505,
+      "loss": 1.0506,
+      "step": 8532
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00017557221056752057,
+      "loss": 0.9805,
+      "step": 8533
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001755305562852141,
+      "loss": 1.052,
+      "step": 8534
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00017548890308041992,
+      "loss": 0.9776,
+      "step": 8535
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00017544725095497223,
+      "loss": 1.0833,
+      "step": 8536
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001754055999107051,
+      "loss": 1.0241,
+      "step": 8537
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00017536394994945273,
+      "loss": 0.9811,
+      "step": 8538
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00017532230107304911,
+      "loss": 0.9585,
+      "step": 8539
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001752806532833283,
+      "loss": 1.0952,
+      "step": 8540
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00017523900658212414,
+      "loss": 0.942,
+      "step": 8541
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001751973609712707,
+      "loss": 1.057,
+      "step": 8542
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00017515571645260178,
+      "loss": 0.942,
+      "step": 8543
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00017511407302795115,
+      "loss": 1.0102,
+      "step": 8544
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001750724306991527,
+      "loss": 1.1092,
+      "step": 8545
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00017503078946804002,
+      "loss": 0.9896,
+      "step": 8546
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001749891493364468,
+      "loss": 1.0816,
+      "step": 8547
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00017494751030620675,
+      "loss": 1.0389,
+      "step": 8548
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00017490587237915342,
+      "loss": 1.0537,
+      "step": 8549
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00017486423555712025,
+      "loss": 0.9078,
+      "step": 8550
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00017482259984194084,
+      "loss": 1.0187,
+      "step": 8551
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001747809652354485,
+      "loss": 0.987,
+      "step": 8552
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00017473933173947664,
+      "loss": 0.9301,
+      "step": 8553
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001746976993558587,
+      "loss": 1.003,
+      "step": 8554
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00017465606808642788,
+      "loss": 0.9232,
+      "step": 8555
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00017461443793301737,
+      "loss": 1.095,
+      "step": 8556
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00017457280889746044,
+      "loss": 1.0637,
+      "step": 8557
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001745311809815902,
+      "loss": 0.9893,
+      "step": 8558
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001744895541872396,
+      "loss": 1.0748,
+      "step": 8559
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001744479285162419,
+      "loss": 1.0018,
+      "step": 8560
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00017440630397043001,
+      "loss": 0.9851,
+      "step": 8561
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00017436468055163679,
+      "loss": 1.0418,
+      "step": 8562
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001743230582616952,
+      "loss": 1.0069,
+      "step": 8563
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00017428143710243807,
+      "loss": 1.0234,
+      "step": 8564
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00017423981707569809,
+      "loss": 1.0733,
+      "step": 8565
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00017419819818330817,
+      "loss": 0.9718,
+      "step": 8566
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00017415658042710096,
+      "loss": 1.0532,
+      "step": 8567
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.000174114963808909,
+      "loss": 0.9988,
+      "step": 8568
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00017407334833056487,
+      "loss": 0.9561,
+      "step": 8569
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00017403173399390125,
+      "loss": 0.9871,
+      "step": 8570
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001739901208007505,
+      "loss": 1.0026,
+      "step": 8571
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001739485087529451,
+      "loss": 1.0655,
+      "step": 8572
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00017390689785231746,
+      "loss": 1.1237,
+      "step": 8573
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00017386528810069992,
+      "loss": 1.0133,
+      "step": 8574
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00017382367949992468,
+      "loss": 0.9466,
+      "step": 8575
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001737820720518241,
+      "loss": 0.97,
+      "step": 8576
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001737404657582303,
+      "loss": 0.9818,
+      "step": 8577
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00017369886062097533,
+      "loss": 1.0444,
+      "step": 8578
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00017365725664189144,
+      "loss": 0.9074,
+      "step": 8579
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00017361565382281057,
+      "loss": 0.9322,
+      "step": 8580
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00017357405216556472,
+      "loss": 0.9483,
+      "step": 8581
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001735324516719858,
+      "loss": 1.0933,
+      "step": 8582
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001734908523439057,
+      "loss": 0.9863,
+      "step": 8583
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00017344925418315618,
+      "loss": 0.9902,
+      "step": 8584
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00017340765719156916,
+      "loss": 1.0872,
+      "step": 8585
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00017336606137097627,
+      "loss": 1.0213,
+      "step": 8586
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001733244667232092,
+      "loss": 0.8969,
+      "step": 8587
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00017328287325009956,
+      "loss": 1.0369,
+      "step": 8588
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00017324128095347896,
+      "loss": 1.049,
+      "step": 8589
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001731996898351788,
+      "loss": 1.0767,
+      "step": 8590
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00017315809989703072,
+      "loss": 0.9975,
+      "step": 8591
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00017311651114086603,
+      "loss": 1.0172,
+      "step": 8592
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001730749235685161,
+      "loss": 1.0485,
+      "step": 8593
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001730333371818122,
+      "loss": 1.0109,
+      "step": 8594
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001729917519825857,
+      "loss": 1.1096,
+      "step": 8595
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00017295016797266773,
+      "loss": 1.0597,
+      "step": 8596
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00017290858515388934,
+      "loss": 0.9597,
+      "step": 8597
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00017286700352808185,
+      "loss": 1.0614,
+      "step": 8598
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00017282542309707616,
+      "loss": 1.0263,
+      "step": 8599
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001727838438627033,
+      "loss": 0.9844,
+      "step": 8600
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00017274226582679423,
+      "loss": 0.992,
+      "step": 8601
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00017270068899117983,
+      "loss": 0.9998,
+      "step": 8602
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00017265911335769086,
+      "loss": 1.0325,
+      "step": 8603
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00017261753892815825,
+      "loss": 1.0836,
+      "step": 8604
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00017257596570441263,
+      "loss": 0.9528,
+      "step": 8605
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00017253439368828468,
+      "loss": 0.9963,
+      "step": 8606
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001724928228816051,
+      "loss": 1.0524,
+      "step": 8607
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00017245125328620438,
+      "loss": 1.0174,
+      "step": 8608
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00017240968490391298,
+      "loss": 1.1565,
+      "step": 8609
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00017236811773656155,
+      "loss": 1.0198,
+      "step": 8610
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001723265517859804,
+      "loss": 0.9749,
+      "step": 8611
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001722849870539998,
+      "loss": 1.0241,
+      "step": 8612
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00017224342354245023,
+      "loss": 1.0046,
+      "step": 8613
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00017220186125316183,
+      "loss": 0.9787,
+      "step": 8614
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00017216030018796471,
+      "loss": 1.0146,
+      "step": 8615
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00017211874034868925,
+      "loss": 1.0682,
+      "step": 8616
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00017207718173716536,
+      "loss": 0.9453,
+      "step": 8617
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00017203562435522308,
+      "loss": 1.0023,
+      "step": 8618
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001719940682046925,
+      "loss": 0.9892,
+      "step": 8619
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00017195251328740345,
+      "loss": 1.0195,
+      "step": 8620
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00017191095960518584,
+      "loss": 1.0177,
+      "step": 8621
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00017186940715986938,
+      "loss": 1.0458,
+      "step": 8622
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00017182785595328402,
+      "loss": 0.9849,
+      "step": 8623
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00017178630598725935,
+      "loss": 0.9966,
+      "step": 8624
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00017174475726362503,
+      "loss": 1.0254,
+      "step": 8625
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001717032097842107,
+      "loss": 1.0396,
+      "step": 8626
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00017166166355084587,
+      "loss": 1.0247,
+      "step": 8627
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00017162011856536,
+      "loss": 0.9926,
+      "step": 8628
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001715785748295826,
+      "loss": 1.0466,
+      "step": 8629
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.000171537032345343,
+      "loss": 1.0646,
+      "step": 8630
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001714954911144705,
+      "loss": 1.0691,
+      "step": 8631
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00017145395113879447,
+      "loss": 1.0281,
+      "step": 8632
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00017141241242014406,
+      "loss": 1.0254,
+      "step": 8633
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00017137087496034832,
+      "loss": 1.0426,
+      "step": 8634
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00017132933876123652,
+      "loss": 0.9996,
+      "step": 8635
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00017128780382463769,
+      "loss": 1.0102,
+      "step": 8636
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001712462701523807,
+      "loss": 0.9969,
+      "step": 8637
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001712047377462946,
+      "loss": 0.9635,
+      "step": 8638
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00017116320660820827,
+      "loss": 0.9859,
+      "step": 8639
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001711216767399504,
+      "loss": 1.1152,
+      "step": 8640
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001710801481433499,
+      "loss": 0.9387,
+      "step": 8641
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00017103862082023545,
+      "loss": 1.0013,
+      "step": 8642
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00017099709477243566,
+      "loss": 1.0038,
+      "step": 8643
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001709555700017792,
+      "loss": 1.0237,
+      "step": 8644
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00017091404651009456,
+      "loss": 0.9715,
+      "step": 8645
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00017087252429921028,
+      "loss": 0.9795,
+      "step": 8646
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00017083100337095464,
+      "loss": 1.0731,
+      "step": 8647
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00017078948372715623,
+      "loss": 1.0565,
+      "step": 8648
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00017074796536964323,
+      "loss": 0.9908,
+      "step": 8649
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00017070644830024395,
+      "loss": 0.9812,
+      "step": 8650
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001706649325207866,
+      "loss": 1.0029,
+      "step": 8651
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001706234180330993,
+      "loss": 1.028,
+      "step": 8652
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00017058190483901006,
+      "loss": 1.0433,
+      "step": 8653
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00017054039294034712,
+      "loss": 0.9815,
+      "step": 8654
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00017049888233893832,
+      "loss": 1.049,
+      "step": 8655
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00017045737303661158,
+      "loss": 1.1013,
+      "step": 8656
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001704158650351948,
+      "loss": 1.0241,
+      "step": 8657
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00017037435833651578,
+      "loss": 0.9639,
+      "step": 8658
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001703328529424022,
+      "loss": 1.0139,
+      "step": 8659
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00017029134885468186,
+      "loss": 1.0108,
+      "step": 8660
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00017024984607518238,
+      "loss": 1.022,
+      "step": 8661
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00017020834460573122,
+      "loss": 1.0064,
+      "step": 8662
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00017016684444815604,
+      "loss": 1.0372,
+      "step": 8663
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00017012534560428418,
+      "loss": 1.0168,
+      "step": 8664
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00017008384807594307,
+      "loss": 1.0058,
+      "step": 8665
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00017004235186496015,
+      "loss": 1.035,
+      "step": 8666
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00017000085697316264,
+      "loss": 1.0911,
+      "step": 8667
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00016995936340237773,
+      "loss": 1.007,
+      "step": 8668
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00016991787115443264,
+      "loss": 1.0556,
+      "step": 8669
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001698763802311545,
+      "loss": 1.0357,
+      "step": 8670
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001698348906343702,
+      "loss": 1.0617,
+      "step": 8671
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016979340236590698,
+      "loss": 1.0286,
+      "step": 8672
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001697519154275917,
+      "loss": 1.0431,
+      "step": 8673
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016971042982125115,
+      "loss": 1.0903,
+      "step": 8674
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016966894554871215,
+      "loss": 0.9317,
+      "step": 8675
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001696274626118016,
+      "loss": 0.9748,
+      "step": 8676
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016958598101234607,
+      "loss": 1.0586,
+      "step": 8677
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016954450075217216,
+      "loss": 0.9815,
+      "step": 8678
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016950302183310666,
+      "loss": 1.0415,
+      "step": 8679
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016946154425697592,
+      "loss": 1.0342,
+      "step": 8680
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016942006802560643,
+      "loss": 0.9674,
+      "step": 8681
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016937859314082468,
+      "loss": 1.0273,
+      "step": 8682
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016933711960445696,
+      "loss": 0.9825,
+      "step": 8683
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016929564741832945,
+      "loss": 1.038,
+      "step": 8684
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001692541765842686,
+      "loss": 0.9984,
+      "step": 8685
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016921270710410047,
+      "loss": 1.0154,
+      "step": 8686
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016917123897965112,
+      "loss": 1.0465,
+      "step": 8687
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016912977221274667,
+      "loss": 1.047,
+      "step": 8688
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016908830680521312,
+      "loss": 0.9914,
+      "step": 8689
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001690468427588762,
+      "loss": 1.068,
+      "step": 8690
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001690053800755621,
+      "loss": 1.0315,
+      "step": 8691
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016896391875709645,
+      "loss": 0.9779,
+      "step": 8692
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016892245880530498,
+      "loss": 0.9385,
+      "step": 8693
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016888100022201345,
+      "loss": 1.0646,
+      "step": 8694
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016883954300904748,
+      "loss": 1.0316,
+      "step": 8695
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016879808716823254,
+      "loss": 1.0419,
+      "step": 8696
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001687566327013943,
+      "loss": 1.047,
+      "step": 8697
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001687151796103581,
+      "loss": 0.9809,
+      "step": 8698
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016867372789694936,
+      "loss": 0.9923,
+      "step": 8699
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001686322775629934,
+      "loss": 0.9517,
+      "step": 8700
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001685908286103155,
+      "loss": 0.9757,
+      "step": 8701
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016854938104074086,
+      "loss": 0.9713,
+      "step": 8702
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016850793485609453,
+      "loss": 0.9416,
+      "step": 8703
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016846649005820176,
+      "loss": 1.0295,
+      "step": 8704
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016842504664888749,
+      "loss": 0.9199,
+      "step": 8705
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016838360462997665,
+      "loss": 0.9913,
+      "step": 8706
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016834216400329422,
+      "loss": 1.0647,
+      "step": 8707
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016830072477066498,
+      "loss": 0.9787,
+      "step": 8708
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016825928693391362,
+      "loss": 1.024,
+      "step": 8709
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016821785049486504,
+      "loss": 1.0057,
+      "step": 8710
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016817641545534384,
+      "loss": 1.0066,
+      "step": 8711
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001681349818171745,
+      "loss": 0.9891,
+      "step": 8712
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016809354958218168,
+      "loss": 1.0113,
+      "step": 8713
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016805211875218983,
+      "loss": 1.0052,
+      "step": 8714
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001680106893290232,
+      "loss": 1.0574,
+      "step": 8715
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001679692613145064,
+      "loss": 0.9133,
+      "step": 8716
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016792783471046354,
+      "loss": 1.0232,
+      "step": 8717
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016788640951871883,
+      "loss": 1.005,
+      "step": 8718
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016784498574109652,
+      "loss": 1.0207,
+      "step": 8719
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001678035633794207,
+      "loss": 0.9981,
+      "step": 8720
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016776214243551522,
+      "loss": 1.0937,
+      "step": 8721
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016772072291120434,
+      "loss": 1.0088,
+      "step": 8722
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0001676793048083118,
+      "loss": 1.0022,
+      "step": 8723
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0001676378881286614,
+      "loss": 1.0519,
+      "step": 8724
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016759647287407708,
+      "loss": 0.9734,
+      "step": 8725
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016755505904638248,
+      "loss": 1.0518,
+      "step": 8726
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016751364664740125,
+      "loss": 1.0293,
+      "step": 8727
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016747223567895693,
+      "loss": 1.0787,
+      "step": 8728
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016743082614287315,
+      "loss": 1.0033,
+      "step": 8729
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016738941804097336,
+      "loss": 1.0212,
+      "step": 8730
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016734801137508094,
+      "loss": 1.0184,
+      "step": 8731
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016730660614701926,
+      "loss": 0.9709,
+      "step": 8732
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0001672652023586116,
+      "loss": 0.9499,
+      "step": 8733
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016722380001168107,
+      "loss": 0.9524,
+      "step": 8734
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016718239910805098,
+      "loss": 1.0598,
+      "step": 8735
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016714099964954438,
+      "loss": 1.0503,
+      "step": 8736
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0001670996016379842,
+      "loss": 1.008,
+      "step": 8737
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016705820507519355,
+      "loss": 1.0953,
+      "step": 8738
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016701680996299522,
+      "loss": 1.0165,
+      "step": 8739
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.000166975416303212,
+      "loss": 1.0202,
+      "step": 8740
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0001669340240976668,
+      "loss": 1.0904,
+      "step": 8741
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016689263334818225,
+      "loss": 1.1176,
+      "step": 8742
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.000166851244056581,
+      "loss": 0.9519,
+      "step": 8743
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016680985622468563,
+      "loss": 1.1237,
+      "step": 8744
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0001667684698543187,
+      "loss": 0.9996,
+      "step": 8745
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016672708494730248,
+      "loss": 1.0257,
+      "step": 8746
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016668570150545962,
+      "loss": 1.029,
+      "step": 8747
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016664431953061225,
+      "loss": 0.9767,
+      "step": 8748
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0001666029390245827,
+      "loss": 0.9923,
+      "step": 8749
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016656155998919314,
+      "loss": 1.0287,
+      "step": 8750
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016652018242626571,
+      "loss": 0.9925,
+      "step": 8751
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016647880633762242,
+      "loss": 1.0525,
+      "step": 8752
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016643743172508533,
+      "loss": 1.0449,
+      "step": 8753
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016639605859047638,
+      "loss": 0.9633,
+      "step": 8754
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016635468693561742,
+      "loss": 1.0835,
+      "step": 8755
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016631331676233017,
+      "loss": 1.0552,
+      "step": 8756
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016627194807243648,
+      "loss": 0.9321,
+      "step": 8757
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.000166230580867758,
+      "loss": 1.0344,
+      "step": 8758
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016618921515011622,
+      "loss": 1.0278,
+      "step": 8759
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016614785092133286,
+      "loss": 1.0463,
+      "step": 8760
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016610648818322926,
+      "loss": 1.0518,
+      "step": 8761
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016606512693762688,
+      "loss": 0.9902,
+      "step": 8762
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0001660237671863471,
+      "loss": 1.0758,
+      "step": 8763
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016598240893121112,
+      "loss": 0.9534,
+      "step": 8764
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0001659410521740401,
+      "loss": 0.9769,
+      "step": 8765
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016589969691665539,
+      "loss": 1.0432,
+      "step": 8766
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016585834316087792,
+      "loss": 0.945,
+      "step": 8767
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0001658169909085287,
+      "loss": 1.0613,
+      "step": 8768
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016577564016142875,
+      "loss": 1.0264,
+      "step": 8769
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0001657342909213989,
+      "loss": 1.004,
+      "step": 8770
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016569294319025992,
+      "loss": 1.0415,
+      "step": 8771
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016565159696983267,
+      "loss": 1.052,
+      "step": 8772
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016561025226193777,
+      "loss": 1.0118,
+      "step": 8773
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016556890906839583,
+      "loss": 1.0009,
+      "step": 8774
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016552756739102745,
+      "loss": 1.0133,
+      "step": 8775
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016548622723165304,
+      "loss": 1.0118,
+      "step": 8776
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016544488859209294,
+      "loss": 0.9407,
+      "step": 8777
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001654035514741677,
+      "loss": 1.0243,
+      "step": 8778
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001653622158796975,
+      "loss": 1.0462,
+      "step": 8779
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016532088181050257,
+      "loss": 1.0674,
+      "step": 8780
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016527954926840298,
+      "loss": 0.9662,
+      "step": 8781
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016523821825521892,
+      "loss": 1.0247,
+      "step": 8782
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016519688877277038,
+      "loss": 0.9974,
+      "step": 8783
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016515556082287715,
+      "loss": 0.9513,
+      "step": 8784
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016511423440735933,
+      "loss": 0.9873,
+      "step": 8785
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016507290952803662,
+      "loss": 0.9775,
+      "step": 8786
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016503158618672874,
+      "loss": 0.9844,
+      "step": 8787
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016499026438525545,
+      "loss": 1.0506,
+      "step": 8788
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016494894412543627,
+      "loss": 1.009,
+      "step": 8789
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001649076254090907,
+      "loss": 1.0137,
+      "step": 8790
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016486630823803837,
+      "loss": 1.0864,
+      "step": 8791
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001648249926140986,
+      "loss": 1.0288,
+      "step": 8792
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016478367853909066,
+      "loss": 1.0128,
+      "step": 8793
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016474236601483392,
+      "loss": 1.0526,
+      "step": 8794
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016470105504314754,
+      "loss": 1.0224,
+      "step": 8795
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016465974562585053,
+      "loss": 1.067,
+      "step": 8796
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016461843776476219,
+      "loss": 0.9187,
+      "step": 8797
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016457713146170134,
+      "loss": 1.0523,
+      "step": 8798
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001645358267184869,
+      "loss": 1.0499,
+      "step": 8799
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016449452353693784,
+      "loss": 0.9794,
+      "step": 8800
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016445322191887288,
+      "loss": 0.9568,
+      "step": 8801
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001644119218661106,
+      "loss": 1.024,
+      "step": 8802
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001643706233804699,
+      "loss": 1.027,
+      "step": 8803
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016432932646376927,
+      "loss": 0.9635,
+      "step": 8804
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001642880311178271,
+      "loss": 0.9967,
+      "step": 8805
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016424673734446202,
+      "loss": 1.0459,
+      "step": 8806
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016420544514549225,
+      "loss": 1.0243,
+      "step": 8807
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001641641545227362,
+      "loss": 1.0529,
+      "step": 8808
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001641228654780119,
+      "loss": 0.9699,
+      "step": 8809
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001640815780131378,
+      "loss": 1.0016,
+      "step": 8810
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016404029212993183,
+      "loss": 1.0303,
+      "step": 8811
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016399900783021202,
+      "loss": 0.9663,
+      "step": 8812
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001639577251157964,
+      "loss": 0.9986,
+      "step": 8813
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016391644398850276,
+      "loss": 0.9961,
+      "step": 8814
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001638751644501489,
+      "loss": 0.9504,
+      "step": 8815
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016383388650255268,
+      "loss": 0.9488,
+      "step": 8816
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016379261014753176,
+      "loss": 0.9813,
+      "step": 8817
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016375133538690365,
+      "loss": 0.9437,
+      "step": 8818
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016371006222248594,
+      "loss": 1.0229,
+      "step": 8819
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016366879065609613,
+      "loss": 1.0163,
+      "step": 8820
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016362752068955146,
+      "loss": 0.9834,
+      "step": 8821
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016358625232466948,
+      "loss": 1.0062,
+      "step": 8822
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016354498556326731,
+      "loss": 1.0072,
+      "step": 8823
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016350372040716213,
+      "loss": 1.027,
+      "step": 8824
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016346245685817113,
+      "loss": 1.0087,
+      "step": 8825
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016342119491811127,
+      "loss": 1.0343,
+      "step": 8826
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016337993458879945,
+      "loss": 1.0379,
+      "step": 8827
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016333867587205276,
+      "loss": 0.9945,
+      "step": 8828
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016329741876968795,
+      "loss": 1.0257,
+      "step": 8829
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016325616328352168,
+      "loss": 0.9346,
+      "step": 8830
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001632149094153708,
+      "loss": 1.0519,
+      "step": 8831
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001631736571670518,
+      "loss": 1.0296,
+      "step": 8832
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016313240654038122,
+      "loss": 1.0012,
+      "step": 8833
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016309115753717563,
+      "loss": 1.0634,
+      "step": 8834
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001630499101592514,
+      "loss": 1.013,
+      "step": 8835
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001630086644084248,
+      "loss": 1.0153,
+      "step": 8836
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016296742028651214,
+      "loss": 1.0892,
+      "step": 8837
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016292617779532957,
+      "loss": 0.956,
+      "step": 8838
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016288493693669324,
+      "loss": 1.0185,
+      "step": 8839
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001628436977124191,
+      "loss": 1.0258,
+      "step": 8840
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016280246012432325,
+      "loss": 1.013,
+      "step": 8841
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016276122417422157,
+      "loss": 1.0515,
+      "step": 8842
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016271998986392978,
+      "loss": 0.9831,
+      "step": 8843
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016267875719526374,
+      "loss": 0.9931,
+      "step": 8844
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016263752617003908,
+      "loss": 1.0069,
+      "step": 8845
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016259629679007135,
+      "loss": 0.9945,
+      "step": 8846
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016255506905717623,
+      "loss": 0.9563,
+      "step": 8847
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001625138429731691,
+      "loss": 0.9974,
+      "step": 8848
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016247261853986536,
+      "loss": 1.0201,
+      "step": 8849
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016243139575908034,
+      "loss": 1.0797,
+      "step": 8850
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016239017463262926,
+      "loss": 1.005,
+      "step": 8851
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016234895516232725,
+      "loss": 1.0659,
+      "step": 8852
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016230773734998954,
+      "loss": 0.9557,
+      "step": 8853
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016226652119743107,
+      "loss": 0.982,
+      "step": 8854
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016222530670646678,
+      "loss": 1.0047,
+      "step": 8855
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016218409387891162,
+      "loss": 1.0088,
+      "step": 8856
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016214288271658035,
+      "loss": 1.053,
+      "step": 8857
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016210167322128763,
+      "loss": 1.0514,
+      "step": 8858
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001620604653948483,
+      "loss": 1.0654,
+      "step": 8859
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016201925923907683,
+      "loss": 1.0747,
+      "step": 8860
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016197805475578773,
+      "loss": 0.9715,
+      "step": 8861
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016193685194679548,
+      "loss": 0.9728,
+      "step": 8862
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001618956508139144,
+      "loss": 1.0311,
+      "step": 8863
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016185445135895885,
+      "loss": 1.0511,
+      "step": 8864
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016181325358374292,
+      "loss": 1.0503,
+      "step": 8865
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016177205749008092,
+      "loss": 1.0059,
+      "step": 8866
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016173086307978682,
+      "loss": 1.0027,
+      "step": 8867
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016168967035467462,
+      "loss": 0.9455,
+      "step": 8868
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001616484793165583,
+      "loss": 1.0406,
+      "step": 8869
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016160728996725168,
+      "loss": 0.9818,
+      "step": 8870
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016156610230856844,
+      "loss": 1.0517,
+      "step": 8871
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016152491634232244,
+      "loss": 0.9942,
+      "step": 8872
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001614837320703272,
+      "loss": 1.08,
+      "step": 8873
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016144254949439628,
+      "loss": 1.0337,
+      "step": 8874
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016140136861634321,
+      "loss": 1.0284,
+      "step": 8875
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016136018943798135,
+      "loss": 1.0458,
+      "step": 8876
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016131901196112392,
+      "loss": 1.0097,
+      "step": 8877
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001612778361875844,
+      "loss": 0.99,
+      "step": 8878
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016123666211917582,
+      "loss": 1.0041,
+      "step": 8879
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001611954897577113,
+      "loss": 1.0106,
+      "step": 8880
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016115431910500386,
+      "loss": 0.8948,
+      "step": 8881
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001611131501628665,
+      "loss": 0.9665,
+      "step": 8882
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016107198293311196,
+      "loss": 0.9931,
+      "step": 8883
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001610308174175532,
+      "loss": 1.0169,
+      "step": 8884
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001609896536180029,
+      "loss": 0.9808,
+      "step": 8885
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016094849153627366,
+      "loss": 0.9525,
+      "step": 8886
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001609073311741781,
+      "loss": 0.9908,
+      "step": 8887
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016086617253352868,
+      "loss": 1.0208,
+      "step": 8888
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016082501561613788,
+      "loss": 1.0454,
+      "step": 8889
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016078386042381787,
+      "loss": 0.9657,
+      "step": 8890
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016074270695838117,
+      "loss": 1.0162,
+      "step": 8891
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016070155522163984,
+      "loss": 0.9495,
+      "step": 8892
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016066040521540599,
+      "loss": 0.9734,
+      "step": 8893
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016061925694149174,
+      "loss": 1.0362,
+      "step": 8894
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016057811040170897,
+      "loss": 0.9611,
+      "step": 8895
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001605369655978695,
+      "loss": 1.0047,
+      "step": 8896
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016049582253178536,
+      "loss": 0.9674,
+      "step": 8897
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016045468120526814,
+      "loss": 1.0378,
+      "step": 8898
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016041354162012948,
+      "loss": 0.9884,
+      "step": 8899
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016037240377818107,
+      "loss": 0.9651,
+      "step": 8900
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016033126768123435,
+      "loss": 0.982,
+      "step": 8901
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016029013333110066,
+      "loss": 0.975,
+      "step": 8902
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001602490007295915,
+      "loss": 1.0403,
+      "step": 8903
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016020786987851811,
+      "loss": 1.0111,
+      "step": 8904
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016016674077969163,
+      "loss": 1.0445,
+      "step": 8905
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016012561343492325,
+      "loss": 1.0786,
+      "step": 8906
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016008448784602397,
+      "loss": 0.971,
+      "step": 8907
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001600433640148047,
+      "loss": 1.0195,
+      "step": 8908
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016000224194307646,
+      "loss": 0.9752,
+      "step": 8909
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015996112163265,
+      "loss": 1.0208,
+      "step": 8910
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015992000308533597,
+      "loss": 1.0263,
+      "step": 8911
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015987888630294517,
+      "loss": 1.0129,
+      "step": 8912
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001598377712872881,
+      "loss": 0.9253,
+      "step": 8913
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015979665804017527,
+      "loss": 1.0403,
+      "step": 8914
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015975554656341703,
+      "loss": 1.0152,
+      "step": 8915
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001597144368588239,
+      "loss": 1.0194,
+      "step": 8916
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.000159673328928206,
+      "loss": 0.9806,
+      "step": 8917
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015963222277337353,
+      "loss": 0.9685,
+      "step": 8918
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015959111839613667,
+      "loss": 0.9177,
+      "step": 8919
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015955001579830543,
+      "loss": 1.0248,
+      "step": 8920
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015950891498168966,
+      "loss": 1.0771,
+      "step": 8921
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015946781594809942,
+      "loss": 1.0751,
+      "step": 8922
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015942671869934437,
+      "loss": 0.9556,
+      "step": 8923
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015938562323723426,
+      "loss": 1.0138,
+      "step": 8924
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015934452956357874,
+      "loss": 1.014,
+      "step": 8925
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015930343768018738,
+      "loss": 1.0253,
+      "step": 8926
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015926234758886957,
+      "loss": 1.0091,
+      "step": 8927
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015922125929143487,
+      "loss": 1.0542,
+      "step": 8928
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015918017278969252,
+      "loss": 0.9419,
+      "step": 8929
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015913908808545176,
+      "loss": 1.0138,
+      "step": 8930
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015909800518052177,
+      "loss": 1.0252,
+      "step": 8931
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001590569240767116,
+      "loss": 1.0192,
+      "step": 8932
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015901584477583024,
+      "loss": 1.0383,
+      "step": 8933
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015897476727968677,
+      "loss": 0.9927,
+      "step": 8934
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015893369159008987,
+      "loss": 0.9597,
+      "step": 8935
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015889261770884837,
+      "loss": 0.9746,
+      "step": 8936
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.000158851545637771,
+      "loss": 1.0333,
+      "step": 8937
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015881047537866634,
+      "loss": 0.9719,
+      "step": 8938
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001587694069333428,
+      "loss": 1.0355,
+      "step": 8939
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015872834030360904,
+      "loss": 1.0394,
+      "step": 8940
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015868727549127332,
+      "loss": 1.0373,
+      "step": 8941
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015864621249814394,
+      "loss": 0.9172,
+      "step": 8942
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015860515132602905,
+      "loss": 0.9887,
+      "step": 8943
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015856409197673686,
+      "loss": 0.9702,
+      "step": 8944
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015852303445207543,
+      "loss": 0.9611,
+      "step": 8945
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001584819787538526,
+      "loss": 0.9808,
+      "step": 8946
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015844092488387642,
+      "loss": 1.0755,
+      "step": 8947
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015839987284395465,
+      "loss": 0.9962,
+      "step": 8948
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015835882263589494,
+      "loss": 1.0456,
+      "step": 8949
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015831777426150506,
+      "loss": 0.9642,
+      "step": 8950
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015827672772259245,
+      "loss": 0.9528,
+      "step": 8951
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015823568302096463,
+      "loss": 0.9671,
+      "step": 8952
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015819464015842907,
+      "loss": 1.068,
+      "step": 8953
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015815359913679306,
+      "loss": 1.0559,
+      "step": 8954
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015811255995786382,
+      "loss": 0.9702,
+      "step": 8955
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001580715226234485,
+      "loss": 0.9917,
+      "step": 8956
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015803048713535427,
+      "loss": 1.0496,
+      "step": 8957
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015798945349538795,
+      "loss": 1.0218,
+      "step": 8958
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015794842170535663,
+      "loss": 1.0075,
+      "step": 8959
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015790739176706707,
+      "loss": 1.0259,
+      "step": 8960
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.000157866363682326,
+      "loss": 1.0158,
+      "step": 8961
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015782533745294024,
+      "loss": 1.0976,
+      "step": 8962
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015778431308071622,
+      "loss": 1.0335,
+      "step": 8963
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001577432905674604,
+      "loss": 1.0362,
+      "step": 8964
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001577022699149794,
+      "loss": 1.0011,
+      "step": 8965
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001576612511250795,
+      "loss": 0.9883,
+      "step": 8966
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001576202341995668,
+      "loss": 1.0234,
+      "step": 8967
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015757921914024775,
+      "loss": 1.0488,
+      "step": 8968
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015753820594892828,
+      "loss": 1.0342,
+      "step": 8969
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015749719462741443,
+      "loss": 0.9665,
+      "step": 8970
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001574561851775121,
+      "loss": 1.0683,
+      "step": 8971
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001574151776010272,
+      "loss": 1.0038,
+      "step": 8972
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001573741718997655,
+      "loss": 1.0081,
+      "step": 8973
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015733316807553264,
+      "loss": 1.0622,
+      "step": 8974
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015729216613013432,
+      "loss": 1.0655,
+      "step": 8975
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015725116606537598,
+      "loss": 1.0328,
+      "step": 8976
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015721016788306294,
+      "loss": 0.9879,
+      "step": 8977
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015716917158500085,
+      "loss": 0.9762,
+      "step": 8978
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015712817717299479,
+      "loss": 1.0601,
+      "step": 8979
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015708718464884985,
+      "loss": 0.998,
+      "step": 8980
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015704619401437145,
+      "loss": 0.9991,
+      "step": 8981
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015700520527136438,
+      "loss": 1.1273,
+      "step": 8982
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001569642184216335,
+      "loss": 1.0049,
+      "step": 8983
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015692323346698395,
+      "loss": 1.0025,
+      "step": 8984
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001568822504092203,
+      "loss": 1.0286,
+      "step": 8985
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015684126925014723,
+      "loss": 1.0282,
+      "step": 8986
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001568002899915695,
+      "loss": 0.9886,
+      "step": 8987
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015675931263529154,
+      "loss": 1.0313,
+      "step": 8988
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015671833718311766,
+      "loss": 1.0347,
+      "step": 8989
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001566773636368524,
+      "loss": 1.0376,
+      "step": 8990
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015663639199830004,
+      "loss": 1.0066,
+      "step": 8991
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015659542226926457,
+      "loss": 1.051,
+      "step": 8992
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015655445445155036,
+      "loss": 1.0531,
+      "step": 8993
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015651348854696124,
+      "loss": 0.9738,
+      "step": 8994
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001564725245573012,
+      "loss": 1.0523,
+      "step": 8995
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015643156248437397,
+      "loss": 1.0054,
+      "step": 8996
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015639060232998354,
+      "loss": 0.9477,
+      "step": 8997
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015634964409593348,
+      "loss": 1.0611,
+      "step": 8998
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015630868778402726,
+      "loss": 0.975,
+      "step": 8999
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015626773339606867,
+      "loss": 1.1019,
+      "step": 9000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015622678093386093,
+      "loss": 1.0001,
+      "step": 9001
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001561858303992074,
+      "loss": 1.0129,
+      "step": 9002
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015614488179391137,
+      "loss": 0.9715,
+      "step": 9003
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015610393511977607,
+      "loss": 1.0288,
+      "step": 9004
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015606299037860443,
+      "loss": 1.0651,
+      "step": 9005
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001560220475721997,
+      "loss": 0.9512,
+      "step": 9006
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015598110670236458,
+      "loss": 1.0193,
+      "step": 9007
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001559401677709019,
+      "loss": 1.0214,
+      "step": 9008
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015589923077961458,
+      "loss": 1.0251,
+      "step": 9009
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015585829573030516,
+      "loss": 0.9968,
+      "step": 9010
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015581736262477617,
+      "loss": 1.0243,
+      "step": 9011
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015577643146483022,
+      "loss": 1.0249,
+      "step": 9012
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015573550225226975,
+      "loss": 1.0502,
+      "step": 9013
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001556945749888968,
+      "loss": 1.0058,
+      "step": 9014
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015565364967651393,
+      "loss": 1.049,
+      "step": 9015
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015561272631692312,
+      "loss": 1.0373,
+      "step": 9016
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015557180491192635,
+      "loss": 0.9662,
+      "step": 9017
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015553088546332584,
+      "loss": 1.0327,
+      "step": 9018
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015548996797292333,
+      "loss": 1.0219,
+      "step": 9019
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015544905244252054,
+      "loss": 1.0289,
+      "step": 9020
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015540813887391936,
+      "loss": 1.0171,
+      "step": 9021
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015536722726892133,
+      "loss": 1.0113,
+      "step": 9022
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015532631762932801,
+      "loss": 1.066,
+      "step": 9023
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015528540995694075,
+      "loss": 0.8535,
+      "step": 9024
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015524450425356113,
+      "loss": 1.0393,
+      "step": 9025
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015520360052099033,
+      "loss": 0.9994,
+      "step": 9026
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001551626987610294,
+      "loss": 0.9697,
+      "step": 9027
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001551217989754797,
+      "loss": 1.0385,
+      "step": 9028
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015508090116614213,
+      "loss": 1.0119,
+      "step": 9029
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015504000533481756,
+      "loss": 0.8949,
+      "step": 9030
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015499911148330697,
+      "loss": 0.9513,
+      "step": 9031
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015495821961341113,
+      "loss": 1.0071,
+      "step": 9032
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001549173297269305,
+      "loss": 1.0099,
+      "step": 9033
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015487644182566595,
+      "loss": 1.0725,
+      "step": 9034
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015483555591141781,
+      "loss": 1.0147,
+      "step": 9035
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015479467198598642,
+      "loss": 1.0859,
+      "step": 9036
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001547537900511723,
+      "loss": 1.0492,
+      "step": 9037
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001547129101087757,
+      "loss": 1.0106,
+      "step": 9038
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015467203216059646,
+      "loss": 1.0894,
+      "step": 9039
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015463115620843499,
+      "loss": 0.9715,
+      "step": 9040
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015459028225409108,
+      "loss": 0.9856,
+      "step": 9041
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015454941029936458,
+      "loss": 0.9522,
+      "step": 9042
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015450854034605545,
+      "loss": 0.9903,
+      "step": 9043
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015446767239596335,
+      "loss": 1.0823,
+      "step": 9044
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015442680645088775,
+      "loss": 1.1091,
+      "step": 9045
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015438594251262835,
+      "loss": 0.9365,
+      "step": 9046
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015434508058298455,
+      "loss": 1.0166,
+      "step": 9047
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001543042206637557,
+      "loss": 1.0613,
+      "step": 9048
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015426336275674096,
+      "loss": 1.0541,
+      "step": 9049
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001542225068637397,
+      "loss": 1.0445,
+      "step": 9050
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.000154181652986551,
+      "loss": 0.9617,
+      "step": 9051
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001541408011269736,
+      "loss": 1.0658,
+      "step": 9052
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001540999512868067,
+      "loss": 1.0237,
+      "step": 9053
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.000154059103467849,
+      "loss": 0.9923,
+      "step": 9054
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015401825767189916,
+      "loss": 1.0725,
+      "step": 9055
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015397741390075603,
+      "loss": 0.999,
+      "step": 9056
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015393657215621807,
+      "loss": 0.9718,
+      "step": 9057
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001538957324400836,
+      "loss": 0.9762,
+      "step": 9058
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001538548947541512,
+      "loss": 0.9741,
+      "step": 9059
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001538140591002191,
+      "loss": 0.969,
+      "step": 9060
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001537732254800854,
+      "loss": 0.9943,
+      "step": 9061
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015373239389554835,
+      "loss": 1.0605,
+      "step": 9062
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015369156434840597,
+      "loss": 1.0074,
+      "step": 9063
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015365073684045602,
+      "loss": 1.0009,
+      "step": 9064
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001536099113734965,
+      "loss": 1.1218,
+      "step": 9065
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001535690879493251,
+      "loss": 0.9449,
+      "step": 9066
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015352826656973942,
+      "loss": 1.0228,
+      "step": 9067
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001534874472365372,
+      "loss": 0.9615,
+      "step": 9068
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001534466299515158,
+      "loss": 1.039,
+      "step": 9069
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015340581471647258,
+      "loss": 1.0467,
+      "step": 9070
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015336500153320494,
+      "loss": 1.0154,
+      "step": 9071
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015332419040351005,
+      "loss": 0.9482,
+      "step": 9072
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001532833813291849,
+      "loss": 1.0067,
+      "step": 9073
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015324257431202672,
+      "loss": 1.0461,
+      "step": 9074
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001532017693538324,
+      "loss": 1.0243,
+      "step": 9075
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015316096645639875,
+      "loss": 0.9772,
+      "step": 9076
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001531201656215224,
+      "loss": 1.044,
+      "step": 9077
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015307936685100024,
+      "loss": 0.9956,
+      "step": 9078
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015303857014662876,
+      "loss": 1.0421,
+      "step": 9079
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001529977755102043,
+      "loss": 1.0408,
+      "step": 9080
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015295698294352347,
+      "loss": 0.9075,
+      "step": 9081
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001529161924483825,
+      "loss": 1.0215,
+      "step": 9082
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001528754040265775,
+      "loss": 1.0423,
+      "step": 9083
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00015283461767990478,
+      "loss": 1.0572,
+      "step": 9084
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001527938334101602,
+      "loss": 0.9413,
+      "step": 9085
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00015275305121913968,
+      "loss": 1.0326,
+      "step": 9086
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00015271227110863922,
+      "loss": 1.0658,
+      "step": 9087
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00015267149308045448,
+      "loss": 1.028,
+      "step": 9088
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001526307171363811,
+      "loss": 0.9535,
+      "step": 9089
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00015258994327821472,
+      "loss": 1.0129,
+      "step": 9090
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00015254917150775077,
+      "loss": 1.0286,
+      "step": 9091
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00015250840182678456,
+      "loss": 1.0436,
+      "step": 9092
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00015246763423711158,
+      "loss": 0.9744,
+      "step": 9093
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00015242686874052688,
+      "loss": 1.0993,
+      "step": 9094
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00015238610533882557,
+      "loss": 1.0328,
+      "step": 9095
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00015234534403380274,
+      "loss": 0.9937,
+      "step": 9096
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00015230458482725327,
+      "loss": 1.0453,
+      "step": 9097
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00015226382772097195,
+      "loss": 1.0017,
+      "step": 9098
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001522230727167536,
+      "loss": 0.9057,
+      "step": 9099
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00015218231981639288,
+      "loss": 0.954,
+      "step": 9100
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00015214156902168424,
+      "loss": 0.9893,
+      "step": 9101
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00015210082033442226,
+      "loss": 0.9896,
+      "step": 9102
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00015206007375640125,
+      "loss": 1.059,
+      "step": 9103
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00015201932928941543,
+      "loss": 1.0143,
+      "step": 9104
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.000151978586935259,
+      "loss": 0.9773,
+      "step": 9105
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001519378466957262,
+      "loss": 0.9137,
+      "step": 9106
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001518971085726109,
+      "loss": 0.9869,
+      "step": 9107
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00015185637256770696,
+      "loss": 1.0478,
+      "step": 9108
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001518156386828083,
+      "loss": 1.0096,
+      "step": 9109
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001517749069197086,
+      "loss": 1.1102,
+      "step": 9110
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001517341772802014,
+      "loss": 0.9501,
+      "step": 9111
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00015169344976608036,
+      "loss": 0.9965,
+      "step": 9112
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00015165272437913887,
+      "loss": 1.0022,
+      "step": 9113
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001516120011211702,
+      "loss": 0.997,
+      "step": 9114
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00015157127999396774,
+      "loss": 0.9364,
+      "step": 9115
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00015153056099932456,
+      "loss": 0.9947,
+      "step": 9116
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001514898441390336,
+      "loss": 1.0066,
+      "step": 9117
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001514491294148881,
+      "loss": 0.9482,
+      "step": 9118
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00015140841682868074,
+      "loss": 1.0216,
+      "step": 9119
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00015136770638220432,
+      "loss": 0.9317,
+      "step": 9120
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001513269980772516,
+      "loss": 1.0242,
+      "step": 9121
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00015128629191561515,
+      "loss": 1.0497,
+      "step": 9122
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00015124558789908731,
+      "loss": 0.9931,
+      "step": 9123
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00015120488602946074,
+      "loss": 1.0226,
+      "step": 9124
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00015116418630852757,
+      "loss": 0.9753,
+      "step": 9125
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00015112348873808003,
+      "loss": 0.9717,
+      "step": 9126
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00015108279331991033,
+      "loss": 1.0499,
+      "step": 9127
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00015104210005581042,
+      "loss": 1.0065,
+      "step": 9128
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00015100140894757228,
+      "loss": 1.0228,
+      "step": 9129
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00015096071999698757,
+      "loss": 0.9611,
+      "step": 9130
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00015092003320584828,
+      "loss": 0.9792,
+      "step": 9131
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00015087934857594592,
+      "loss": 0.9614,
+      "step": 9132
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00015083866610907202,
+      "loss": 1.0252,
+      "step": 9133
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00015079798580701814,
+      "loss": 1.0397,
+      "step": 9134
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00015075730767157553,
+      "loss": 0.9635,
+      "step": 9135
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001507166317045354,
+      "loss": 0.9874,
+      "step": 9136
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00015067595790768909,
+      "loss": 1.0105,
+      "step": 9137
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00015063528628282761,
+      "loss": 0.9648,
+      "step": 9138
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00015059461683174184,
+      "loss": 0.9679,
+      "step": 9139
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001505539495562228,
+      "loss": 0.9716,
+      "step": 9140
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00015051328445806118,
+      "loss": 1.0005,
+      "step": 9141
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00015047262153904762,
+      "loss": 1.0431,
+      "step": 9142
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00015043196080097285,
+      "loss": 1.0228,
+      "step": 9143
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00015039130224562733,
+      "loss": 0.9995,
+      "step": 9144
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001503506458748014,
+      "loss": 0.9728,
+      "step": 9145
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00015030999169028542,
+      "loss": 1.0855,
+      "step": 9146
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00015026933969386957,
+      "loss": 0.9868,
+      "step": 9147
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00015022868988734387,
+      "loss": 1.0558,
+      "step": 9148
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001501880422724985,
+      "loss": 0.9801,
+      "step": 9149
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00015014739685112333,
+      "loss": 0.9281,
+      "step": 9150
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00015010675362500813,
+      "loss": 1.0105,
+      "step": 9151
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00015006611259594268,
+      "loss": 0.994,
+      "step": 9152
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00015002547376571657,
+      "loss": 1.0154,
+      "step": 9153
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014998483713611927,
+      "loss": 0.9937,
+      "step": 9154
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014994420270894037,
+      "loss": 0.952,
+      "step": 9155
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014990357048596912,
+      "loss": 0.9807,
+      "step": 9156
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014986294046899482,
+      "loss": 1.0032,
+      "step": 9157
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014982231265980644,
+      "loss": 1.0763,
+      "step": 9158
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014978168706019322,
+      "loss": 0.9483,
+      "step": 9159
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014974106367194402,
+      "loss": 1.0036,
+      "step": 9160
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014970044249684763,
+      "loss": 1.0202,
+      "step": 9161
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014965982353669298,
+      "loss": 0.9897,
+      "step": 9162
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001496192067932686,
+      "loss": 1.0008,
+      "step": 9163
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014957859226836303,
+      "loss": 1.0647,
+      "step": 9164
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014953797996376483,
+      "loss": 1.0536,
+      "step": 9165
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014949736988126232,
+      "loss": 1.0654,
+      "step": 9166
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014945676202264365,
+      "loss": 1.0542,
+      "step": 9167
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014941615638969718,
+      "loss": 1.0495,
+      "step": 9168
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014937555298421087,
+      "loss": 1.0087,
+      "step": 9169
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001493349518079727,
+      "loss": 1.015,
+      "step": 9170
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014929435286277062,
+      "loss": 1.0246,
+      "step": 9171
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001492537561503923,
+      "loss": 1.0397,
+      "step": 9172
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014921316167262538,
+      "loss": 0.9969,
+      "step": 9173
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014917256943125759,
+      "loss": 0.9423,
+      "step": 9174
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014913197942807634,
+      "loss": 1.0088,
+      "step": 9175
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014909139166486898,
+      "loss": 1.067,
+      "step": 9176
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014905080614342287,
+      "loss": 1.0059,
+      "step": 9177
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014901022286552512,
+      "loss": 0.988,
+      "step": 9178
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014896964183296277,
+      "loss": 0.986,
+      "step": 9179
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014892906304752297,
+      "loss": 0.9972,
+      "step": 9180
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001488884865109925,
+      "loss": 0.9894,
+      "step": 9181
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014884791222515814,
+      "loss": 1.0148,
+      "step": 9182
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001488073401918066,
+      "loss": 1.0259,
+      "step": 9183
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014876677041272447,
+      "loss": 0.9862,
+      "step": 9184
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014872620288969827,
+      "loss": 1.0231,
+      "step": 9185
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014868563762451427,
+      "loss": 0.9195,
+      "step": 9186
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014864507461895894,
+      "loss": 1.058,
+      "step": 9187
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014860451387481836,
+      "loss": 1.0452,
+      "step": 9188
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014856395539387864,
+      "loss": 0.9938,
+      "step": 9189
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0001485233991779258,
+      "loss": 1.0125,
+      "step": 9190
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014848284522874576,
+      "loss": 1.0042,
+      "step": 9191
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014844229354812414,
+      "loss": 1.0674,
+      "step": 9192
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014840174413784687,
+      "loss": 0.9531,
+      "step": 9193
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014836119699969942,
+      "loss": 1.0124,
+      "step": 9194
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014832065213546728,
+      "loss": 0.9283,
+      "step": 9195
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0001482801095469359,
+      "loss": 0.9447,
+      "step": 9196
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014823956923589054,
+      "loss": 0.9688,
+      "step": 9197
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014819903120411629,
+      "loss": 1.009,
+      "step": 9198
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014815849545339843,
+      "loss": 1.0655,
+      "step": 9199
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014811796198552189,
+      "loss": 1.0164,
+      "step": 9200
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014807743080227147,
+      "loss": 0.9944,
+      "step": 9201
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014803690190543208,
+      "loss": 1.0259,
+      "step": 9202
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014799637529678837,
+      "loss": 0.9886,
+      "step": 9203
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014795585097812484,
+      "loss": 1.0087,
+      "step": 9204
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0001479153289512261,
+      "loss": 1.0069,
+      "step": 9205
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014787480921787653,
+      "loss": 0.9891,
+      "step": 9206
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014783429177986034,
+      "loss": 1.0354,
+      "step": 9207
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014779377663896179,
+      "loss": 1.0482,
+      "step": 9208
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014775326379696494,
+      "loss": 1.0815,
+      "step": 9209
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014771275325565377,
+      "loss": 0.9808,
+      "step": 9210
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014767224501681207,
+      "loss": 0.9713,
+      "step": 9211
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014763173908222382,
+      "loss": 1.0677,
+      "step": 9212
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014759123545367254,
+      "loss": 0.9988,
+      "step": 9213
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014755073413294184,
+      "loss": 1.0043,
+      "step": 9214
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014751023512181525,
+      "loss": 1.0673,
+      "step": 9215
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014746973842207612,
+      "loss": 1.011,
+      "step": 9216
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014742924403550762,
+      "loss": 1.0647,
+      "step": 9217
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0001473887519638931,
+      "loss": 1.0743,
+      "step": 9218
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0001473482622090155,
+      "loss": 1.044,
+      "step": 9219
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014730777477265785,
+      "loss": 0.9441,
+      "step": 9220
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.000147267289656603,
+      "loss": 1.0054,
+      "step": 9221
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014722680686263373,
+      "loss": 1.0545,
+      "step": 9222
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014718632639253255,
+      "loss": 0.9332,
+      "step": 9223
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014714584824808226,
+      "loss": 1.023,
+      "step": 9224
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0001471053724310652,
+      "loss": 0.9566,
+      "step": 9225
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014706489894326366,
+      "loss": 1.0177,
+      "step": 9226
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014702442778645998,
+      "loss": 1.028,
+      "step": 9227
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0001469839589624363,
+      "loss": 0.9998,
+      "step": 9228
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014694349247297453,
+      "loss": 1.0316,
+      "step": 9229
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014690302831985686,
+      "loss": 1.0674,
+      "step": 9230
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014686256650486498,
+      "loss": 0.9993,
+      "step": 9231
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0001468221070297806,
+      "loss": 1.0745,
+      "step": 9232
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014678164989638542,
+      "loss": 1.0457,
+      "step": 9233
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014674119510646097,
+      "loss": 0.9768,
+      "step": 9234
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014670074266178858,
+      "loss": 1.0621,
+      "step": 9235
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014666029256414973,
+      "loss": 1.0409,
+      "step": 9236
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014661984481532554,
+      "loss": 1.0502,
+      "step": 9237
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014657939941709714,
+      "loss": 0.9812,
+      "step": 9238
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001465389563712455,
+      "loss": 1.0087,
+      "step": 9239
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014649851567955166,
+      "loss": 0.933,
+      "step": 9240
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014645807734379632,
+      "loss": 1.0126,
+      "step": 9241
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014641764136576012,
+      "loss": 1.0043,
+      "step": 9242
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014637720774722385,
+      "loss": 1.0342,
+      "step": 9243
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014633677648996792,
+      "loss": 0.9797,
+      "step": 9244
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014629634759577264,
+      "loss": 1.0116,
+      "step": 9245
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001462559210664184,
+      "loss": 0.9788,
+      "step": 9246
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014621549690368532,
+      "loss": 1.0038,
+      "step": 9247
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014617507510935343,
+      "loss": 1.0173,
+      "step": 9248
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014613465568520288,
+      "loss": 1.0018,
+      "step": 9249
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014609423863301343,
+      "loss": 0.9296,
+      "step": 9250
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001460538239545648,
+      "loss": 1.034,
+      "step": 9251
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014601341165163673,
+      "loss": 1.0622,
+      "step": 9252
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014597300172600877,
+      "loss": 1.0079,
+      "step": 9253
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014593259417946026,
+      "loss": 0.9474,
+      "step": 9254
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014589218901377073,
+      "loss": 1.0112,
+      "step": 9255
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014585178623071935,
+      "loss": 1.029,
+      "step": 9256
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014581138583208514,
+      "loss": 0.9494,
+      "step": 9257
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001457709878196473,
+      "loss": 1.012,
+      "step": 9258
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014573059219518466,
+      "loss": 0.981,
+      "step": 9259
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.000145690198960476,
+      "loss": 0.9956,
+      "step": 9260
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014564980811730018,
+      "loss": 0.9736,
+      "step": 9261
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014560941966743575,
+      "loss": 1.0488,
+      "step": 9262
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014556903361266117,
+      "loss": 1.0248,
+      "step": 9263
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014552864995475483,
+      "loss": 1.0404,
+      "step": 9264
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014548826869549515,
+      "loss": 0.9556,
+      "step": 9265
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001454478898366602,
+      "loss": 0.9962,
+      "step": 9266
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014540751338002804,
+      "loss": 0.9849,
+      "step": 9267
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014536713932737676,
+      "loss": 1.0981,
+      "step": 9268
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001453267676804842,
+      "loss": 1.0136,
+      "step": 9269
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014528639844112802,
+      "loss": 0.9743,
+      "step": 9270
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014524603161108606,
+      "loss": 1.0106,
+      "step": 9271
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014520566719213574,
+      "loss": 0.9569,
+      "step": 9272
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014516530518605448,
+      "loss": 0.9868,
+      "step": 9273
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014512494559461978,
+      "loss": 1.0155,
+      "step": 9274
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014508458841960877,
+      "loss": 1.0177,
+      "step": 9275
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014504423366279854,
+      "loss": 1.0336,
+      "step": 9276
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001450038813259662,
+      "loss": 1.0856,
+      "step": 9277
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014496353141088864,
+      "loss": 0.9499,
+      "step": 9278
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014492318391934257,
+      "loss": 1.0144,
+      "step": 9279
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014488283885310488,
+      "loss": 1.0279,
+      "step": 9280
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014484249621395204,
+      "loss": 0.9947,
+      "step": 9281
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014480215600366054,
+      "loss": 1.0129,
+      "step": 9282
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014476181822400684,
+      "loss": 0.9999,
+      "step": 9283
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014472148287676714,
+      "loss": 1.0149,
+      "step": 9284
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014468114996371752,
+      "loss": 1.0169,
+      "step": 9285
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014464081948663426,
+      "loss": 1.0722,
+      "step": 9286
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001446004914472932,
+      "loss": 0.9813,
+      "step": 9287
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014456016584747015,
+      "loss": 1.0243,
+      "step": 9288
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014451984268894092,
+      "loss": 1.0766,
+      "step": 9289
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001444795219734811,
+      "loss": 1.0113,
+      "step": 9290
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001444392037028662,
+      "loss": 0.9854,
+      "step": 9291
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014439888787887162,
+      "loss": 1.0488,
+      "step": 9292
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014435857450327276,
+      "loss": 1.0633,
+      "step": 9293
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014431826357784474,
+      "loss": 1.0228,
+      "step": 9294
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014427795510436265,
+      "loss": 0.9885,
+      "step": 9295
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014423764908460155,
+      "loss": 1.0215,
+      "step": 9296
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014419734552033626,
+      "loss": 1.0446,
+      "step": 9297
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014415704441334148,
+      "loss": 0.9879,
+      "step": 9298
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014411674576539203,
+      "loss": 0.9848,
+      "step": 9299
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014407644957826235,
+      "loss": 1.0709,
+      "step": 9300
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014403615585372687,
+      "loss": 0.949,
+      "step": 9301
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014399586459356,
+      "loss": 1.1105,
+      "step": 9302
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014395557579953593,
+      "loss": 1.0334,
+      "step": 9303
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001439152894734287,
+      "loss": 1.0088,
+      "step": 9304
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014387500561701248,
+      "loss": 1.0452,
+      "step": 9305
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014383472423206106,
+      "loss": 0.9632,
+      "step": 9306
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014379444532034824,
+      "loss": 1.013,
+      "step": 9307
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014375416888364774,
+      "loss": 0.9207,
+      "step": 9308
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014371389492373314,
+      "loss": 1.0201,
+      "step": 9309
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014367362344237775,
+      "loss": 0.9952,
+      "step": 9310
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014363335444135517,
+      "loss": 1.028,
+      "step": 9311
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014359308792243853,
+      "loss": 1.0399,
+      "step": 9312
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014355282388740087,
+      "loss": 0.9989,
+      "step": 9313
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001435125623380154,
+      "loss": 0.9512,
+      "step": 9314
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014347230327605494,
+      "loss": 0.9853,
+      "step": 9315
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001434320467032923,
+      "loss": 1.038,
+      "step": 9316
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001433917926215001,
+      "loss": 1.0049,
+      "step": 9317
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014335154103245113,
+      "loss": 1.0215,
+      "step": 9318
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001433112919379177,
+      "loss": 1.0015,
+      "step": 9319
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014327104533967222,
+      "loss": 0.9977,
+      "step": 9320
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.000143230801239487,
+      "loss": 1.0592,
+      "step": 9321
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014319055963913418,
+      "loss": 1.0022,
+      "step": 9322
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014315032054038564,
+      "loss": 1.0371,
+      "step": 9323
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014311008394501354,
+      "loss": 1.0286,
+      "step": 9324
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014306984985478963,
+      "loss": 0.9272,
+      "step": 9325
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001430296182714855,
+      "loss": 0.9424,
+      "step": 9326
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001429893891968729,
+      "loss": 0.9607,
+      "step": 9327
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014294916263272328,
+      "loss": 1.0518,
+      "step": 9328
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014290893858080788,
+      "loss": 0.983,
+      "step": 9329
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014286871704289816,
+      "loss": 1.0229,
+      "step": 9330
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001428284980207652,
+      "loss": 1.0088,
+      "step": 9331
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014278828151617997,
+      "loss": 1.0276,
+      "step": 9332
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014274806753091356,
+      "loss": 1.0276,
+      "step": 9333
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014270785606673668,
+      "loss": 0.9823,
+      "step": 9334
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014266764712542,
+      "loss": 0.9721,
+      "step": 9335
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014262744070873428,
+      "loss": 1.0161,
+      "step": 9336
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001425872368184499,
+      "loss": 1.0296,
+      "step": 9337
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014254703545633722,
+      "loss": 0.9784,
+      "step": 9338
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001425068366241666,
+      "loss": 1.0045,
+      "step": 9339
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014246664032370815,
+      "loss": 1.0118,
+      "step": 9340
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014242644655673177,
+      "loss": 0.9661,
+      "step": 9341
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014238625532500768,
+      "loss": 1.0597,
+      "step": 9342
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014234606663030553,
+      "loss": 0.9213,
+      "step": 9343
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014230588047439504,
+      "loss": 0.9732,
+      "step": 9344
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001422656968590458,
+      "loss": 1.0158,
+      "step": 9345
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001422255157860273,
+      "loss": 1.0047,
+      "step": 9346
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014218533725710898,
+      "loss": 1.0106,
+      "step": 9347
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014214516127405996,
+      "loss": 1.0143,
+      "step": 9348
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014210498783864955,
+      "loss": 1.0096,
+      "step": 9349
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014206481695264674,
+      "loss": 0.9949,
+      "step": 9350
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001420246486178204,
+      "loss": 0.9769,
+      "step": 9351
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001419844828359394,
+      "loss": 1.0551,
+      "step": 9352
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001419443196087724,
+      "loss": 0.9717,
+      "step": 9353
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014190415893808797,
+      "loss": 1.0202,
+      "step": 9354
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014186400082565469,
+      "loss": 0.973,
+      "step": 9355
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014182384527324085,
+      "loss": 1.0003,
+      "step": 9356
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014178369228261465,
+      "loss": 1.0226,
+      "step": 9357
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014174354185554432,
+      "loss": 1.0919,
+      "step": 9358
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014170339399379787,
+      "loss": 1.0298,
+      "step": 9359
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001416632486991431,
+      "loss": 0.9947,
+      "step": 9360
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014162310597334797,
+      "loss": 0.9814,
+      "step": 9361
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001415829658181801,
+      "loss": 0.9229,
+      "step": 9362
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014154282823540702,
+      "loss": 1.0049,
+      "step": 9363
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014150269322679625,
+      "loss": 1.043,
+      "step": 9364
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014146256079411513,
+      "loss": 1.0088,
+      "step": 9365
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014142243093913078,
+      "loss": 0.9985,
+      "step": 9366
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001413823036636105,
+      "loss": 1.0239,
+      "step": 9367
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001413421789693212,
+      "loss": 1.0752,
+      "step": 9368
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014130205685802976,
+      "loss": 0.9399,
+      "step": 9369
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014126193733150294,
+      "loss": 1.1506,
+      "step": 9370
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001412218203915075,
+      "loss": 0.9785,
+      "step": 9371
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001411817060398099,
+      "loss": 1.0176,
+      "step": 9372
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014114159427817654,
+      "loss": 1.0417,
+      "step": 9373
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001411014851083739,
+      "loss": 0.9046,
+      "step": 9374
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014106137853216806,
+      "loss": 0.9949,
+      "step": 9375
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014102127455132516,
+      "loss": 0.9524,
+      "step": 9376
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014098117316761118,
+      "loss": 0.9422,
+      "step": 9377
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014094107438279195,
+      "loss": 0.9425,
+      "step": 9378
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001409009781986332,
+      "loss": 1.0551,
+      "step": 9379
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014086088461690065,
+      "loss": 0.9895,
+      "step": 9380
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001408207936393598,
+      "loss": 1.0079,
+      "step": 9381
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.000140780705267776,
+      "loss": 1.0123,
+      "step": 9382
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014074061950391462,
+      "loss": 0.9783,
+      "step": 9383
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001407005363495408,
+      "loss": 0.9758,
+      "step": 9384
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001406604558064195,
+      "loss": 1.0125,
+      "step": 9385
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014062037787631584,
+      "loss": 0.9926,
+      "step": 9386
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014058030256099458,
+      "loss": 0.9733,
+      "step": 9387
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014054022986222038,
+      "loss": 0.9897,
+      "step": 9388
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.000140500159781758,
+      "loss": 0.8851,
+      "step": 9389
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001404600923213718,
+      "loss": 1.0308,
+      "step": 9390
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00014042002748282606,
+      "loss": 0.9385,
+      "step": 9391
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00014037996526788523,
+      "loss": 0.9253,
+      "step": 9392
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00014033990567831336,
+      "loss": 1.0171,
+      "step": 9393
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001402998487158745,
+      "loss": 0.988,
+      "step": 9394
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00014025979438233254,
+      "loss": 1.0267,
+      "step": 9395
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00014021974267945127,
+      "loss": 1.0221,
+      "step": 9396
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001401796936089944,
+      "loss": 0.9978,
+      "step": 9397
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00014013964717272538,
+      "loss": 1.0975,
+      "step": 9398
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00014009960337240778,
+      "loss": 1.0365,
+      "step": 9399
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00014005956220980496,
+      "loss": 1.025,
+      "step": 9400
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00014001952368667997,
+      "loss": 0.9839,
+      "step": 9401
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00013997948780479605,
+      "loss": 0.958,
+      "step": 9402
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00013993945456591616,
+      "loss": 0.9306,
+      "step": 9403
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.000139899423971803,
+      "loss": 1.0503,
+      "step": 9404
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00013985939602421957,
+      "loss": 1.0377,
+      "step": 9405
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00013981937072492836,
+      "loss": 1.0275,
+      "step": 9406
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00013977934807569186,
+      "loss": 1.0102,
+      "step": 9407
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00013973932807827258,
+      "loss": 1.0212,
+      "step": 9408
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00013969931073443275,
+      "loss": 1.0237,
+      "step": 9409
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00013965929604593436,
+      "loss": 0.997,
+      "step": 9410
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00013961928401453978,
+      "loss": 1.0497,
+      "step": 9411
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00013957927464201077,
+      "loss": 1.0412,
+      "step": 9412
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00013953926793010908,
+      "loss": 1.027,
+      "step": 9413
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00013949926388059653,
+      "loss": 0.9762,
+      "step": 9414
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00013945926249523465,
+      "loss": 0.9919,
+      "step": 9415
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001394192637757848,
+      "loss": 1.0012,
+      "step": 9416
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00013937926772400854,
+      "loss": 1.0205,
+      "step": 9417
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00013933927434166694,
+      "loss": 1.0037,
+      "step": 9418
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001392992836305211,
+      "loss": 1.0209,
+      "step": 9419
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00013925929559233214,
+      "loss": 1.0859,
+      "step": 9420
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001392193102288608,
+      "loss": 1.0658,
+      "step": 9421
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00013917932754186783,
+      "loss": 1.0184,
+      "step": 9422
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00013913934753311403,
+      "loss": 0.9785,
+      "step": 9423
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00013909937020435982,
+      "loss": 0.9456,
+      "step": 9424
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001390593955573656,
+      "loss": 1.0223,
+      "step": 9425
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00013901942359389156,
+      "loss": 0.9639,
+      "step": 9426
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00013897945431569806,
+      "loss": 1.047,
+      "step": 9427
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.000138939487724545,
+      "loss": 0.985,
+      "step": 9428
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001388995238221923,
+      "loss": 0.973,
+      "step": 9429
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00013885956261039988,
+      "loss": 0.9339,
+      "step": 9430
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00013881960409092741,
+      "loss": 0.934,
+      "step": 9431
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00013877964826553441,
+      "loss": 1.0508,
+      "step": 9432
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00013873969513598035,
+      "loss": 1.0513,
+      "step": 9433
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00013869974470402457,
+      "loss": 0.9899,
+      "step": 9434
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00013865979697142626,
+      "loss": 1.0407,
+      "step": 9435
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001386198519399446,
+      "loss": 0.9848,
+      "step": 9436
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00013857990961133854,
+      "loss": 1.0308,
+      "step": 9437
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001385399699873669,
+      "loss": 1.0308,
+      "step": 9438
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00013850003306978845,
+      "loss": 0.9882,
+      "step": 9439
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001384600988603618,
+      "loss": 0.9953,
+      "step": 9440
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001384201673608454,
+      "loss": 0.9582,
+      "step": 9441
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013838023857299777,
+      "loss": 1.0247,
+      "step": 9442
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001383403124985771,
+      "loss": 0.9083,
+      "step": 9443
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013830038913934152,
+      "loss": 1.0007,
+      "step": 9444
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013826046849704907,
+      "loss": 1.0668,
+      "step": 9445
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013822055057345766,
+      "loss": 0.969,
+      "step": 9446
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.000138180635370325,
+      "loss": 0.9636,
+      "step": 9447
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013814072288940893,
+      "loss": 1.0129,
+      "step": 9448
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013810081313246685,
+      "loss": 1.0271,
+      "step": 9449
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013806090610125627,
+      "loss": 1.0216,
+      "step": 9450
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013802100179753442,
+      "loss": 0.9309,
+      "step": 9451
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013798110022305855,
+      "loss": 0.9847,
+      "step": 9452
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013794120137958568,
+      "loss": 1.0866,
+      "step": 9453
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013790130526887271,
+      "loss": 1.0241,
+      "step": 9454
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013786141189267665,
+      "loss": 1.0143,
+      "step": 9455
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013782152125275406,
+      "loss": 1.0245,
+      "step": 9456
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013778163335086152,
+      "loss": 0.9765,
+      "step": 9457
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013774174818875556,
+      "loss": 0.9693,
+      "step": 9458
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013770186576819252,
+      "loss": 0.9666,
+      "step": 9459
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013766198609092847,
+      "loss": 1.0476,
+      "step": 9460
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013762210915871972,
+      "loss": 1.0134,
+      "step": 9461
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013758223497332218,
+      "loss": 1.0566,
+      "step": 9462
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013754236353649167,
+      "loss": 0.9712,
+      "step": 9463
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.000137502494849984,
+      "loss": 0.983,
+      "step": 9464
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013746262891555473,
+      "loss": 0.9438,
+      "step": 9465
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001374227657349593,
+      "loss": 0.928,
+      "step": 9466
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013738290530995322,
+      "loss": 0.9918,
+      "step": 9467
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013734304764229167,
+      "loss": 0.9331,
+      "step": 9468
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013730319273372977,
+      "loss": 0.9667,
+      "step": 9469
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013726334058602262,
+      "loss": 1.0182,
+      "step": 9470
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013722349120092502,
+      "loss": 1.0384,
+      "step": 9471
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001371836445801917,
+      "loss": 1.0936,
+      "step": 9472
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013714380072557745,
+      "loss": 0.978,
+      "step": 9473
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013710395963883668,
+      "loss": 0.9949,
+      "step": 9474
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013706412132172383,
+      "loss": 0.9912,
+      "step": 9475
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013702428577599322,
+      "loss": 1.027,
+      "step": 9476
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013698445300339891,
+      "loss": 1.0442,
+      "step": 9477
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013694462300569506,
+      "loss": 1.0069,
+      "step": 9478
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013690479578463543,
+      "loss": 1.023,
+      "step": 9479
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013686497134197395,
+      "loss": 0.9365,
+      "step": 9480
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013682514967946428,
+      "loss": 0.9802,
+      "step": 9481
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013678533079885985,
+      "loss": 0.8947,
+      "step": 9482
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001367455147019142,
+      "loss": 1.0863,
+      "step": 9483
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001367057013903806,
+      "loss": 1.0555,
+      "step": 9484
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013666589086601215,
+      "loss": 0.9974,
+      "step": 9485
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013662608313056207,
+      "loss": 1.0161,
+      "step": 9486
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013658627818578323,
+      "loss": 1.034,
+      "step": 9487
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013654647603342834,
+      "loss": 1.0423,
+      "step": 9488
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001365066766752502,
+      "loss": 1.0778,
+      "step": 9489
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013646688011300138,
+      "loss": 1.0216,
+      "step": 9490
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013642708634843418,
+      "loss": 0.9674,
+      "step": 9491
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013638729538330112,
+      "loss": 0.9824,
+      "step": 9492
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013634750721935426,
+      "loss": 1.0224,
+      "step": 9493
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00013630772185834568,
+      "loss": 0.9514,
+      "step": 9494
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001362679393020274,
+      "loss": 1.0368,
+      "step": 9495
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001362281595521512,
+      "loss": 1.0708,
+      "step": 9496
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001361883826104687,
+      "loss": 0.9953,
+      "step": 9497
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001361486084787317,
+      "loss": 0.9734,
+      "step": 9498
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00013610883715869148,
+      "loss": 0.9264,
+      "step": 9499
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00013606906865209933,
+      "loss": 1.0956,
+      "step": 9500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00013602930296070662,
+      "loss": 1.0453,
+      "step": 9501
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00013598954008626436,
+      "loss": 1.0158,
+      "step": 9502
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00013594978003052344,
+      "loss": 1.0202,
+      "step": 9503
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00013591002279523474,
+      "loss": 0.9796,
+      "step": 9504
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00013587026838214904,
+      "loss": 0.9807,
+      "step": 9505
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001358305167930168,
+      "loss": 0.9194,
+      "step": 9506
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00013579076802958858,
+      "loss": 0.9891,
+      "step": 9507
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00013575102209361473,
+      "loss": 0.9964,
+      "step": 9508
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001357112789868454,
+      "loss": 1.0107,
+      "step": 9509
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001356715387110306,
+      "loss": 1.0538,
+      "step": 9510
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001356318012679205,
+      "loss": 0.9804,
+      "step": 9511
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001355920666592648,
+      "loss": 0.9596,
+      "step": 9512
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00013555233488681326,
+      "loss": 1.0457,
+      "step": 9513
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00013551260595231544,
+      "loss": 1.0591,
+      "step": 9514
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00013547287985752083,
+      "loss": 1.0194,
+      "step": 9515
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00013543315660417867,
+      "loss": 1.009,
+      "step": 9516
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00013539343619403836,
+      "loss": 0.9548,
+      "step": 9517
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001353537186288489,
+      "loss": 0.9965,
+      "step": 9518
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00013531400391035916,
+      "loss": 1.037,
+      "step": 9519
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00013527429204031812,
+      "loss": 1.0272,
+      "step": 9520
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00013523458302047443,
+      "loss": 0.9593,
+      "step": 9521
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00013519487685257663,
+      "loss": 0.9758,
+      "step": 9522
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001351551735383733,
+      "loss": 1.0548,
+      "step": 9523
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00013511547307961264,
+      "loss": 0.9606,
+      "step": 9524
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00013507577547804296,
+      "loss": 1.0188,
+      "step": 9525
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00013503608073541233,
+      "loss": 0.9972,
+      "step": 9526
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001349963888534687,
+      "loss": 0.9098,
+      "step": 9527
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001349566998339598,
+      "loss": 1.0139,
+      "step": 9528
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00013491701367863355,
+      "loss": 1.0162,
+      "step": 9529
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001348773303892374,
+      "loss": 1.0142,
+      "step": 9530
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00013483764996751878,
+      "loss": 0.9861,
+      "step": 9531
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00013479797241522503,
+      "loss": 1.0188,
+      "step": 9532
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001347582977341034,
+      "loss": 1.0262,
+      "step": 9533
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00013471862592590096,
+      "loss": 0.9718,
+      "step": 9534
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00013467895699236456,
+      "loss": 1.0327,
+      "step": 9535
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00013463929093524116,
+      "loss": 1.0655,
+      "step": 9536
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00013459962775627738,
+      "loss": 1.0183,
+      "step": 9537
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001345599674572198,
+      "loss": 1.0207,
+      "step": 9538
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001345203100398149,
+      "loss": 0.9639,
+      "step": 9539
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00013448065550580895,
+      "loss": 1.0318,
+      "step": 9540
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00013444100385694805,
+      "loss": 1.0291,
+      "step": 9541
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00013440135509497846,
+      "loss": 1.0461,
+      "step": 9542
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.000134361709221646,
+      "loss": 1.0792,
+      "step": 9543
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00013432206623869647,
+      "loss": 1.024,
+      "step": 9544
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00013428242614787559,
+      "loss": 1.0191,
+      "step": 9545
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001342427889509289,
+      "loss": 1.0492,
+      "step": 9546
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00013420315464960175,
+      "loss": 1.0273,
+      "step": 9547
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00013416352324563958,
+      "loss": 1.0495,
+      "step": 9548
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00013412389474078745,
+      "loss": 0.9891,
+      "step": 9549
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001340842691367904,
+      "loss": 1.0013,
+      "step": 9550
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001340446464353934,
+      "loss": 1.0612,
+      "step": 9551
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00013400502663834125,
+      "loss": 1.0138,
+      "step": 9552
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001339654097473785,
+      "loss": 1.0159,
+      "step": 9553
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00013392579576424979,
+      "loss": 1.0739,
+      "step": 9554
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001338861846906995,
+      "loss": 0.9904,
+      "step": 9555
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00013384657652847185,
+      "loss": 0.9659,
+      "step": 9556
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001338069712793111,
+      "loss": 0.9895,
+      "step": 9557
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001337673689449611,
+      "loss": 1.0519,
+      "step": 9558
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001337277695271659,
+      "loss": 1.0362,
+      "step": 9559
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001336881730276691,
+      "loss": 1.0841,
+      "step": 9560
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001336485794482145,
+      "loss": 1.0109,
+      "step": 9561
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00013360898879054552,
+      "loss": 0.9349,
+      "step": 9562
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00013356940105640553,
+      "loss": 1.0222,
+      "step": 9563
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001335298162475378,
+      "loss": 0.9961,
+      "step": 9564
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00013349023436568544,
+      "loss": 0.9813,
+      "step": 9565
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00013345065541259135,
+      "loss": 1.0686,
+      "step": 9566
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00013341107938999856,
+      "loss": 0.9713,
+      "step": 9567
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00013337150629964971,
+      "loss": 0.9866,
+      "step": 9568
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001333319361432874,
+      "loss": 1.0028,
+      "step": 9569
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00013329236892265414,
+      "loss": 0.9731,
+      "step": 9570
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001332528046394922,
+      "loss": 0.9725,
+      "step": 9571
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00013321324329554379,
+      "loss": 1.0178,
+      "step": 9572
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00013317368489255111,
+      "loss": 1.015,
+      "step": 9573
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00013313412943225607,
+      "loss": 0.9853,
+      "step": 9574
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001330945769164004,
+      "loss": 0.9743,
+      "step": 9575
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00013305502734672594,
+      "loss": 0.9846,
+      "step": 9576
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00013301548072497415,
+      "loss": 0.9709,
+      "step": 9577
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00013297593705288642,
+      "loss": 1.0098,
+      "step": 9578
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001329363963322042,
+      "loss": 1.0593,
+      "step": 9579
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00013289685856466866,
+      "loss": 0.9509,
+      "step": 9580
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00013285732375202067,
+      "loss": 0.9691,
+      "step": 9581
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00013281779189600135,
+      "loss": 1.0345,
+      "step": 9582
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00013277826299835138,
+      "loss": 0.9837,
+      "step": 9583
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00013273873706081145,
+      "loss": 0.9783,
+      "step": 9584
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00013269921408512196,
+      "loss": 0.9796,
+      "step": 9585
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001326596940730235,
+      "loss": 1.003,
+      "step": 9586
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00013262017702625627,
+      "loss": 0.9771,
+      "step": 9587
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00013258066294656034,
+      "loss": 1.0093,
+      "step": 9588
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00013254115183567575,
+      "loss": 0.9833,
+      "step": 9589
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00013250164369534243,
+      "loss": 0.9658,
+      "step": 9590
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00013246213852729995,
+      "loss": 1.0142,
+      "step": 9591
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00013242263633328813,
+      "loss": 0.957,
+      "step": 9592
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00013238313711504636,
+      "loss": 0.9326,
+      "step": 9593
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00013234364087431395,
+      "loss": 0.974,
+      "step": 9594
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001323041476128302,
+      "loss": 1.0031,
+      "step": 9595
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00013226465733233412,
+      "loss": 1.0183,
+      "step": 9596
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00013222517003456464,
+      "loss": 1.0843,
+      "step": 9597
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00013218568572126072,
+      "loss": 0.9276,
+      "step": 9598
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00013214620439416094,
+      "loss": 0.9855,
+      "step": 9599
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00013210672605500388,
+      "loss": 1.0374,
+      "step": 9600
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00013206725070552801,
+      "loss": 0.9976,
+      "step": 9601
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00013202777834747163,
+      "loss": 1.0589,
+      "step": 9602
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00013198830898257275,
+      "loss": 0.9677,
+      "step": 9603
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00013194884261256962,
+      "loss": 1.0247,
+      "step": 9604
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00013190937923920006,
+      "loss": 1.016,
+      "step": 9605
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001318699188642018,
+      "loss": 1.0598,
+      "step": 9606
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00013183046148931252,
+      "loss": 1.0215,
+      "step": 9607
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00013179100711626976,
+      "loss": 0.9532,
+      "step": 9608
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001317515557468107,
+      "loss": 1.0212,
+      "step": 9609
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00013171210738267285,
+      "loss": 1.0079,
+      "step": 9610
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001316726620255932,
+      "loss": 1.045,
+      "step": 9611
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00013163321967730874,
+      "loss": 0.9285,
+      "step": 9612
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00013159378033955624,
+      "loss": 0.9691,
+      "step": 9613
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00013155434401407253,
+      "loss": 1.0514,
+      "step": 9614
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001315149107025941,
+      "loss": 1.0535,
+      "step": 9615
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00013147548040685735,
+      "loss": 1.073,
+      "step": 9616
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00013143605312859874,
+      "loss": 1.0812,
+      "step": 9617
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00013139662886955443,
+      "loss": 0.9936,
+      "step": 9618
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00013135720763146033,
+      "loss": 0.9639,
+      "step": 9619
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00013131778941605254,
+      "loss": 0.9806,
+      "step": 9620
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00013127837422506668,
+      "loss": 1.0416,
+      "step": 9621
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00013123896206023843,
+      "loss": 1.0225,
+      "step": 9622
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001311995529233034,
+      "loss": 0.9423,
+      "step": 9623
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001311601468159969,
+      "loss": 1.0073,
+      "step": 9624
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00013112074374005417,
+      "loss": 1.0004,
+      "step": 9625
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00013108134369721038,
+      "loss": 1.023,
+      "step": 9626
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00013104194668920046,
+      "loss": 1.0366,
+      "step": 9627
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00013100255271775922,
+      "loss": 0.9499,
+      "step": 9628
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00013096316178462148,
+      "loss": 1.1174,
+      "step": 9629
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001309237738915218,
+      "loss": 1.0438,
+      "step": 9630
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00013088438904019454,
+      "loss": 1.0454,
+      "step": 9631
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001308450072323741,
+      "loss": 0.9398,
+      "step": 9632
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00013080562846979463,
+      "loss": 0.985,
+      "step": 9633
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00013076625275419007,
+      "loss": 1.0152,
+      "step": 9634
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00013072688008729455,
+      "loss": 0.9581,
+      "step": 9635
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001306875104708417,
+      "loss": 1.0152,
+      "step": 9636
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00013064814390656516,
+      "loss": 0.9743,
+      "step": 9637
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00013060878039619845,
+      "loss": 1.0593,
+      "step": 9638
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.000130569419941475,
+      "loss": 1.0118,
+      "step": 9639
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.000130530062544128,
+      "loss": 1.0075,
+      "step": 9640
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00013049070820589048,
+      "loss": 0.9943,
+      "step": 9641
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00013045135692849556,
+      "loss": 1.0844,
+      "step": 9642
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00013041200871367602,
+      "loss": 0.9804,
+      "step": 9643
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00013037266356316448,
+      "loss": 0.9739,
+      "step": 9644
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00013033332147869363,
+      "loss": 0.9342,
+      "step": 9645
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001302939824619958,
+      "loss": 0.9475,
+      "step": 9646
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00013025464651480326,
+      "loss": 1.0492,
+      "step": 9647
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001302153136388483,
+      "loss": 0.9931,
+      "step": 9648
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00013017598383586285,
+      "loss": 0.9389,
+      "step": 9649
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001301366571075788,
+      "loss": 1.0857,
+      "step": 9650
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00013009733345572795,
+      "loss": 0.9652,
+      "step": 9651
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001300580128820419,
+      "loss": 0.9782,
+      "step": 9652
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00013001869538825205,
+      "loss": 0.9982,
+      "step": 9653
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012997938097608985,
+      "loss": 0.8958,
+      "step": 9654
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012994006964728652,
+      "loss": 0.9794,
+      "step": 9655
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012990076140357304,
+      "loss": 1.018,
+      "step": 9656
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012986145624668046,
+      "loss": 0.939,
+      "step": 9657
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001298221541783395,
+      "loss": 1.0434,
+      "step": 9658
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001297828552002808,
+      "loss": 1.0058,
+      "step": 9659
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012974355931423502,
+      "loss": 1.0391,
+      "step": 9660
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012970426652193244,
+      "loss": 1.0117,
+      "step": 9661
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012966497682510335,
+      "loss": 1.0452,
+      "step": 9662
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001296256902254779,
+      "loss": 1.0253,
+      "step": 9663
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012958640672478607,
+      "loss": 1.0901,
+      "step": 9664
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012954712632475766,
+      "loss": 1.0636,
+      "step": 9665
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012950784902712238,
+      "loss": 1.0817,
+      "step": 9666
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012946857483360987,
+      "loss": 1.0473,
+      "step": 9667
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012942930374594959,
+      "loss": 1.0337,
+      "step": 9668
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012939003576587074,
+      "loss": 0.9526,
+      "step": 9669
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012935077089510258,
+      "loss": 1.0195,
+      "step": 9670
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012931150913537411,
+      "loss": 0.9594,
+      "step": 9671
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012927225048841413,
+      "loss": 0.9417,
+      "step": 9672
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012923299495595157,
+      "loss": 1.0228,
+      "step": 9673
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012919374253971493,
+      "loss": 1.0113,
+      "step": 9674
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012915449324143267,
+      "loss": 0.9819,
+      "step": 9675
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012911524706283323,
+      "loss": 0.9587,
+      "step": 9676
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012907600400564477,
+      "loss": 0.9559,
+      "step": 9677
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012903676407159526,
+      "loss": 1.0519,
+      "step": 9678
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012899752726241282,
+      "loss": 1.0577,
+      "step": 9679
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012895829357982517,
+      "loss": 0.9645,
+      "step": 9680
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012891906302555986,
+      "loss": 1.0477,
+      "step": 9681
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012887983560134456,
+      "loss": 1.0371,
+      "step": 9682
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001288406113089066,
+      "loss": 0.9537,
+      "step": 9683
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012880139014997307,
+      "loss": 0.9353,
+      "step": 9684
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001287621721262713,
+      "loss": 1.0257,
+      "step": 9685
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012872295723952818,
+      "loss": 0.9438,
+      "step": 9686
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001286837454914705,
+      "loss": 1.0057,
+      "step": 9687
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.000128644536883825,
+      "loss": 0.9414,
+      "step": 9688
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001286053314183182,
+      "loss": 0.9609,
+      "step": 9689
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012856612909667643,
+      "loss": 0.9366,
+      "step": 9690
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001285269299206261,
+      "loss": 1.0637,
+      "step": 9691
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012848773389189337,
+      "loss": 0.9933,
+      "step": 9692
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012844854101220414,
+      "loss": 0.9249,
+      "step": 9693
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012840935128328424,
+      "loss": 1.0057,
+      "step": 9694
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012837016470685947,
+      "loss": 0.9878,
+      "step": 9695
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001283309812846554,
+      "loss": 0.9905,
+      "step": 9696
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012829180101839739,
+      "loss": 0.9741,
+      "step": 9697
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001282526239098109,
+      "loss": 1.017,
+      "step": 9698
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.000128213449960621,
+      "loss": 0.9464,
+      "step": 9699
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012817427917255264,
+      "loss": 0.993,
+      "step": 9700
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001281351115473309,
+      "loss": 1.0315,
+      "step": 9701
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012809594708668035,
+      "loss": 0.8958,
+      "step": 9702
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001280567857923256,
+      "loss": 1.0446,
+      "step": 9703
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012801762766599128,
+      "loss": 1.06,
+      "step": 9704
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012797847270940162,
+      "loss": 1.0008,
+      "step": 9705
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012793932092428076,
+      "loss": 0.9868,
+      "step": 9706
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012790017231235283,
+      "loss": 0.9236,
+      "step": 9707
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012786102687534172,
+      "loss": 0.9734,
+      "step": 9708
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012782188461497109,
+      "loss": 0.9823,
+      "step": 9709
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012778274553296477,
+      "loss": 0.9537,
+      "step": 9710
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012774360963104612,
+      "loss": 1.0663,
+      "step": 9711
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012770447691093847,
+      "loss": 1.0121,
+      "step": 9712
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012766534737436514,
+      "loss": 1.0483,
+      "step": 9713
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012762622102304913,
+      "loss": 0.9608,
+      "step": 9714
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001275870978587133,
+      "loss": 1.0078,
+      "step": 9715
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001275479778830806,
+      "loss": 0.9788,
+      "step": 9716
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001275088610978736,
+      "loss": 0.9286,
+      "step": 9717
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012746974750481478,
+      "loss": 1.0905,
+      "step": 9718
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001274306371056265,
+      "loss": 1.0277,
+      "step": 9719
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012739152990203103,
+      "loss": 1.0395,
+      "step": 9720
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012735242589575045,
+      "loss": 1.0035,
+      "step": 9721
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012731332508850664,
+      "loss": 1.0129,
+      "step": 9722
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001272742274820215,
+      "loss": 1.0467,
+      "step": 9723
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012723513307801672,
+      "loss": 0.9779,
+      "step": 9724
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012719604187821367,
+      "loss": 0.9815,
+      "step": 9725
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001271569538843339,
+      "loss": 0.9967,
+      "step": 9726
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012711786909809852,
+      "loss": 1.017,
+      "step": 9727
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012707878752122863,
+      "loss": 0.9656,
+      "step": 9728
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001270397091554453,
+      "loss": 0.9545,
+      "step": 9729
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012700063400246928,
+      "loss": 1.0089,
+      "step": 9730
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012696156206402123,
+      "loss": 0.9818,
+      "step": 9731
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012692249334182173,
+      "loss": 1.0097,
+      "step": 9732
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012688342783759112,
+      "loss": 0.9779,
+      "step": 9733
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012684436555304962,
+      "loss": 0.9912,
+      "step": 9734
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012680530648991747,
+      "loss": 0.9011,
+      "step": 9735
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012676625064991456,
+      "loss": 0.9573,
+      "step": 9736
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012672719803476066,
+      "loss": 1.0824,
+      "step": 9737
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012668814864617555,
+      "loss": 0.9518,
+      "step": 9738
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012664910248587872,
+      "loss": 0.9734,
+      "step": 9739
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001266100595555895,
+      "loss": 0.9568,
+      "step": 9740
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012657101985702728,
+      "loss": 1.0133,
+      "step": 9741
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012653198339191113,
+      "loss": 1.0093,
+      "step": 9742
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012649295016195996,
+      "loss": 0.9722,
+      "step": 9743
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012645392016889266,
+      "loss": 0.9814,
+      "step": 9744
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012641489341442793,
+      "loss": 1.0108,
+      "step": 9745
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012637586990028426,
+      "loss": 1.0256,
+      "step": 9746
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012633684962818,
+      "loss": 1.0536,
+      "step": 9747
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012629783259983357,
+      "loss": 0.969,
+      "step": 9748
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.000126258818816963,
+      "loss": 1.0263,
+      "step": 9749
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001262198082812862,
+      "loss": 1.0155,
+      "step": 9750
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001261808009945211,
+      "loss": 1.0319,
+      "step": 9751
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012614179695838535,
+      "loss": 0.9629,
+      "step": 9752
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012610279617459642,
+      "loss": 0.9367,
+      "step": 9753
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012606379864487182,
+      "loss": 0.9939,
+      "step": 9754
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001260248043709288,
+      "loss": 1.0786,
+      "step": 9755
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012598581335448436,
+      "loss": 1.0271,
+      "step": 9756
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012594682559725562,
+      "loss": 0.9621,
+      "step": 9757
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012590784110095933,
+      "loss": 0.9913,
+      "step": 9758
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001258688598673121,
+      "loss": 0.9424,
+      "step": 9759
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001258298818980306,
+      "loss": 0.995,
+      "step": 9760
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001257909071948312,
+      "loss": 0.9778,
+      "step": 9761
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012575193575943008,
+      "loss": 0.984,
+      "step": 9762
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012571296759354342,
+      "loss": 0.9681,
+      "step": 9763
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012567400269888716,
+      "loss": 0.9815,
+      "step": 9764
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012563504107717703,
+      "loss": 1.0855,
+      "step": 9765
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012559608273012885,
+      "loss": 0.9544,
+      "step": 9766
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012555712765945814,
+      "loss": 0.9925,
+      "step": 9767
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012551817586688014,
+      "loss": 1.0245,
+      "step": 9768
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001254792273541103,
+      "loss": 1.0202,
+      "step": 9769
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012544028212286353,
+      "loss": 0.9878,
+      "step": 9770
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001254013401748549,
+      "loss": 1.056,
+      "step": 9771
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001253624015117991,
+      "loss": 0.9695,
+      "step": 9772
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012532346613541094,
+      "loss": 1.0565,
+      "step": 9773
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001252845340474049,
+      "loss": 1.0428,
+      "step": 9774
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012524560524949528,
+      "loss": 0.9943,
+      "step": 9775
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012520667974339638,
+      "loss": 0.9894,
+      "step": 9776
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001251677575308223,
+      "loss": 0.9755,
+      "step": 9777
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012512883861348685,
+      "loss": 0.9572,
+      "step": 9778
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.000125089922993104,
+      "loss": 0.9525,
+      "step": 9779
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012505101067138732,
+      "loss": 0.9989,
+      "step": 9780
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012501210165005028,
+      "loss": 0.9808,
+      "step": 9781
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012497319593080634,
+      "loss": 0.9996,
+      "step": 9782
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001249342935153686,
+      "loss": 1.0934,
+      "step": 9783
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012489539440545012,
+      "loss": 1.0166,
+      "step": 9784
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012485649860276396,
+      "loss": 0.918,
+      "step": 9785
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001248176061090228,
+      "loss": 0.9633,
+      "step": 9786
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012477871692593927,
+      "loss": 1.0039,
+      "step": 9787
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001247398310552259,
+      "loss": 1.0371,
+      "step": 9788
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.000124700948498595,
+      "loss": 1.0432,
+      "step": 9789
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012466206925775873,
+      "loss": 0.9928,
+      "step": 9790
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001246231933344292,
+      "loss": 0.9834,
+      "step": 9791
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012458432073031832,
+      "loss": 1.0083,
+      "step": 9792
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012454545144713775,
+      "loss": 0.96,
+      "step": 9793
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001245065854865992,
+      "loss": 1.025,
+      "step": 9794
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012446772285041411,
+      "loss": 0.9772,
+      "step": 9795
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001244288635402937,
+      "loss": 0.9129,
+      "step": 9796
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012439000755794932,
+      "loss": 0.9217,
+      "step": 9797
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012435115490509188,
+      "loss": 1.0452,
+      "step": 9798
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012431230558343227,
+      "loss": 0.9879,
+      "step": 9799
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012427345959468118,
+      "loss": 1.0537,
+      "step": 9800
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012423461694054927,
+      "loss": 0.9362,
+      "step": 9801
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012419577762274697,
+      "loss": 1.007,
+      "step": 9802
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012415694164298442,
+      "loss": 0.9345,
+      "step": 9803
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012411810900297198,
+      "loss": 1.0323,
+      "step": 9804
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001240792797044196,
+      "loss": 1.0318,
+      "step": 9805
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.000124040453749037,
+      "loss": 0.9867,
+      "step": 9806
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.000124001631138534,
+      "loss": 0.9131,
+      "step": 9807
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001239628118746201,
+      "loss": 0.9737,
+      "step": 9808
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012392399595900465,
+      "loss": 0.9827,
+      "step": 9809
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001238851833933971,
+      "loss": 0.9867,
+      "step": 9810
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001238463741795064,
+      "loss": 0.9844,
+      "step": 9811
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012380756831904157,
+      "loss": 1.0272,
+      "step": 9812
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001237687658137114,
+      "loss": 1.0183,
+      "step": 9813
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001237299666652246,
+      "loss": 1.0192,
+      "step": 9814
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012369117087528958,
+      "loss": 0.9401,
+      "step": 9815
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012365237844561488,
+      "loss": 0.9195,
+      "step": 9816
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012361358937790863,
+      "loss": 1.0597,
+      "step": 9817
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001235748036738789,
+      "loss": 1.0106,
+      "step": 9818
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012353602133523367,
+      "loss": 0.9019,
+      "step": 9819
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012349724236368067,
+      "loss": 0.9971,
+      "step": 9820
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001234584667609275,
+      "loss": 0.9375,
+      "step": 9821
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012341969452868176,
+      "loss": 0.9865,
+      "step": 9822
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012338092566865073,
+      "loss": 1.0149,
+      "step": 9823
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012334216018254155,
+      "loss": 1.1235,
+      "step": 9824
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012330339807206133,
+      "loss": 1.0338,
+      "step": 9825
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012326463933891692,
+      "loss": 0.9741,
+      "step": 9826
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001232258839848151,
+      "loss": 1.0214,
+      "step": 9827
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012318713201146233,
+      "loss": 1.0106,
+      "step": 9828
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012314838342056524,
+      "loss": 1.0058,
+      "step": 9829
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012310963821383003,
+      "loss": 1.0877,
+      "step": 9830
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001230708963929628,
+      "loss": 0.9214,
+      "step": 9831
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001230321579596697,
+      "loss": 1.0462,
+      "step": 9832
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012299342291565644,
+      "loss": 1.0321,
+      "step": 9833
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012295469126262867,
+      "loss": 0.9176,
+      "step": 9834
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001229159630022921,
+      "loss": 0.9851,
+      "step": 9835
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001228772381363521,
+      "loss": 0.9984,
+      "step": 9836
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001228385166665138,
+      "loss": 1.0436,
+      "step": 9837
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012279979859448244,
+      "loss": 1.0788,
+      "step": 9838
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012276108392196287,
+      "loss": 0.9411,
+      "step": 9839
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001227223726506599,
+      "loss": 1.0154,
+      "step": 9840
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012268366478227825,
+      "loss": 1.0247,
+      "step": 9841
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012264496031852236,
+      "loss": 0.9397,
+      "step": 9842
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012260625926109663,
+      "loss": 0.9295,
+      "step": 9843
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012256756161170521,
+      "loss": 1.0325,
+      "step": 9844
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001225288673720522,
+      "loss": 0.9086,
+      "step": 9845
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001224901765438414,
+      "loss": 1.0069,
+      "step": 9846
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012245148912877672,
+      "loss": 1.0026,
+      "step": 9847
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001224128051285617,
+      "loss": 0.9488,
+      "step": 9848
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001223741245448997,
+      "loss": 1.0001,
+      "step": 9849
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012233544737949413,
+      "loss": 1.0313,
+      "step": 9850
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012229677363404808,
+      "loss": 0.9848,
+      "step": 9851
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001222581033102646,
+      "loss": 0.9619,
+      "step": 9852
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012221943640984643,
+      "loss": 0.9634,
+      "step": 9853
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012218077293449638,
+      "loss": 1.0273,
+      "step": 9854
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.000122142112885917,
+      "loss": 0.9957,
+      "step": 9855
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0001221034562658106,
+      "loss": 1.0417,
+      "step": 9856
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012206480307587948,
+      "loss": 0.9947,
+      "step": 9857
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012202615331782576,
+      "loss": 0.9782,
+      "step": 9858
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012198750699335124,
+      "loss": 0.998,
+      "step": 9859
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012194886410415792,
+      "loss": 1.0109,
+      "step": 9860
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012191022465194733,
+      "loss": 0.9725,
+      "step": 9861
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012187158863842092,
+      "loss": 0.9939,
+      "step": 9862
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012183295606528013,
+      "loss": 1.0671,
+      "step": 9863
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012179432693422608,
+      "loss": 1.0228,
+      "step": 9864
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012175570124695971,
+      "loss": 1.0433,
+      "step": 9865
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012171707900518212,
+      "loss": 0.9997,
+      "step": 9866
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0001216784602105939,
+      "loss": 1.0421,
+      "step": 9867
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012163984486489566,
+      "loss": 0.9545,
+      "step": 9868
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012160123296978782,
+      "loss": 0.9972,
+      "step": 9869
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012156262452697067,
+      "loss": 1.0199,
+      "step": 9870
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012152401953814428,
+      "loss": 1.0131,
+      "step": 9871
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012148541800500868,
+      "loss": 1.0725,
+      "step": 9872
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012144681992926373,
+      "loss": 1.032,
+      "step": 9873
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012140822531260896,
+      "loss": 1.0153,
+      "step": 9874
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012136963415674403,
+      "loss": 1.077,
+      "step": 9875
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012133104646336821,
+      "loss": 0.8784,
+      "step": 9876
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012129246223418068,
+      "loss": 0.9477,
+      "step": 9877
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012125388147088062,
+      "loss": 1.0033,
+      "step": 9878
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012121530417516686,
+      "loss": 1.0631,
+      "step": 9879
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012117673034873816,
+      "loss": 0.9815,
+      "step": 9880
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012113815999329307,
+      "loss": 1.0369,
+      "step": 9881
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012109959311053015,
+      "loss": 0.945,
+      "step": 9882
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012106102970214761,
+      "loss": 0.9346,
+      "step": 9883
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012102246976984353,
+      "loss": 1.0533,
+      "step": 9884
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012098391331531604,
+      "loss": 0.9957,
+      "step": 9885
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012094536034026292,
+      "loss": 0.965,
+      "step": 9886
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0001209068108463818,
+      "loss": 1.0231,
+      "step": 9887
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012086826483537028,
+      "loss": 1.0501,
+      "step": 9888
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012082972230892573,
+      "loss": 1.0351,
+      "step": 9889
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012079118326874521,
+      "loss": 0.9591,
+      "step": 9890
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012075264771652603,
+      "loss": 1.037,
+      "step": 9891
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012071411565396497,
+      "loss": 0.974,
+      "step": 9892
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012067558708275881,
+      "loss": 0.9244,
+      "step": 9893
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012063706200460416,
+      "loss": 0.9774,
+      "step": 9894
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0001205985404211975,
+      "loss": 0.9896,
+      "step": 9895
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.000120560022334235,
+      "loss": 1.0413,
+      "step": 9896
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012052150774541298,
+      "loss": 1.0416,
+      "step": 9897
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012048299665642735,
+      "loss": 0.9409,
+      "step": 9898
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012044448906897389,
+      "loss": 0.9597,
+      "step": 9899
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012040598498474842,
+      "loss": 0.998,
+      "step": 9900
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012036748440544639,
+      "loss": 1.0306,
+      "step": 9901
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012032898733276305,
+      "loss": 0.9692,
+      "step": 9902
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012029049376839388,
+      "loss": 0.9931,
+      "step": 9903
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012025200371403376,
+      "loss": 0.9815,
+      "step": 9904
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012021351717137767,
+      "loss": 0.9741,
+      "step": 9905
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00012017503414212028,
+      "loss": 0.9063,
+      "step": 9906
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0001201365546279563,
+      "loss": 1.0019,
+      "step": 9907
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00012009807863058013,
+      "loss": 1.0097,
+      "step": 9908
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00012005960615168598,
+      "loss": 0.9917,
+      "step": 9909
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00012002113719296818,
+      "loss": 0.9615,
+      "step": 9910
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011998267175612054,
+      "loss": 1.0462,
+      "step": 9911
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011994420984283692,
+      "loss": 1.074,
+      "step": 9912
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011990575145481107,
+      "loss": 0.9657,
+      "step": 9913
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011986729659373641,
+      "loss": 0.9817,
+      "step": 9914
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011982884526130629,
+      "loss": 1.0443,
+      "step": 9915
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011979039745921405,
+      "loss": 1.0867,
+      "step": 9916
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011975195318915265,
+      "loss": 0.9229,
+      "step": 9917
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011971351245281495,
+      "loss": 0.9859,
+      "step": 9918
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011967507525189374,
+      "loss": 1.0003,
+      "step": 9919
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011963664158808159,
+      "loss": 0.9503,
+      "step": 9920
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011959821146307089,
+      "loss": 1.0685,
+      "step": 9921
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011955978487855397,
+      "loss": 0.9962,
+      "step": 9922
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011952136183622295,
+      "loss": 1.0039,
+      "step": 9923
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011948294233776974,
+      "loss": 1.01,
+      "step": 9924
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011944452638488615,
+      "loss": 0.9026,
+      "step": 9925
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0001194061139792639,
+      "loss": 1.0195,
+      "step": 9926
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011936770512259428,
+      "loss": 0.9837,
+      "step": 9927
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011932929981656888,
+      "loss": 1.0204,
+      "step": 9928
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011929089806287876,
+      "loss": 1.0682,
+      "step": 9929
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011925249986321492,
+      "loss": 0.9897,
+      "step": 9930
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011921410521926826,
+      "loss": 0.9537,
+      "step": 9931
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0001191757141327295,
+      "loss": 1.0163,
+      "step": 9932
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011913732660528918,
+      "loss": 1.0003,
+      "step": 9933
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011909894263863756,
+      "loss": 1.013,
+      "step": 9934
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011906056223446514,
+      "loss": 0.9945,
+      "step": 9935
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011902218539446184,
+      "loss": 1.0172,
+      "step": 9936
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011898381212031758,
+      "loss": 1.009,
+      "step": 9937
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011894544241372219,
+      "loss": 1.0816,
+      "step": 9938
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011890707627636523,
+      "loss": 1.0015,
+      "step": 9939
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011886871370993614,
+      "loss": 0.9902,
+      "step": 9940
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0001188303547161243,
+      "loss": 0.9605,
+      "step": 9941
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0001187919992966188,
+      "loss": 1.0057,
+      "step": 9942
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011875364745310858,
+      "loss": 0.9967,
+      "step": 9943
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011871529918728255,
+      "loss": 0.9788,
+      "step": 9944
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011867695450082931,
+      "loss": 1.0,
+      "step": 9945
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011863861339543732,
+      "loss": 1.0177,
+      "step": 9946
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011860027587279509,
+      "loss": 0.8757,
+      "step": 9947
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011856194193459069,
+      "loss": 0.9833,
+      "step": 9948
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011852361158251218,
+      "loss": 1.0304,
+      "step": 9949
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011848528481824748,
+      "loss": 0.9609,
+      "step": 9950
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011844696164348428,
+      "loss": 0.9589,
+      "step": 9951
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011840864205991002,
+      "loss": 1.0153,
+      "step": 9952
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011837032606921233,
+      "loss": 1.0435,
+      "step": 9953
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011833201367307834,
+      "loss": 0.977,
+      "step": 9954
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011829370487319509,
+      "loss": 0.9661,
+      "step": 9955
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011825539967124961,
+      "loss": 0.9596,
+      "step": 9956
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011821709806892863,
+      "loss": 1.0338,
+      "step": 9957
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011817880006791866,
+      "loss": 1.0186,
+      "step": 9958
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011814050566990632,
+      "loss": 0.9594,
+      "step": 9959
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011810221487657785,
+      "loss": 0.9706,
+      "step": 9960
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011806392768961933,
+      "loss": 1.0438,
+      "step": 9961
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011802564411071674,
+      "loss": 0.9934,
+      "step": 9962
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.000117987364141556,
+      "loss": 1.0234,
+      "step": 9963
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011794908778382266,
+      "loss": 0.9041,
+      "step": 9964
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011791081503920219,
+      "loss": 1.0346,
+      "step": 9965
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011787254590938008,
+      "loss": 0.9587,
+      "step": 9966
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011783428039604145,
+      "loss": 0.9809,
+      "step": 9967
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011779601850087123,
+      "loss": 0.9362,
+      "step": 9968
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0001177577602255544,
+      "loss": 0.9655,
+      "step": 9969
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011771950557177565,
+      "loss": 0.9702,
+      "step": 9970
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011768125454121936,
+      "loss": 0.9385,
+      "step": 9971
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011764300713557017,
+      "loss": 1.0104,
+      "step": 9972
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011760476335651218,
+      "loss": 1.0093,
+      "step": 9973
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011756652320572943,
+      "loss": 0.978,
+      "step": 9974
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011752828668490585,
+      "loss": 0.9598,
+      "step": 9975
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011749005379572521,
+      "loss": 0.9983,
+      "step": 9976
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011745182453987103,
+      "loss": 0.932,
+      "step": 9977
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011741359891902681,
+      "loss": 1.0118,
+      "step": 9978
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011737537693487582,
+      "loss": 1.011,
+      "step": 9979
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011733715858910106,
+      "loss": 1.0044,
+      "step": 9980
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011729894388338562,
+      "loss": 0.9799,
+      "step": 9981
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0001172607328194122,
+      "loss": 0.9433,
+      "step": 9982
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011722252539886334,
+      "loss": 0.9752,
+      "step": 9983
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0001171843216234217,
+      "loss": 1.0128,
+      "step": 9984
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011714612149476947,
+      "loss": 1.0648,
+      "step": 9985
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011710792501458883,
+      "loss": 1.0209,
+      "step": 9986
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011706973218456169,
+      "loss": 0.9982,
+      "step": 9987
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011703154300636996,
+      "loss": 0.9677,
+      "step": 9988
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011699335748169527,
+      "loss": 0.9577,
+      "step": 9989
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011695517561221903,
+      "loss": 1.0518,
+      "step": 9990
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011691699739962274,
+      "loss": 0.9866,
+      "step": 9991
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011687882284558749,
+      "loss": 1.0008,
+      "step": 9992
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0001168406519517943,
+      "loss": 1.007,
+      "step": 9993
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011680248471992405,
+      "loss": 0.9984,
+      "step": 9994
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011676432115165741,
+      "loss": 1.0144,
+      "step": 9995
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011672616124867489,
+      "loss": 0.9413,
+      "step": 9996
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011668800501265691,
+      "loss": 1.06,
+      "step": 9997
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0001166498524452837,
+      "loss": 0.9025,
+      "step": 9998
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011661170354823516,
+      "loss": 1.0022,
+      "step": 9999
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011657355832319141,
+      "loss": 1.0226,
+      "step": 10000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011653541677183199,
+      "loss": 0.967,
+      "step": 10001
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011649727889583655,
+      "loss": 1.0061,
+      "step": 10002
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011645914469688452,
+      "loss": 0.9608,
+      "step": 10003
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011642101417665502,
+      "loss": 1.0217,
+      "step": 10004
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011638288733682721,
+      "loss": 0.9931,
+      "step": 10005
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011634476417907998,
+      "loss": 1.1154,
+      "step": 10006
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011630664470509218,
+      "loss": 1.0332,
+      "step": 10007
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011626852891654226,
+      "loss": 0.9713,
+      "step": 10008
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011623041681510866,
+      "loss": 0.9794,
+      "step": 10009
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011619230840246982,
+      "loss": 0.9964,
+      "step": 10010
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011615420368030354,
+      "loss": 1.0152,
+      "step": 10011
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011611610265028812,
+      "loss": 1.0056,
+      "step": 10012
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011607800531410107,
+      "loss": 0.9298,
+      "step": 10013
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011603991167342014,
+      "loss": 1.0367,
+      "step": 10014
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011600182172992272,
+      "loss": 0.9732,
+      "step": 10015
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011596373548528609,
+      "loss": 1.0051,
+      "step": 10016
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011592565294118749,
+      "loss": 0.959,
+      "step": 10017
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011588757409930367,
+      "loss": 1.0077,
+      "step": 10018
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0001158494989613117,
+      "loss": 0.9641,
+      "step": 10019
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.000115811427528888,
+      "loss": 1.0281,
+      "step": 10020
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011577335980370918,
+      "loss": 0.9808,
+      "step": 10021
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011573529578745154,
+      "loss": 1.0379,
+      "step": 10022
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011569723548179114,
+      "loss": 1.0051,
+      "step": 10023
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011565917888840403,
+      "loss": 0.9971,
+      "step": 10024
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.000115621126008966,
+      "loss": 0.9946,
+      "step": 10025
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011558307684515282,
+      "loss": 0.9978,
+      "step": 10026
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011554503139863984,
+      "loss": 1.0029,
+      "step": 10027
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011550698967110242,
+      "loss": 1.0605,
+      "step": 10028
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011546895166421587,
+      "loss": 1.03,
+      "step": 10029
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011543091737965488,
+      "loss": 1.0579,
+      "step": 10030
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011539288681909468,
+      "loss": 0.9662,
+      "step": 10031
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011535485998420968,
+      "loss": 0.9372,
+      "step": 10032
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011531683687667448,
+      "loss": 0.9186,
+      "step": 10033
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011527881749816347,
+      "loss": 0.9878,
+      "step": 10034
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011524080185035071,
+      "loss": 1.1015,
+      "step": 10035
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011520278993491032,
+      "loss": 0.9108,
+      "step": 10036
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0001151647817535161,
+      "loss": 1.0316,
+      "step": 10037
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011512677730784185,
+      "loss": 1.0113,
+      "step": 10038
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011508877659956096,
+      "loss": 0.9676,
+      "step": 10039
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011505077963034683,
+      "loss": 1.0055,
+      "step": 10040
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011501278640187274,
+      "loss": 0.859,
+      "step": 10041
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0001149747969158116,
+      "loss": 0.9107,
+      "step": 10042
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0001149368111738363,
+      "loss": 0.962,
+      "step": 10043
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0001148988291776196,
+      "loss": 0.9543,
+      "step": 10044
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011486085092883409,
+      "loss": 0.9744,
+      "step": 10045
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.000114822876429152,
+      "loss": 1.0518,
+      "step": 10046
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011478490568024555,
+      "loss": 1.0052,
+      "step": 10047
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011474693868378699,
+      "loss": 0.9441,
+      "step": 10048
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011470897544144783,
+      "loss": 0.9415,
+      "step": 10049
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011467101595490018,
+      "loss": 0.9322,
+      "step": 10050
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0001146330602258153,
+      "loss": 0.9824,
+      "step": 10051
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011459510825586468,
+      "loss": 0.9551,
+      "step": 10052
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011455716004671963,
+      "loss": 1.0272,
+      "step": 10053
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011451921560005102,
+      "loss": 0.9979,
+      "step": 10054
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011448127491752979,
+      "loss": 1.042,
+      "step": 10055
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011444333800082666,
+      "loss": 0.945,
+      "step": 10056
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0001144054048516123,
+      "loss": 0.9315,
+      "step": 10057
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011436747547155686,
+      "loss": 1.0385,
+      "step": 10058
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011432954986233085,
+      "loss": 0.9527,
+      "step": 10059
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011429162802560416,
+      "loss": 1.0468,
+      "step": 10060
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011425370996304654,
+      "loss": 1.0469,
+      "step": 10061
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011421579567632802,
+      "loss": 0.9489,
+      "step": 10062
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011417788516711794,
+      "loss": 1.0145,
+      "step": 10063
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011413997843708575,
+      "loss": 0.9683,
+      "step": 10064
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011410207548790067,
+      "loss": 1.0164,
+      "step": 10065
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011406417632123184,
+      "loss": 0.948,
+      "step": 10066
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011402628093874805,
+      "loss": 0.9775,
+      "step": 10067
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011398838934211793,
+      "loss": 1.0275,
+      "step": 10068
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011395050153301028,
+      "loss": 0.9381,
+      "step": 10069
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0001139126175130933,
+      "loss": 1.0156,
+      "step": 10070
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011387473728403528,
+      "loss": 0.987,
+      "step": 10071
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011383686084750434,
+      "loss": 0.988,
+      "step": 10072
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0001137989882051682,
+      "loss": 1.024,
+      "step": 10073
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011376111935869472,
+      "loss": 1.0484,
+      "step": 10074
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011372325430975142,
+      "loss": 0.9698,
+      "step": 10075
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011368539306000575,
+      "loss": 0.9876,
+      "step": 10076
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011364753561112471,
+      "loss": 0.9242,
+      "step": 10077
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011360968196477568,
+      "loss": 1.0524,
+      "step": 10078
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011357183212262536,
+      "loss": 0.9136,
+      "step": 10079
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011353398608634033,
+      "loss": 0.9734,
+      "step": 10080
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011349614385758749,
+      "loss": 0.9657,
+      "step": 10081
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011345830543803291,
+      "loss": 0.9579,
+      "step": 10082
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011342047082934293,
+      "loss": 0.9371,
+      "step": 10083
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011338264003318359,
+      "loss": 0.9543,
+      "step": 10084
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011334481305122086,
+      "loss": 1.0489,
+      "step": 10085
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011330698988512027,
+      "loss": 0.9942,
+      "step": 10086
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011326917053654749,
+      "loss": 1.0025,
+      "step": 10087
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011323135500716791,
+      "loss": 0.9667,
+      "step": 10088
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011319354329864653,
+      "loss": 1.0619,
+      "step": 10089
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011315573541264872,
+      "loss": 0.9624,
+      "step": 10090
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011311793135083916,
+      "loss": 0.9606,
+      "step": 10091
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011308013111488243,
+      "loss": 1.0084,
+      "step": 10092
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011304233470644337,
+      "loss": 0.966,
+      "step": 10093
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011300454212718607,
+      "loss": 1.0148,
+      "step": 10094
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011296675337877496,
+      "loss": 0.9453,
+      "step": 10095
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011292896846287375,
+      "loss": 1.0496,
+      "step": 10096
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011289118738114668,
+      "loss": 0.9737,
+      "step": 10097
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011285341013525723,
+      "loss": 0.9622,
+      "step": 10098
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011281563672686879,
+      "loss": 1.0378,
+      "step": 10099
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011277786715764503,
+      "loss": 0.9636,
+      "step": 10100
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011274010142924889,
+      "loss": 0.9707,
+      "step": 10101
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011270233954334348,
+      "loss": 1.0448,
+      "step": 10102
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011266458150159165,
+      "loss": 0.9835,
+      "step": 10103
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0001126268273056561,
+      "loss": 1.034,
+      "step": 10104
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011258907695719923,
+      "loss": 0.9524,
+      "step": 10105
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011255133045788344,
+      "loss": 0.9983,
+      "step": 10106
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011251358780937096,
+      "loss": 0.9041,
+      "step": 10107
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0001124758490133236,
+      "loss": 0.9546,
+      "step": 10108
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011243811407140346,
+      "loss": 0.9542,
+      "step": 10109
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011240038298527204,
+      "loss": 0.9295,
+      "step": 10110
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001123626557565907,
+      "loss": 1.0729,
+      "step": 10111
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011232493238702107,
+      "loss": 0.9628,
+      "step": 10112
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011228721287822402,
+      "loss": 1.009,
+      "step": 10113
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011224949723186067,
+      "loss": 1.017,
+      "step": 10114
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011221178544959176,
+      "loss": 1.0742,
+      "step": 10115
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011217407753307808,
+      "loss": 1.0096,
+      "step": 10116
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011213637348397989,
+      "loss": 0.9962,
+      "step": 10117
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001120986733039576,
+      "loss": 0.9578,
+      "step": 10118
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001120609769946714,
+      "loss": 1.0334,
+      "step": 10119
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011202328455778106,
+      "loss": 0.929,
+      "step": 10120
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011198559599494652,
+      "loss": 1.0423,
+      "step": 10121
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011194791130782731,
+      "loss": 1.1061,
+      "step": 10122
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011191023049808303,
+      "loss": 1.0114,
+      "step": 10123
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011187255356737274,
+      "loss": 1.0375,
+      "step": 10124
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011183488051735564,
+      "loss": 0.976,
+      "step": 10125
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011179721134969079,
+      "loss": 0.9573,
+      "step": 10126
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011175954606603665,
+      "loss": 1.0441,
+      "step": 10127
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011172188466805216,
+      "loss": 1.114,
+      "step": 10128
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011168422715739556,
+      "loss": 1.0664,
+      "step": 10129
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011164657353572496,
+      "loss": 0.9736,
+      "step": 10130
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011160892380469876,
+      "loss": 0.9926,
+      "step": 10131
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011157127796597463,
+      "loss": 1.0494,
+      "step": 10132
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011153363602121034,
+      "loss": 0.9222,
+      "step": 10133
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011149599797206351,
+      "loss": 1.0335,
+      "step": 10134
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011145836382019157,
+      "loss": 1.0243,
+      "step": 10135
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001114207335672516,
+      "loss": 0.9712,
+      "step": 10136
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011138310721490075,
+      "loss": 0.9966,
+      "step": 10137
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011134548476479594,
+      "loss": 1.0122,
+      "step": 10138
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011130786621859361,
+      "loss": 0.9599,
+      "step": 10139
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001112702515779507,
+      "loss": 1.0244,
+      "step": 10140
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011123264084452328,
+      "loss": 0.9687,
+      "step": 10141
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001111950340199676,
+      "loss": 1.0537,
+      "step": 10142
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011115743110593979,
+      "loss": 1.0101,
+      "step": 10143
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011111983210409551,
+      "loss": 0.8707,
+      "step": 10144
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011108223701609056,
+      "loss": 0.916,
+      "step": 10145
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011104464584358036,
+      "loss": 0.918,
+      "step": 10146
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001110070585882204,
+      "loss": 0.9877,
+      "step": 10147
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011096947525166573,
+      "loss": 0.9999,
+      "step": 10148
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011093189583557113,
+      "loss": 0.9805,
+      "step": 10149
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011089432034159177,
+      "loss": 0.9841,
+      "step": 10150
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011085674877138204,
+      "loss": 0.9726,
+      "step": 10151
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011081918112659648,
+      "loss": 1.0094,
+      "step": 10152
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011078161740888942,
+      "loss": 1.0023,
+      "step": 10153
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011074405761991498,
+      "loss": 0.9763,
+      "step": 10154
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011070650176132702,
+      "loss": 0.9719,
+      "step": 10155
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011066894983477936,
+      "loss": 0.9332,
+      "step": 10156
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011063140184192567,
+      "loss": 0.9271,
+      "step": 10157
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011059385778441913,
+      "loss": 0.9735,
+      "step": 10158
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011055631766391337,
+      "loss": 1.0139,
+      "step": 10159
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011051878148206118,
+      "loss": 1.1328,
+      "step": 10160
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011048124924051554,
+      "loss": 0.9981,
+      "step": 10161
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001104437209409293,
+      "loss": 0.961,
+      "step": 10162
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00011040619658495483,
+      "loss": 1.0492,
+      "step": 10163
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00011036867617424462,
+      "loss": 1.0042,
+      "step": 10164
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00011033115971045088,
+      "loss": 1.0654,
+      "step": 10165
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00011029364719522566,
+      "loss": 1.0105,
+      "step": 10166
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00011025613863022077,
+      "loss": 0.8973,
+      "step": 10167
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001102186340170879,
+      "loss": 0.9334,
+      "step": 10168
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00011018113335747869,
+      "loss": 0.9995,
+      "step": 10169
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00011014363665304421,
+      "loss": 1.0789,
+      "step": 10170
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00011010614390543598,
+      "loss": 0.9822,
+      "step": 10171
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00011006865511630473,
+      "loss": 0.9794,
+      "step": 10172
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00011003117028730145,
+      "loss": 0.9232,
+      "step": 10173
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010999368942007664,
+      "loss": 1.0077,
+      "step": 10174
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010995621251628083,
+      "loss": 1.0495,
+      "step": 10175
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010991873957756438,
+      "loss": 0.9876,
+      "step": 10176
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001098812706055772,
+      "loss": 1.0162,
+      "step": 10177
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010984380560196954,
+      "loss": 0.9442,
+      "step": 10178
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010980634456839096,
+      "loss": 1.0326,
+      "step": 10179
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001097688875064911,
+      "loss": 1.0135,
+      "step": 10180
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001097314344179195,
+      "loss": 0.9814,
+      "step": 10181
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001096939853043252,
+      "loss": 1.0443,
+      "step": 10182
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010965654016735739,
+      "loss": 0.9885,
+      "step": 10183
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010961909900866494,
+      "loss": 0.9402,
+      "step": 10184
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010958166182989669,
+      "loss": 1.0171,
+      "step": 10185
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010954422863270097,
+      "loss": 1.006,
+      "step": 10186
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001095067994187263,
+      "loss": 0.9293,
+      "step": 10187
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010946937418962089,
+      "loss": 0.9515,
+      "step": 10188
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010943195294703257,
+      "loss": 1.0219,
+      "step": 10189
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010939453569260947,
+      "loss": 1.0204,
+      "step": 10190
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010935712242799904,
+      "loss": 1.0606,
+      "step": 10191
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010931971315484888,
+      "loss": 1.0732,
+      "step": 10192
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010928230787480634,
+      "loss": 0.9431,
+      "step": 10193
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010924490658951841,
+      "loss": 1.0041,
+      "step": 10194
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010920750930063213,
+      "loss": 0.9815,
+      "step": 10195
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010917011600979433,
+      "loss": 1.0006,
+      "step": 10196
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010913272671865167,
+      "loss": 1.0045,
+      "step": 10197
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010909534142885043,
+      "loss": 0.9831,
+      "step": 10198
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010905796014203696,
+      "loss": 1.0122,
+      "step": 10199
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010902058285985743,
+      "loss": 0.9681,
+      "step": 10200
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010898320958395758,
+      "loss": 0.9733,
+      "step": 10201
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010894584031598323,
+      "loss": 1.0211,
+      "step": 10202
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010890847505757993,
+      "loss": 1.0393,
+      "step": 10203
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010887111381039314,
+      "loss": 0.9737,
+      "step": 10204
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010883375657606789,
+      "loss": 0.942,
+      "step": 10205
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010879640335624933,
+      "loss": 0.9037,
+      "step": 10206
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010875905415258234,
+      "loss": 1.0286,
+      "step": 10207
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010872170896671137,
+      "loss": 1.0041,
+      "step": 10208
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010868436780028126,
+      "loss": 0.9266,
+      "step": 10209
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010864703065493605,
+      "loss": 0.9924,
+      "step": 10210
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010860969753231998,
+      "loss": 1.0027,
+      "step": 10211
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001085723684340771,
+      "loss": 0.9756,
+      "step": 10212
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.000108535043361851,
+      "loss": 0.9478,
+      "step": 10213
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010849772231728543,
+      "loss": 0.946,
+      "step": 10214
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001084604053020238,
+      "loss": 1.0389,
+      "step": 10215
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010842309231770941,
+      "loss": 1.0313,
+      "step": 10216
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010838578336598524,
+      "loss": 0.9405,
+      "step": 10217
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010834847844849423,
+      "loss": 0.878,
+      "step": 10218
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010831117756687917,
+      "loss": 1.0352,
+      "step": 10219
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010827388072278237,
+      "loss": 0.9965,
+      "step": 10220
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010823658791784657,
+      "loss": 0.9513,
+      "step": 10221
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010819929915371365,
+      "loss": 1.0247,
+      "step": 10222
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010816201443202576,
+      "loss": 1.0626,
+      "step": 10223
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010812473375442476,
+      "loss": 0.9408,
+      "step": 10224
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010808745712255217,
+      "loss": 1.0131,
+      "step": 10225
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010805018453804965,
+      "loss": 1.0254,
+      "step": 10226
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010801291600255821,
+      "loss": 0.9849,
+      "step": 10227
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010797565151771934,
+      "loss": 0.9632,
+      "step": 10228
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001079383910851737,
+      "loss": 0.9499,
+      "step": 10229
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010790113470656216,
+      "loss": 0.9363,
+      "step": 10230
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010786388238352536,
+      "loss": 1.0222,
+      "step": 10231
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010782663411770357,
+      "loss": 0.9847,
+      "step": 10232
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010778938991073709,
+      "loss": 0.9629,
+      "step": 10233
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010775214976426595,
+      "loss": 0.9791,
+      "step": 10234
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010771491367993017,
+      "loss": 0.9956,
+      "step": 10235
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001076776816593692,
+      "loss": 0.9109,
+      "step": 10236
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010764045370422265,
+      "loss": 0.9387,
+      "step": 10237
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010760322981612999,
+      "loss": 1.0651,
+      "step": 10238
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010756600999673003,
+      "loss": 0.9651,
+      "step": 10239
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010752879424766218,
+      "loss": 0.9603,
+      "step": 10240
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010749158257056493,
+      "loss": 1.0258,
+      "step": 10241
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010745437496707698,
+      "loss": 0.9694,
+      "step": 10242
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010741717143883687,
+      "loss": 0.9983,
+      "step": 10243
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010737997198748266,
+      "loss": 1.0034,
+      "step": 10244
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010734277661465255,
+      "loss": 0.9802,
+      "step": 10245
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010730558532198442,
+      "loss": 0.971,
+      "step": 10246
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010726839811111608,
+      "loss": 1.052,
+      "step": 10247
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001072312149836849,
+      "loss": 1.0686,
+      "step": 10248
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010719403594132829,
+      "loss": 0.9789,
+      "step": 10249
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010715686098568359,
+      "loss": 1.0249,
+      "step": 10250
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010711969011838747,
+      "loss": 1.0339,
+      "step": 10251
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010708252334107709,
+      "loss": 0.9618,
+      "step": 10252
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010704536065538891,
+      "loss": 0.9717,
+      "step": 10253
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001070082020629595,
+      "loss": 1.0053,
+      "step": 10254
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010697104756542498,
+      "loss": 0.9346,
+      "step": 10255
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010693389716442153,
+      "loss": 1.0076,
+      "step": 10256
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010689675086158515,
+      "loss": 0.9589,
+      "step": 10257
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010685960865855134,
+      "loss": 0.996,
+      "step": 10258
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.000106822470556956,
+      "loss": 0.9868,
+      "step": 10259
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010678533655843424,
+      "loss": 1.0384,
+      "step": 10260
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001067482066646213,
+      "loss": 0.9174,
+      "step": 10261
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010671108087715236,
+      "loss": 1.0067,
+      "step": 10262
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010667395919766203,
+      "loss": 0.9862,
+      "step": 10263
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010663684162778508,
+      "loss": 0.9298,
+      "step": 10264
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010659972816915598,
+      "loss": 1.0278,
+      "step": 10265
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010656261882340906,
+      "loss": 1.0329,
+      "step": 10266
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010652551359217832,
+      "loss": 1.0026,
+      "step": 10267
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010648841247709773,
+      "loss": 0.9871,
+      "step": 10268
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010645131547980117,
+      "loss": 1.0287,
+      "step": 10269
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010641422260192188,
+      "loss": 0.9922,
+      "step": 10270
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010637713384509367,
+      "loss": 1.0447,
+      "step": 10271
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010634004921094942,
+      "loss": 1.0034,
+      "step": 10272
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010630296870112229,
+      "loss": 0.9707,
+      "step": 10273
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010626589231724517,
+      "loss": 0.9079,
+      "step": 10274
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010622882006095056,
+      "loss": 0.9068,
+      "step": 10275
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.000106191751933871,
+      "loss": 1.0406,
+      "step": 10276
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010615468793763885,
+      "loss": 0.9727,
+      "step": 10277
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010611762807388625,
+      "loss": 0.9387,
+      "step": 10278
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010608057234424498,
+      "loss": 1.051,
+      "step": 10279
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010604352075034687,
+      "loss": 1.0016,
+      "step": 10280
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010600647329382355,
+      "loss": 1.0299,
+      "step": 10281
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010596942997630628,
+      "loss": 0.9843,
+      "step": 10282
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010593239079942634,
+      "loss": 0.9875,
+      "step": 10283
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001058953557648147,
+      "loss": 1.0371,
+      "step": 10284
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010585832487410237,
+      "loss": 1.032,
+      "step": 10285
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010582129812891976,
+      "loss": 0.9853,
+      "step": 10286
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010578427553089747,
+      "loss": 0.9742,
+      "step": 10287
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010574725708166584,
+      "loss": 0.9311,
+      "step": 10288
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010571024278285478,
+      "loss": 0.9403,
+      "step": 10289
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010567323263609447,
+      "loss": 0.9511,
+      "step": 10290
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010563622664301447,
+      "loss": 0.9235,
+      "step": 10291
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010559922480524442,
+      "loss": 0.9033,
+      "step": 10292
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010556222712441375,
+      "loss": 0.9306,
+      "step": 10293
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001055252336021515,
+      "loss": 0.8299,
+      "step": 10294
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010548824424008672,
+      "loss": 0.925,
+      "step": 10295
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010545125903984833,
+      "loss": 0.9049,
+      "step": 10296
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010541427800306498,
+      "loss": 0.8903,
+      "step": 10297
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010537730113136499,
+      "loss": 0.9201,
+      "step": 10298
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010534032842637672,
+      "loss": 0.9199,
+      "step": 10299
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010530335988972835,
+      "loss": 0.877,
+      "step": 10300
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010526639552304753,
+      "loss": 0.8762,
+      "step": 10301
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010522943532796232,
+      "loss": 0.9438,
+      "step": 10302
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010519247930610005,
+      "loss": 0.9547,
+      "step": 10303
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001051555274590881,
+      "loss": 0.8639,
+      "step": 10304
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010511857978855375,
+      "loss": 0.7975,
+      "step": 10305
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010508163629612387,
+      "loss": 0.9944,
+      "step": 10306
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010504469698342538,
+      "loss": 0.8838,
+      "step": 10307
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010500776185208465,
+      "loss": 0.9001,
+      "step": 10308
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001049708309037285,
+      "loss": 0.9082,
+      "step": 10309
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001049339041399829,
+      "loss": 0.8955,
+      "step": 10310
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010489698156247398,
+      "loss": 0.8464,
+      "step": 10311
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010486006317282778,
+      "loss": 0.9632,
+      "step": 10312
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010482314897266978,
+      "loss": 0.843,
+      "step": 10313
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001047862389636256,
+      "loss": 0.8579,
+      "step": 10314
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010474933314732059,
+      "loss": 0.9376,
+      "step": 10315
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010471243152537993,
+      "loss": 0.9194,
+      "step": 10316
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010467553409942846,
+      "loss": 0.8627,
+      "step": 10317
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010463864087109101,
+      "loss": 0.8563,
+      "step": 10318
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001046017518419923,
+      "loss": 0.9195,
+      "step": 10319
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010456486701375645,
+      "loss": 0.8828,
+      "step": 10320
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010452798638800803,
+      "loss": 0.934,
+      "step": 10321
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010449110996637083,
+      "loss": 0.8539,
+      "step": 10322
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010445423775046881,
+      "loss": 0.8955,
+      "step": 10323
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010441736974192569,
+      "loss": 0.9136,
+      "step": 10324
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010438050594236477,
+      "loss": 0.9273,
+      "step": 10325
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001043436463534095,
+      "loss": 0.8995,
+      "step": 10326
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010430679097668293,
+      "loss": 0.8413,
+      "step": 10327
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010426993981380808,
+      "loss": 0.9144,
+      "step": 10328
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010423309286640755,
+      "loss": 0.9336,
+      "step": 10329
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010419625013610397,
+      "loss": 0.8877,
+      "step": 10330
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010415941162451974,
+      "loss": 0.8721,
+      "step": 10331
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001041225773332769,
+      "loss": 1.0128,
+      "step": 10332
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010408574726399769,
+      "loss": 0.9511,
+      "step": 10333
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010404892141830371,
+      "loss": 0.9023,
+      "step": 10334
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010401209979781676,
+      "loss": 0.8592,
+      "step": 10335
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010397528240415813,
+      "loss": 0.8752,
+      "step": 10336
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010393846923894908,
+      "loss": 0.8969,
+      "step": 10337
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010390166030381084,
+      "loss": 0.9185,
+      "step": 10338
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.000103864855600364,
+      "loss": 0.9206,
+      "step": 10339
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010382805513022961,
+      "loss": 0.9198,
+      "step": 10340
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010379125889502792,
+      "loss": 0.9845,
+      "step": 10341
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010375446689637936,
+      "loss": 0.8809,
+      "step": 10342
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010371767913590411,
+      "loss": 0.9513,
+      "step": 10343
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010368089561522196,
+      "loss": 0.9447,
+      "step": 10344
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010364411633595277,
+      "loss": 0.8846,
+      "step": 10345
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010360734129971615,
+      "loss": 0.9185,
+      "step": 10346
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010357057050813149,
+      "loss": 0.8636,
+      "step": 10347
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010353380396281788,
+      "loss": 0.888,
+      "step": 10348
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010349704166539442,
+      "loss": 0.9279,
+      "step": 10349
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010346028361748001,
+      "loss": 0.8423,
+      "step": 10350
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010342352982069306,
+      "loss": 0.9255,
+      "step": 10351
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001033867802766523,
+      "loss": 0.9195,
+      "step": 10352
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010335003498697582,
+      "loss": 0.876,
+      "step": 10353
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010331329395328171,
+      "loss": 0.8904,
+      "step": 10354
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.000103276557177188,
+      "loss": 0.9549,
+      "step": 10355
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010323982466031224,
+      "loss": 0.9216,
+      "step": 10356
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010320309640427197,
+      "loss": 0.8667,
+      "step": 10357
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010316637241068456,
+      "loss": 0.8975,
+      "step": 10358
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010312965268116723,
+      "loss": 0.9626,
+      "step": 10359
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010309293721733677,
+      "loss": 0.927,
+      "step": 10360
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010305622602081002,
+      "loss": 0.8554,
+      "step": 10361
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010301951909320364,
+      "loss": 0.9865,
+      "step": 10362
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010298281643613385,
+      "loss": 0.9783,
+      "step": 10363
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010294611805121694,
+      "loss": 0.9261,
+      "step": 10364
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010290942394006892,
+      "loss": 0.9773,
+      "step": 10365
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010287273410430573,
+      "loss": 0.9263,
+      "step": 10366
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001028360485455428,
+      "loss": 0.9657,
+      "step": 10367
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001027993672653957,
+      "loss": 0.8425,
+      "step": 10368
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010276269026547973,
+      "loss": 0.9028,
+      "step": 10369
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010272601754740974,
+      "loss": 0.9279,
+      "step": 10370
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010268934911280093,
+      "loss": 0.8281,
+      "step": 10371
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010265268496326777,
+      "loss": 0.8303,
+      "step": 10372
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010261602510042485,
+      "loss": 0.9089,
+      "step": 10373
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010257936952588652,
+      "loss": 0.9119,
+      "step": 10374
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001025427182412668,
+      "loss": 0.9272,
+      "step": 10375
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010250607124817968,
+      "loss": 0.8637,
+      "step": 10376
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010246942854823893,
+      "loss": 0.9064,
+      "step": 10377
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010243279014305817,
+      "loss": 0.9072,
+      "step": 10378
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010239615603425065,
+      "loss": 0.8948,
+      "step": 10379
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010235952622342962,
+      "loss": 0.9137,
+      "step": 10380
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010232290071220813,
+      "loss": 0.8977,
+      "step": 10381
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010228627950219876,
+      "loss": 0.9277,
+      "step": 10382
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010224966259501442,
+      "loss": 0.9153,
+      "step": 10383
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010221304999226736,
+      "loss": 0.8252,
+      "step": 10384
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010217644169556984,
+      "loss": 0.8823,
+      "step": 10385
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010213983770653399,
+      "loss": 0.892,
+      "step": 10386
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010210323802677155,
+      "loss": 0.8362,
+      "step": 10387
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010206664265789428,
+      "loss": 0.8668,
+      "step": 10388
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010203005160151349,
+      "loss": 0.9971,
+      "step": 10389
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010199346485924075,
+      "loss": 0.8852,
+      "step": 10390
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010195688243268691,
+      "loss": 0.9034,
+      "step": 10391
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010192030432346298,
+      "loss": 0.8687,
+      "step": 10392
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010188373053317973,
+      "loss": 0.9149,
+      "step": 10393
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010184716106344753,
+      "loss": 0.8891,
+      "step": 10394
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010181059591587685,
+      "loss": 0.8436,
+      "step": 10395
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010177403509207777,
+      "loss": 0.9081,
+      "step": 10396
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010173747859366037,
+      "loss": 0.8124,
+      "step": 10397
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010170092642223423,
+      "loss": 0.8637,
+      "step": 10398
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010166437857940903,
+      "loss": 0.9288,
+      "step": 10399
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001016278350667942,
+      "loss": 0.8402,
+      "step": 10400
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010159129588599871,
+      "loss": 0.9121,
+      "step": 10401
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001015547610386319,
+      "loss": 0.9097,
+      "step": 10402
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010151823052630234,
+      "loss": 0.8879,
+      "step": 10403
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001014817043506187,
+      "loss": 0.8261,
+      "step": 10404
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010144518251318951,
+      "loss": 0.8814,
+      "step": 10405
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010140866501562287,
+      "loss": 0.9464,
+      "step": 10406
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010137215185952688,
+      "loss": 0.9504,
+      "step": 10407
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010133564304650942,
+      "loss": 0.8997,
+      "step": 10408
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010129913857817821,
+      "loss": 0.89,
+      "step": 10409
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010126263845614057,
+      "loss": 0.8717,
+      "step": 10410
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010122614268200388,
+      "loss": 0.9197,
+      "step": 10411
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001011896512573753,
+      "loss": 0.9156,
+      "step": 10412
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010115316418386149,
+      "loss": 0.9351,
+      "step": 10413
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010111668146306947,
+      "loss": 0.918,
+      "step": 10414
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010108020309660555,
+      "loss": 0.8829,
+      "step": 10415
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001010437290860762,
+      "loss": 0.8673,
+      "step": 10416
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010100725943308738,
+      "loss": 0.9005,
+      "step": 10417
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010097079413924513,
+      "loss": 0.8616,
+      "step": 10418
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010093433320615525,
+      "loss": 0.9103,
+      "step": 10419
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010089787663542307,
+      "loss": 0.9277,
+      "step": 10420
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0001008614244286543,
+      "loss": 0.8719,
+      "step": 10421
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010082497658745384,
+      "loss": 0.9545,
+      "step": 10422
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0001007885331134268,
+      "loss": 0.9444,
+      "step": 10423
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010075209400817795,
+      "loss": 0.8728,
+      "step": 10424
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010071565927331186,
+      "loss": 0.905,
+      "step": 10425
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010067922891043293,
+      "loss": 0.8682,
+      "step": 10426
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010064280292114536,
+      "loss": 0.8756,
+      "step": 10427
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010060638130705329,
+      "loss": 0.8681,
+      "step": 10428
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010056996406976038,
+      "loss": 0.8227,
+      "step": 10429
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010053355121087033,
+      "loss": 0.9259,
+      "step": 10430
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010049714273198667,
+      "loss": 0.9098,
+      "step": 10431
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010046073863471242,
+      "loss": 0.8702,
+      "step": 10432
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0001004243389206509,
+      "loss": 0.922,
+      "step": 10433
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0001003879435914048,
+      "loss": 0.9272,
+      "step": 10434
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010035155264857685,
+      "loss": 0.8509,
+      "step": 10435
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010031516609376956,
+      "loss": 0.904,
+      "step": 10436
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010027878392858512,
+      "loss": 0.8867,
+      "step": 10437
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010024240615462566,
+      "loss": 0.8966,
+      "step": 10438
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010020603277349308,
+      "loss": 0.8547,
+      "step": 10439
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010016966378678916,
+      "loss": 0.9052,
+      "step": 10440
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010013329919611526,
+      "loss": 0.8932,
+      "step": 10441
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010009693900307278,
+      "loss": 0.8535,
+      "step": 10442
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010006058320926288,
+      "loss": 0.9284,
+      "step": 10443
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010002423181628637,
+      "loss": 0.9002,
+      "step": 10444
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.998788482574408e-05,
+      "loss": 0.9087,
+      "step": 10445
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.995154223923651e-05,
+      "loss": 0.8879,
+      "step": 10446
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.991520405836407e-05,
+      "loss": 0.8876,
+      "step": 10447
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.98788702847268e-05,
+      "loss": 0.9045,
+      "step": 10448
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.984254091992473e-05,
+      "loss": 0.8947,
+      "step": 10449
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.980621596555768e-05,
+      "loss": 0.9164,
+      "step": 10450
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.976989542322501e-05,
+      "loss": 0.8972,
+      "step": 10451
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.973357929452639e-05,
+      "loss": 0.9204,
+      "step": 10452
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.969726758106077e-05,
+      "loss": 0.8834,
+      "step": 10453
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.966096028442722e-05,
+      "loss": 0.8599,
+      "step": 10454
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.96246574062246e-05,
+      "loss": 0.9445,
+      "step": 10455
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.958835894805136e-05,
+      "loss": 0.868,
+      "step": 10456
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.955206491150596e-05,
+      "loss": 0.8666,
+      "step": 10457
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.951577529818668e-05,
+      "loss": 0.9134,
+      "step": 10458
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.94794901096915e-05,
+      "loss": 0.9152,
+      "step": 10459
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.94432093476182e-05,
+      "loss": 0.9035,
+      "step": 10460
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.940693301356439e-05,
+      "loss": 0.9388,
+      "step": 10461
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.93706611091276e-05,
+      "loss": 0.872,
+      "step": 10462
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.933439363590487e-05,
+      "loss": 0.871,
+      "step": 10463
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.929813059549351e-05,
+      "loss": 0.9357,
+      "step": 10464
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.926187198949013e-05,
+      "loss": 0.9049,
+      "step": 10465
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.922561781949149e-05,
+      "loss": 0.9317,
+      "step": 10466
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.918936808709406e-05,
+      "loss": 0.8386,
+      "step": 10467
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.915312279389402e-05,
+      "loss": 0.8983,
+      "step": 10468
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.911688194148754e-05,
+      "loss": 0.8869,
+      "step": 10469
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.908064553147024e-05,
+      "loss": 0.847,
+      "step": 10470
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.904441356543814e-05,
+      "loss": 0.8761,
+      "step": 10471
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.900818604498645e-05,
+      "loss": 0.9087,
+      "step": 10472
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.897196297171056e-05,
+      "loss": 0.9662,
+      "step": 10473
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.893574434720558e-05,
+      "loss": 0.8893,
+      "step": 10474
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.88995301730663e-05,
+      "loss": 0.8886,
+      "step": 10475
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.886332045088747e-05,
+      "loss": 0.7925,
+      "step": 10476
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.882711518226357e-05,
+      "loss": 0.8656,
+      "step": 10477
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.879091436878897e-05,
+      "loss": 0.9178,
+      "step": 10478
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.875471801205764e-05,
+      "loss": 0.9542,
+      "step": 10479
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.871852611366359e-05,
+      "loss": 0.8907,
+      "step": 10480
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.868233867520054e-05,
+      "loss": 0.8366,
+      "step": 10481
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.864615569826184e-05,
+      "loss": 0.914,
+      "step": 10482
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.860997718444105e-05,
+      "loss": 0.9181,
+      "step": 10483
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.857380313533111e-05,
+      "loss": 0.9177,
+      "step": 10484
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.853763355252501e-05,
+      "loss": 0.9156,
+      "step": 10485
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.850146843761556e-05,
+      "loss": 0.8802,
+      "step": 10486
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.846530779219512e-05,
+      "loss": 0.8377,
+      "step": 10487
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.842915161785613e-05,
+      "loss": 0.8488,
+      "step": 10488
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.839299991619069e-05,
+      "loss": 0.8208,
+      "step": 10489
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.835685268879085e-05,
+      "loss": 0.8963,
+      "step": 10490
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.832070993724816e-05,
+      "loss": 0.9734,
+      "step": 10491
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.82845716631543e-05,
+      "loss": 0.9208,
+      "step": 10492
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.824843786810065e-05,
+      "loss": 0.8542,
+      "step": 10493
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.821230855367823e-05,
+      "loss": 0.9144,
+      "step": 10494
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.817618372147806e-05,
+      "loss": 0.8435,
+      "step": 10495
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.814006337309089e-05,
+      "loss": 0.9564,
+      "step": 10496
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.810394751010741e-05,
+      "loss": 0.923,
+      "step": 10497
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.806783613411776e-05,
+      "loss": 0.8869,
+      "step": 10498
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.803172924671222e-05,
+      "loss": 0.8812,
+      "step": 10499
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.799562684948082e-05,
+      "loss": 0.9193,
+      "step": 10500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.79595289440131e-05,
+      "loss": 0.8872,
+      "step": 10501
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.792343553189896e-05,
+      "loss": 0.9262,
+      "step": 10502
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.788734661472751e-05,
+      "loss": 0.9114,
+      "step": 10503
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.785126219408802e-05,
+      "loss": 0.9095,
+      "step": 10504
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.781518227156954e-05,
+      "loss": 0.8502,
+      "step": 10505
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.777910684876073e-05,
+      "loss": 0.9782,
+      "step": 10506
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.774303592725018e-05,
+      "loss": 0.8392,
+      "step": 10507
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.770696950862632e-05,
+      "loss": 0.8615,
+      "step": 10508
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.767090759447742e-05,
+      "loss": 0.7738,
+      "step": 10509
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.76348501863913e-05,
+      "loss": 0.8587,
+      "step": 10510
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.75987972859558e-05,
+      "loss": 0.8926,
+      "step": 10511
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.756274889475862e-05,
+      "loss": 0.9363,
+      "step": 10512
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.752670501438692e-05,
+      "loss": 0.8443,
+      "step": 10513
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.749066564642817e-05,
+      "loss": 0.9566,
+      "step": 10514
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.745463079246917e-05,
+      "loss": 0.94,
+      "step": 10515
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.741860045409676e-05,
+      "loss": 0.8371,
+      "step": 10516
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.738257463289765e-05,
+      "loss": 0.8633,
+      "step": 10517
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.734655333045806e-05,
+      "loss": 0.918,
+      "step": 10518
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.731053654836424e-05,
+      "loss": 0.9046,
+      "step": 10519
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.727452428820225e-05,
+      "loss": 0.8801,
+      "step": 10520
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.723851655155791e-05,
+      "loss": 0.9616,
+      "step": 10521
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.720251334001672e-05,
+      "loss": 0.866,
+      "step": 10522
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.71665146551641e-05,
+      "loss": 0.8786,
+      "step": 10523
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.713052049858538e-05,
+      "loss": 0.8556,
+      "step": 10524
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.709453087186538e-05,
+      "loss": 0.9012,
+      "step": 10525
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.7058545776589e-05,
+      "loss": 0.9269,
+      "step": 10526
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.702256521434083e-05,
+      "loss": 0.8783,
+      "step": 10527
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.698658918670536e-05,
+      "loss": 0.8773,
+      "step": 10528
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.695061769526664e-05,
+      "loss": 0.9432,
+      "step": 10529
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.691465074160873e-05,
+      "loss": 0.8783,
+      "step": 10530
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.687868832731553e-05,
+      "loss": 0.8549,
+      "step": 10531
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.684273045397043e-05,
+      "loss": 0.9483,
+      "step": 10532
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.68067771231571e-05,
+      "loss": 0.8616,
+      "step": 10533
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.677082833645858e-05,
+      "loss": 0.9814,
+      "step": 10534
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.673488409545792e-05,
+      "loss": 0.849,
+      "step": 10535
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.669894440173797e-05,
+      "loss": 0.9118,
+      "step": 10536
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.666300925688121e-05,
+      "loss": 0.8557,
+      "step": 10537
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.662707866247014e-05,
+      "loss": 0.8483,
+      "step": 10538
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.659115262008694e-05,
+      "loss": 0.8846,
+      "step": 10539
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.655523113131369e-05,
+      "loss": 0.8522,
+      "step": 10540
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.651931419773203e-05,
+      "loss": 0.8505,
+      "step": 10541
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.648340182092365e-05,
+      "loss": 0.8884,
+      "step": 10542
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.644749400247007e-05,
+      "loss": 0.8718,
+      "step": 10543
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.641159074395219e-05,
+      "loss": 0.9533,
+      "step": 10544
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.637569204695136e-05,
+      "loss": 0.8724,
+      "step": 10545
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.633979791304812e-05,
+      "loss": 0.9504,
+      "step": 10546
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.630390834382319e-05,
+      "loss": 0.8858,
+      "step": 10547
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.626802334085703e-05,
+      "loss": 0.8674,
+      "step": 10548
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.623214290572966e-05,
+      "loss": 0.8738,
+      "step": 10549
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.619626704002122e-05,
+      "loss": 0.8671,
+      "step": 10550
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.616039574531133e-05,
+      "loss": 0.8702,
+      "step": 10551
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.612452902317983e-05,
+      "loss": 0.9577,
+      "step": 10552
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.608866687520593e-05,
+      "loss": 0.9692,
+      "step": 10553
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.605280930296885e-05,
+      "loss": 0.9021,
+      "step": 10554
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.60169563080477e-05,
+      "loss": 0.9576,
+      "step": 10555
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.598110789202108e-05,
+      "loss": 0.9059,
+      "step": 10556
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.594526405646767e-05,
+      "loss": 0.8469,
+      "step": 10557
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.590942480296583e-05,
+      "loss": 0.8682,
+      "step": 10558
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.587359013309385e-05,
+      "loss": 0.8452,
+      "step": 10559
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.583776004842955e-05,
+      "loss": 0.9271,
+      "step": 10560
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.580193455055077e-05,
+      "loss": 0.92,
+      "step": 10561
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.576611364103519e-05,
+      "loss": 0.9228,
+      "step": 10562
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.573029732145993e-05,
+      "loss": 0.8926,
+      "step": 10563
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.569448559340247e-05,
+      "loss": 0.8919,
+      "step": 10564
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.565867845843955e-05,
+      "loss": 0.8515,
+      "step": 10565
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.562287591814803e-05,
+      "loss": 0.9315,
+      "step": 10566
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.558707797410451e-05,
+      "loss": 0.9405,
+      "step": 10567
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.555128462788527e-05,
+      "loss": 0.8557,
+      "step": 10568
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.551549588106647e-05,
+      "loss": 0.8954,
+      "step": 10569
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.547971173522414e-05,
+      "loss": 0.8775,
+      "step": 10570
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.544393219193406e-05,
+      "loss": 0.8749,
+      "step": 10571
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.540815725277165e-05,
+      "loss": 0.8954,
+      "step": 10572
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.537238691931234e-05,
+      "loss": 0.8652,
+      "step": 10573
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.533662119313134e-05,
+      "loss": 0.8754,
+      "step": 10574
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.530086007580345e-05,
+      "loss": 0.8208,
+      "step": 10575
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.526510356890348e-05,
+      "loss": 0.8842,
+      "step": 10576
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.522935167400599e-05,
+      "loss": 0.88,
+      "step": 10577
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.519360439268536e-05,
+      "loss": 0.8261,
+      "step": 10578
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.51578617265156e-05,
+      "loss": 0.9297,
+      "step": 10579
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.512212367707068e-05,
+      "loss": 0.9149,
+      "step": 10580
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.508639024592443e-05,
+      "loss": 0.9118,
+      "step": 10581
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.505066143465014e-05,
+      "loss": 0.8913,
+      "step": 10582
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.501493724482142e-05,
+      "loss": 0.9019,
+      "step": 10583
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.497921767801117e-05,
+      "loss": 0.8163,
+      "step": 10584
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.494350273579238e-05,
+      "loss": 0.9252,
+      "step": 10585
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.490779241973782e-05,
+      "loss": 0.8762,
+      "step": 10586
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.487208673141985e-05,
+      "loss": 0.8558,
+      "step": 10587
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.483638567241085e-05,
+      "loss": 0.8979,
+      "step": 10588
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.480068924428289e-05,
+      "loss": 0.9289,
+      "step": 10589
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.476499744860798e-05,
+      "loss": 0.8811,
+      "step": 10590
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.472931028695763e-05,
+      "loss": 0.86,
+      "step": 10591
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.469362776090342e-05,
+      "loss": 0.9009,
+      "step": 10592
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.465794987201668e-05,
+      "loss": 0.815,
+      "step": 10593
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.462227662186829e-05,
+      "loss": 0.8586,
+      "step": 10594
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.458660801202939e-05,
+      "loss": 0.856,
+      "step": 10595
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.455094404407045e-05,
+      "loss": 0.909,
+      "step": 10596
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.451528471956199e-05,
+      "loss": 0.8754,
+      "step": 10597
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.447963004007437e-05,
+      "loss": 0.9406,
+      "step": 10598
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.444398000717745e-05,
+      "loss": 0.8872,
+      "step": 10599
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.440833462244119e-05,
+      "loss": 0.888,
+      "step": 10600
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.437269388743522e-05,
+      "loss": 0.8272,
+      "step": 10601
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.433705780372908e-05,
+      "loss": 0.9756,
+      "step": 10602
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.43014263728918e-05,
+      "loss": 0.9201,
+      "step": 10603
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.426579959649255e-05,
+      "loss": 0.8891,
+      "step": 10604
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.42301774761002e-05,
+      "loss": 0.8187,
+      "step": 10605
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.419456001328322e-05,
+      "loss": 0.942,
+      "step": 10606
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.415894720961012e-05,
+      "loss": 0.9095,
+      "step": 10607
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.412333906664905e-05,
+      "loss": 0.9018,
+      "step": 10608
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.408773558596818e-05,
+      "loss": 0.8784,
+      "step": 10609
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.405213676913507e-05,
+      "loss": 0.9156,
+      "step": 10610
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.401654261771746e-05,
+      "loss": 0.9325,
+      "step": 10611
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.39809531332828e-05,
+      "loss": 0.8757,
+      "step": 10612
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.394536831739799e-05,
+      "loss": 0.9122,
+      "step": 10613
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.390978817163039e-05,
+      "loss": 0.8283,
+      "step": 10614
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.387421269754648e-05,
+      "loss": 0.9981,
+      "step": 10615
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.383864189671294e-05,
+      "loss": 0.8768,
+      "step": 10616
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.38030757706962e-05,
+      "loss": 0.9502,
+      "step": 10617
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.376751432106227e-05,
+      "loss": 0.8408,
+      "step": 10618
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.373195754937715e-05,
+      "loss": 0.8538,
+      "step": 10619
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.369640545720662e-05,
+      "loss": 0.9073,
+      "step": 10620
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.366085804611626e-05,
+      "loss": 0.8751,
+      "step": 10621
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.362531531767125e-05,
+      "loss": 0.9608,
+      "step": 10622
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.35897772734368e-05,
+      "loss": 0.9592,
+      "step": 10623
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.355424391497793e-05,
+      "loss": 0.9149,
+      "step": 10624
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.351871524385906e-05,
+      "loss": 0.9383,
+      "step": 10625
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.348319126164507e-05,
+      "loss": 0.8837,
+      "step": 10626
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.344767196989999e-05,
+      "loss": 0.9672,
+      "step": 10627
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.341215737018807e-05,
+      "loss": 0.8925,
+      "step": 10628
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.337664746407305e-05,
+      "loss": 0.8435,
+      "step": 10629
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.334114225311868e-05,
+      "loss": 0.9327,
+      "step": 10630
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.33056417388885e-05,
+      "loss": 0.8649,
+      "step": 10631
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.327014592294557e-05,
+      "loss": 0.866,
+      "step": 10632
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.323465480685323e-05,
+      "loss": 0.8701,
+      "step": 10633
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.319916839217413e-05,
+      "loss": 0.8689,
+      "step": 10634
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.316368668047098e-05,
+      "loss": 0.9071,
+      "step": 10635
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.312820967330627e-05,
+      "loss": 0.8595,
+      "step": 10636
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.309273737224209e-05,
+      "loss": 0.8986,
+      "step": 10637
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.305726977884057e-05,
+      "loss": 0.8796,
+      "step": 10638
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.302180689466349e-05,
+      "loss": 0.9064,
+      "step": 10639
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.298634872127254e-05,
+      "loss": 0.9373,
+      "step": 10640
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.2950895260229e-05,
+      "loss": 0.9014,
+      "step": 10641
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.291544651309409e-05,
+      "loss": 0.9104,
+      "step": 10642
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.288000248142891e-05,
+      "loss": 0.9579,
+      "step": 10643
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.2844563166794e-05,
+      "loss": 0.8905,
+      "step": 10644
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.280912857075024e-05,
+      "loss": 0.8717,
+      "step": 10645
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.277369869485775e-05,
+      "loss": 0.8972,
+      "step": 10646
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.273827354067674e-05,
+      "loss": 0.8558,
+      "step": 10647
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.27028531097673e-05,
+      "loss": 0.9627,
+      "step": 10648
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.266743740368898e-05,
+      "loss": 0.8068,
+      "step": 10649
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.263202642400138e-05,
+      "loss": 0.9147,
+      "step": 10650
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.259662017226381e-05,
+      "loss": 0.9124,
+      "step": 10651
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.256121865003549e-05,
+      "loss": 0.877,
+      "step": 10652
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.252582185887514e-05,
+      "loss": 0.9031,
+      "step": 10653
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.249042980034159e-05,
+      "loss": 0.8613,
+      "step": 10654
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.245504247599337e-05,
+      "loss": 0.8223,
+      "step": 10655
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.241965988738861e-05,
+      "loss": 0.9065,
+      "step": 10656
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.238428203608547e-05,
+      "loss": 0.9221,
+      "step": 10657
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.234890892364181e-05,
+      "loss": 0.9544,
+      "step": 10658
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.231354055161535e-05,
+      "loss": 0.8927,
+      "step": 10659
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.227817692156338e-05,
+      "loss": 0.8333,
+      "step": 10660
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.224281803504322e-05,
+      "loss": 0.8931,
+      "step": 10661
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.220746389361201e-05,
+      "loss": 0.9108,
+      "step": 10662
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.21721144988263e-05,
+      "loss": 0.8501,
+      "step": 10663
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.2136769852243e-05,
+      "loss": 0.8008,
+      "step": 10664
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.210142995541833e-05,
+      "loss": 0.8846,
+      "step": 10665
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.206609480990851e-05,
+      "loss": 0.9095,
+      "step": 10666
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.203076441726963e-05,
+      "loss": 0.8613,
+      "step": 10667
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.199543877905732e-05,
+      "loss": 0.8333,
+      "step": 10668
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.196011789682717e-05,
+      "loss": 0.9672,
+      "step": 10669
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.192480177213459e-05,
+      "loss": 0.8583,
+      "step": 10670
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.188949040653479e-05,
+      "loss": 0.8719,
+      "step": 10671
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.185418380158253e-05,
+      "loss": 0.9105,
+      "step": 10672
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.181888195883262e-05,
+      "loss": 0.9141,
+      "step": 10673
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.178358487983967e-05,
+      "loss": 0.928,
+      "step": 10674
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.174829256615776e-05,
+      "loss": 0.9341,
+      "step": 10675
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.171300501934129e-05,
+      "loss": 0.8569,
+      "step": 10676
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.167772224094389e-05,
+      "loss": 0.9634,
+      "step": 10677
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.164244423251933e-05,
+      "loss": 0.8635,
+      "step": 10678
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.160717099562118e-05,
+      "loss": 0.9126,
+      "step": 10679
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.15719025318025e-05,
+      "loss": 0.838,
+      "step": 10680
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.153663884261645e-05,
+      "loss": 0.8666,
+      "step": 10681
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.150137992961587e-05,
+      "loss": 0.8714,
+      "step": 10682
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.146612579435343e-05,
+      "loss": 0.9239,
+      "step": 10683
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.143087643838142e-05,
+      "loss": 0.8815,
+      "step": 10684
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.139563186325211e-05,
+      "loss": 0.8673,
+      "step": 10685
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.136039207051756e-05,
+      "loss": 0.9002,
+      "step": 10686
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.132515706172944e-05,
+      "loss": 0.8993,
+      "step": 10687
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.128992683843937e-05,
+      "loss": 0.8807,
+      "step": 10688
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.125470140219872e-05,
+      "loss": 0.8673,
+      "step": 10689
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.121948075455873e-05,
+      "loss": 0.9281,
+      "step": 10690
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.118426489707017e-05,
+      "loss": 0.9093,
+      "step": 10691
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.114905383128385e-05,
+      "loss": 0.8973,
+      "step": 10692
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.111384755875036e-05,
+      "loss": 0.8511,
+      "step": 10693
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.107864608101982e-05,
+      "loss": 0.9284,
+      "step": 10694
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.10434493996426e-05,
+      "loss": 0.8192,
+      "step": 10695
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.100825751616835e-05,
+      "loss": 0.9254,
+      "step": 10696
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.097307043214682e-05,
+      "loss": 0.9359,
+      "step": 10697
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.093788814912758e-05,
+      "loss": 0.8176,
+      "step": 10698
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.090271066865972e-05,
+      "loss": 0.9031,
+      "step": 10699
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.086753799229235e-05,
+      "loss": 0.9274,
+      "step": 10700
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.083237012157429e-05,
+      "loss": 0.9098,
+      "step": 10701
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.079720705805423e-05,
+      "loss": 0.8611,
+      "step": 10702
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.076204880328045e-05,
+      "loss": 0.8766,
+      "step": 10703
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.07268953588012e-05,
+      "loss": 0.9124,
+      "step": 10704
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.069174672616456e-05,
+      "loss": 0.8668,
+      "step": 10705
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.065660290691804e-05,
+      "loss": 0.9115,
+      "step": 10706
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.062146390260952e-05,
+      "loss": 0.922,
+      "step": 10707
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.058632971478613e-05,
+      "loss": 0.9644,
+      "step": 10708
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.055120034499513e-05,
+      "loss": 0.8866,
+      "step": 10709
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.051607579478334e-05,
+      "loss": 0.8827,
+      "step": 10710
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.048095606569749e-05,
+      "loss": 0.9032,
+      "step": 10711
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.044584115928416e-05,
+      "loss": 0.8516,
+      "step": 10712
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.041073107708943e-05,
+      "loss": 0.8583,
+      "step": 10713
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.037562582065969e-05,
+      "loss": 0.9184,
+      "step": 10714
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.034052539154051e-05,
+      "loss": 0.9312,
+      "step": 10715
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.030542979127769e-05,
+      "loss": 0.9321,
+      "step": 10716
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.027033902141668e-05,
+      "loss": 0.8852,
+      "step": 10717
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.02352530835026e-05,
+      "loss": 0.8379,
+      "step": 10718
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.020017197908051e-05,
+      "loss": 0.8864,
+      "step": 10719
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.01650957096952e-05,
+      "loss": 0.8488,
+      "step": 10720
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.013002427689135e-05,
+      "loss": 0.9006,
+      "step": 10721
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.009495768221316e-05,
+      "loss": 0.8832,
+      "step": 10722
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.005989592720487e-05,
+      "loss": 0.9144,
+      "step": 10723
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.002483901341052e-05,
+      "loss": 0.8598,
+      "step": 10724
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 8.998978694237359e-05,
+      "loss": 0.9198,
+      "step": 10725
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 8.995473971563788e-05,
+      "loss": 0.8913,
+      "step": 10726
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 8.991969733474652e-05,
+      "loss": 0.9437,
+      "step": 10727
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 8.988465980124268e-05,
+      "loss": 0.8692,
+      "step": 10728
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.984962711666925e-05,
+      "loss": 0.8968,
+      "step": 10729
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.981459928256879e-05,
+      "loss": 0.9675,
+      "step": 10730
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.977957630048381e-05,
+      "loss": 0.9101,
+      "step": 10731
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.974455817195657e-05,
+      "loss": 0.8723,
+      "step": 10732
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.970954489852916e-05,
+      "loss": 0.8068,
+      "step": 10733
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.967453648174322e-05,
+      "loss": 0.826,
+      "step": 10734
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.963953292314045e-05,
+      "loss": 0.8193,
+      "step": 10735
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.960453422426227e-05,
+      "loss": 0.8514,
+      "step": 10736
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.956954038664976e-05,
+      "loss": 0.8507,
+      "step": 10737
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.953455141184385e-05,
+      "loss": 0.9455,
+      "step": 10738
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.949956730138538e-05,
+      "loss": 0.8268,
+      "step": 10739
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.94645880568149e-05,
+      "loss": 0.9187,
+      "step": 10740
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.94296136796726e-05,
+      "loss": 0.902,
+      "step": 10741
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.93946441714986e-05,
+      "loss": 0.8561,
+      "step": 10742
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.935967953383293e-05,
+      "loss": 0.845,
+      "step": 10743
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.932471976821497e-05,
+      "loss": 0.905,
+      "step": 10744
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.92897648761845e-05,
+      "loss": 0.898,
+      "step": 10745
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.925481485928054e-05,
+      "loss": 0.8734,
+      "step": 10746
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.921986971904217e-05,
+      "loss": 0.9382,
+      "step": 10747
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.91849294570083e-05,
+      "loss": 0.9046,
+      "step": 10748
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.914999407471735e-05,
+      "loss": 0.87,
+      "step": 10749
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.911506357370778e-05,
+      "loss": 0.9276,
+      "step": 10750
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.908013795551777e-05,
+      "loss": 0.886,
+      "step": 10751
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.904521722168535e-05,
+      "loss": 0.9289,
+      "step": 10752
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.901030137374809e-05,
+      "loss": 0.9264,
+      "step": 10753
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.897539041324357e-05,
+      "loss": 0.8517,
+      "step": 10754
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.894048434170918e-05,
+      "loss": 0.852,
+      "step": 10755
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.890558316068183e-05,
+      "loss": 0.8405,
+      "step": 10756
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.887068687169862e-05,
+      "loss": 1.0114,
+      "step": 10757
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.883579547629601e-05,
+      "loss": 0.8735,
+      "step": 10758
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.880090897601057e-05,
+      "loss": 0.8399,
+      "step": 10759
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.876602737237852e-05,
+      "loss": 0.9018,
+      "step": 10760
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.873115066693576e-05,
+      "loss": 0.8536,
+      "step": 10761
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.869627886121825e-05,
+      "loss": 0.8795,
+      "step": 10762
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.866141195676132e-05,
+      "loss": 0.9413,
+      "step": 10763
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.862654995510067e-05,
+      "loss": 0.9224,
+      "step": 10764
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.859169285777118e-05,
+      "loss": 0.8823,
+      "step": 10765
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.855684066630787e-05,
+      "loss": 0.9417,
+      "step": 10766
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.852199338224556e-05,
+      "loss": 0.9219,
+      "step": 10767
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.848715100711857e-05,
+      "loss": 0.8854,
+      "step": 10768
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.845231354246127e-05,
+      "loss": 0.8991,
+      "step": 10769
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.841748098980773e-05,
+      "loss": 0.9266,
+      "step": 10770
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.838265335069186e-05,
+      "loss": 0.9105,
+      "step": 10771
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.834783062664718e-05,
+      "loss": 0.9322,
+      "step": 10772
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.831301281920717e-05,
+      "loss": 0.8237,
+      "step": 10773
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.827819992990507e-05,
+      "loss": 0.8901,
+      "step": 10774
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.824339196027369e-05,
+      "loss": 0.8742,
+      "step": 10775
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.82085889118461e-05,
+      "loss": 0.8336,
+      "step": 10776
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.817379078615457e-05,
+      "loss": 0.852,
+      "step": 10777
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.813899758473158e-05,
+      "loss": 0.9322,
+      "step": 10778
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.810420930910928e-05,
+      "loss": 0.828,
+      "step": 10779
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.806942596081945e-05,
+      "loss": 0.8594,
+      "step": 10780
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.803464754139383e-05,
+      "loss": 0.9224,
+      "step": 10781
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.79998740523639e-05,
+      "loss": 0.9455,
+      "step": 10782
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.796510549526095e-05,
+      "loss": 0.9235,
+      "step": 10783
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.793034187161593e-05,
+      "loss": 0.9202,
+      "step": 10784
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.78955831829597e-05,
+      "loss": 0.8787,
+      "step": 10785
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.78608294308229e-05,
+      "loss": 0.8921,
+      "step": 10786
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.782608061673574e-05,
+      "loss": 0.863,
+      "step": 10787
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.779133674222865e-05,
+      "loss": 0.8765,
+      "step": 10788
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.775659780883139e-05,
+      "loss": 0.8979,
+      "step": 10789
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.772186381807379e-05,
+      "loss": 0.8826,
+      "step": 10790
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.768713477148525e-05,
+      "loss": 0.9108,
+      "step": 10791
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.765241067059511e-05,
+      "loss": 0.9262,
+      "step": 10792
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.761769151693253e-05,
+      "loss": 0.8888,
+      "step": 10793
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.758297731202616e-05,
+      "loss": 0.9407,
+      "step": 10794
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.754826805740492e-05,
+      "loss": 0.9332,
+      "step": 10795
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.751356375459703e-05,
+      "loss": 0.9108,
+      "step": 10796
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.747886440513076e-05,
+      "loss": 0.8887,
+      "step": 10797
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.744417001053416e-05,
+      "loss": 0.9367,
+      "step": 10798
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.740948057233485e-05,
+      "loss": 0.8984,
+      "step": 10799
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.737479609206044e-05,
+      "loss": 0.8929,
+      "step": 10800
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.734011657123833e-05,
+      "loss": 0.849,
+      "step": 10801
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.730544201139563e-05,
+      "loss": 0.8998,
+      "step": 10802
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.727077241405912e-05,
+      "loss": 0.8735,
+      "step": 10803
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.723610778075553e-05,
+      "loss": 0.8424,
+      "step": 10804
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.720144811301143e-05,
+      "loss": 0.8964,
+      "step": 10805
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.716679341235281e-05,
+      "loss": 0.8858,
+      "step": 10806
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.713214368030598e-05,
+      "loss": 0.8744,
+      "step": 10807
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.709749891839654e-05,
+      "loss": 0.8895,
+      "step": 10808
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.706285912815013e-05,
+      "loss": 0.913,
+      "step": 10809
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.702822431109218e-05,
+      "loss": 0.9342,
+      "step": 10810
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.699359446874773e-05,
+      "loss": 0.9146,
+      "step": 10811
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.695896960264174e-05,
+      "loss": 0.9339,
+      "step": 10812
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.692434971429892e-05,
+      "loss": 0.9063,
+      "step": 10813
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.688973480524385e-05,
+      "loss": 0.9157,
+      "step": 10814
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.685512487700062e-05,
+      "loss": 0.8532,
+      "step": 10815
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.682051993109339e-05,
+      "loss": 0.9389,
+      "step": 10816
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.678591996904601e-05,
+      "loss": 0.8762,
+      "step": 10817
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.675132499238202e-05,
+      "loss": 0.8681,
+      "step": 10818
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.67167350026248e-05,
+      "loss": 0.9038,
+      "step": 10819
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.668215000129758e-05,
+      "loss": 0.8444,
+      "step": 10820
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.664756998992336e-05,
+      "loss": 0.894,
+      "step": 10821
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.661299497002474e-05,
+      "loss": 0.9789,
+      "step": 10822
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.657842494312426e-05,
+      "loss": 0.9437,
+      "step": 10823
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.654385991074436e-05,
+      "loss": 0.8666,
+      "step": 10824
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.650929987440684e-05,
+      "loss": 0.8645,
+      "step": 10825
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.647474483563384e-05,
+      "loss": 0.8592,
+      "step": 10826
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.644019479594678e-05,
+      "loss": 0.9163,
+      "step": 10827
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.640564975686718e-05,
+      "loss": 0.8562,
+      "step": 10828
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.637110971991626e-05,
+      "loss": 0.8843,
+      "step": 10829
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.633657468661487e-05,
+      "loss": 0.8945,
+      "step": 10830
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.630204465848382e-05,
+      "loss": 0.8361,
+      "step": 10831
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.626751963704362e-05,
+      "loss": 0.88,
+      "step": 10832
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.62329996238147e-05,
+      "loss": 0.8396,
+      "step": 10833
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.619848462031698e-05,
+      "loss": 0.8765,
+      "step": 10834
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.61639746280704e-05,
+      "loss": 0.8628,
+      "step": 10835
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.612946964859467e-05,
+      "loss": 0.877,
+      "step": 10836
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.609496968340904e-05,
+      "loss": 0.9257,
+      "step": 10837
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.606047473403295e-05,
+      "loss": 0.9603,
+      "step": 10838
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.602598480198523e-05,
+      "loss": 0.8779,
+      "step": 10839
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.599149988878466e-05,
+      "loss": 0.8657,
+      "step": 10840
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.595701999594987e-05,
+      "loss": 0.8953,
+      "step": 10841
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.592254512499906e-05,
+      "loss": 0.8466,
+      "step": 10842
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.588807527745047e-05,
+      "loss": 0.8609,
+      "step": 10843
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.585361045482175e-05,
+      "loss": 0.8656,
+      "step": 10844
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.581915065863086e-05,
+      "loss": 0.8761,
+      "step": 10845
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.578469589039502e-05,
+      "loss": 0.8627,
+      "step": 10846
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.575024615163149e-05,
+      "loss": 0.9299,
+      "step": 10847
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.571580144385738e-05,
+      "loss": 0.8575,
+      "step": 10848
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.56813617685893e-05,
+      "loss": 0.8862,
+      "step": 10849
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.564692712734385e-05,
+      "loss": 0.8902,
+      "step": 10850
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.56124975216374e-05,
+      "loss": 0.9325,
+      "step": 10851
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.557807295298613e-05,
+      "loss": 0.9346,
+      "step": 10852
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.554365342290574e-05,
+      "loss": 0.8325,
+      "step": 10853
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.5509238932912e-05,
+      "loss": 0.9278,
+      "step": 10854
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.547482948452042e-05,
+      "loss": 0.9358,
+      "step": 10855
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.544042507924601e-05,
+      "loss": 0.9442,
+      "step": 10856
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.540602571860403e-05,
+      "loss": 0.8941,
+      "step": 10857
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.537163140410907e-05,
+      "loss": 0.8939,
+      "step": 10858
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.533724213727572e-05,
+      "loss": 0.9213,
+      "step": 10859
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.530285791961843e-05,
+      "loss": 0.9316,
+      "step": 10860
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.526847875265114e-05,
+      "loss": 0.9175,
+      "step": 10861
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.523410463788781e-05,
+      "loss": 0.9396,
+      "step": 10862
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.519973557684211e-05,
+      "loss": 0.9087,
+      "step": 10863
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.516537157102756e-05,
+      "loss": 0.8908,
+      "step": 10864
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.51310126219572e-05,
+      "loss": 0.9191,
+      "step": 10865
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.509665873114414e-05,
+      "loss": 0.8706,
+      "step": 10866
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.50623099001012e-05,
+      "loss": 0.8907,
+      "step": 10867
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.502796613034074e-05,
+      "loss": 0.8513,
+      "step": 10868
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.499362742337536e-05,
+      "loss": 0.9293,
+      "step": 10869
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.495929378071694e-05,
+      "loss": 0.8969,
+      "step": 10870
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.492496520387752e-05,
+      "loss": 0.8412,
+      "step": 10871
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.489064169436862e-05,
+      "loss": 0.8762,
+      "step": 10872
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.485632325370174e-05,
+      "loss": 0.9162,
+      "step": 10873
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.482200988338817e-05,
+      "loss": 0.9223,
+      "step": 10874
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.478770158493868e-05,
+      "loss": 0.9186,
+      "step": 10875
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.475339835986431e-05,
+      "loss": 0.897,
+      "step": 10876
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.471910020967542e-05,
+      "loss": 0.8464,
+      "step": 10877
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.468480713588239e-05,
+      "loss": 0.8943,
+      "step": 10878
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.465051913999535e-05,
+      "loss": 0.9223,
+      "step": 10879
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.461623622352408e-05,
+      "loss": 0.8526,
+      "step": 10880
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.45819583879783e-05,
+      "loss": 0.9137,
+      "step": 10881
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.454768563486736e-05,
+      "loss": 0.8839,
+      "step": 10882
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.451341796570062e-05,
+      "loss": 0.9305,
+      "step": 10883
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.44791553819869e-05,
+      "loss": 0.9061,
+      "step": 10884
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.444489788523497e-05,
+      "loss": 0.9363,
+      "step": 10885
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.441064547695348e-05,
+      "loss": 0.9258,
+      "step": 10886
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.43763981586505e-05,
+      "loss": 0.7869,
+      "step": 10887
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.434215593183443e-05,
+      "loss": 0.9158,
+      "step": 10888
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.430791879801287e-05,
+      "loss": 0.9228,
+      "step": 10889
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.427368675869353e-05,
+      "loss": 0.8305,
+      "step": 10890
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.423945981538392e-05,
+      "loss": 0.8851,
+      "step": 10891
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.420523796959108e-05,
+      "loss": 0.8542,
+      "step": 10892
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.4171021222822e-05,
+      "loss": 0.8582,
+      "step": 10893
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.413680957658343e-05,
+      "loss": 0.9214,
+      "step": 10894
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.410260303238198e-05,
+      "loss": 0.8731,
+      "step": 10895
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.406840159172375e-05,
+      "loss": 0.8903,
+      "step": 10896
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.40342052561149e-05,
+      "loss": 0.8388,
+      "step": 10897
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.400001402706134e-05,
+      "loss": 0.8859,
+      "step": 10898
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.396582790606853e-05,
+      "loss": 0.9343,
+      "step": 10899
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.393164689464192e-05,
+      "loss": 0.8682,
+      "step": 10900
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.389747099428669e-05,
+      "loss": 0.8805,
+      "step": 10901
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.386330020650784e-05,
+      "loss": 0.8267,
+      "step": 10902
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.382913453280991e-05,
+      "loss": 0.8891,
+      "step": 10903
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.379497397469752e-05,
+      "loss": 0.9103,
+      "step": 10904
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.376081853367495e-05,
+      "loss": 0.9136,
+      "step": 10905
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.372666821124602e-05,
+      "loss": 0.9455,
+      "step": 10906
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.369252300891486e-05,
+      "loss": 0.9123,
+      "step": 10907
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.36583829281848e-05,
+      "loss": 0.832,
+      "step": 10908
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.36242479705593e-05,
+      "loss": 0.8383,
+      "step": 10909
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.359011813754156e-05,
+      "loss": 0.8772,
+      "step": 10910
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.355599343063434e-05,
+      "loss": 0.8987,
+      "step": 10911
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.35218738513404e-05,
+      "loss": 0.8911,
+      "step": 10912
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.348775940116217e-05,
+      "loss": 0.972,
+      "step": 10913
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.3453650081602e-05,
+      "loss": 0.8941,
+      "step": 10914
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.341954589416172e-05,
+      "loss": 0.8761,
+      "step": 10915
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.338544684034317e-05,
+      "loss": 0.9351,
+      "step": 10916
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.3351352921648e-05,
+      "loss": 0.8805,
+      "step": 10917
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.33172641395773e-05,
+      "loss": 0.8881,
+      "step": 10918
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.328318049563246e-05,
+      "loss": 0.8277,
+      "step": 10919
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.324910199131415e-05,
+      "loss": 0.9353,
+      "step": 10920
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.321502862812309e-05,
+      "loss": 0.8273,
+      "step": 10921
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.318096040755975e-05,
+      "loss": 0.9253,
+      "step": 10922
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.31468973311242e-05,
+      "loss": 0.8673,
+      "step": 10923
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.311283940031656e-05,
+      "loss": 0.8688,
+      "step": 10924
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.307878661663632e-05,
+      "loss": 0.9224,
+      "step": 10925
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.304473898158334e-05,
+      "loss": 0.859,
+      "step": 10926
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.301069649665662e-05,
+      "loss": 0.9251,
+      "step": 10927
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.297665916335535e-05,
+      "loss": 0.865,
+      "step": 10928
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.294262698317842e-05,
+      "loss": 0.9316,
+      "step": 10929
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.29085999576243e-05,
+      "loss": 0.8731,
+      "step": 10930
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.287457808819143e-05,
+      "loss": 0.8677,
+      "step": 10931
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.284056137637797e-05,
+      "loss": 0.904,
+      "step": 10932
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.28065498236819e-05,
+      "loss": 0.9193,
+      "step": 10933
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.277254343160078e-05,
+      "loss": 0.8933,
+      "step": 10934
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.273854220163219e-05,
+      "loss": 0.8452,
+      "step": 10935
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.270454613527342e-05,
+      "loss": 0.9336,
+      "step": 10936
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.267055523402127e-05,
+      "loss": 0.9511,
+      "step": 10937
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.263656949937282e-05,
+      "loss": 0.8932,
+      "step": 10938
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.260258893282442e-05,
+      "loss": 0.8938,
+      "step": 10939
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.256861353587247e-05,
+      "loss": 0.8564,
+      "step": 10940
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.253464331001315e-05,
+      "loss": 0.8811,
+      "step": 10941
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.250067825674219e-05,
+      "loss": 0.8897,
+      "step": 10942
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.246671837755532e-05,
+      "loss": 0.8709,
+      "step": 10943
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.243276367394799e-05,
+      "loss": 0.9264,
+      "step": 10944
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.239881414741541e-05,
+      "loss": 0.9194,
+      "step": 10945
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.236486979945246e-05,
+      "loss": 0.8503,
+      "step": 10946
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.233093063155392e-05,
+      "loss": 0.8732,
+      "step": 10947
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.22969966452144e-05,
+      "loss": 0.892,
+      "step": 10948
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.226306784192793e-05,
+      "loss": 0.901,
+      "step": 10949
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.222914422318885e-05,
+      "loss": 0.8976,
+      "step": 10950
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.219522579049083e-05,
+      "loss": 0.7836,
+      "step": 10951
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.216131254532758e-05,
+      "loss": 0.8673,
+      "step": 10952
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.21274044891923e-05,
+      "loss": 0.8571,
+      "step": 10953
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.209350162357823e-05,
+      "loss": 0.8364,
+      "step": 10954
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.205960394997836e-05,
+      "loss": 0.8905,
+      "step": 10955
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.202571146988514e-05,
+      "loss": 0.8639,
+      "step": 10956
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.199182418479132e-05,
+      "loss": 0.9127,
+      "step": 10957
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.195794209618892e-05,
+      "loss": 0.8642,
+      "step": 10958
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.192406520557001e-05,
+      "loss": 0.8332,
+      "step": 10959
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.189019351442641e-05,
+      "loss": 0.9075,
+      "step": 10960
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.185632702424957e-05,
+      "loss": 0.8844,
+      "step": 10961
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.18224657365308e-05,
+      "loss": 0.8966,
+      "step": 10962
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.178860965276123e-05,
+      "loss": 0.852,
+      "step": 10963
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.175475877443177e-05,
+      "loss": 0.8723,
+      "step": 10964
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.172091310303292e-05,
+      "loss": 0.9405,
+      "step": 10965
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.168707264005511e-05,
+      "loss": 0.9098,
+      "step": 10966
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.165323738698858e-05,
+      "loss": 0.8996,
+      "step": 10967
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.16194073453231e-05,
+      "loss": 0.9094,
+      "step": 10968
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.158558251654861e-05,
+      "loss": 0.9129,
+      "step": 10969
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.15517629021544e-05,
+      "loss": 0.8816,
+      "step": 10970
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.151794850362978e-05,
+      "loss": 0.8653,
+      "step": 10971
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.148413932246384e-05,
+      "loss": 0.8563,
+      "step": 10972
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.145033536014521e-05,
+      "loss": 0.8751,
+      "step": 10973
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.141653661816255e-05,
+      "loss": 0.8368,
+      "step": 10974
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.138274309800418e-05,
+      "loss": 0.8475,
+      "step": 10975
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.134895480115826e-05,
+      "loss": 0.9345,
+      "step": 10976
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.131517172911249e-05,
+      "loss": 0.8706,
+      "step": 10977
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.128139388335463e-05,
+      "loss": 0.8492,
+      "step": 10978
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.12476212653721e-05,
+      "loss": 0.8277,
+      "step": 10979
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.121385387665201e-05,
+      "loss": 0.8447,
+      "step": 10980
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.118009171868133e-05,
+      "loss": 0.9293,
+      "step": 10981
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.114633479294678e-05,
+      "loss": 0.879,
+      "step": 10982
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.111258310093492e-05,
+      "loss": 0.8175,
+      "step": 10983
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.107883664413188e-05,
+      "loss": 0.8881,
+      "step": 10984
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.104509542402373e-05,
+      "loss": 0.8719,
+      "step": 10985
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.101135944209634e-05,
+      "loss": 0.8234,
+      "step": 10986
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.097762869983509e-05,
+      "loss": 0.866,
+      "step": 10987
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.09439031987256e-05,
+      "loss": 0.888,
+      "step": 10988
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.09101829402527e-05,
+      "loss": 0.839,
+      "step": 10989
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.08764679259014e-05,
+      "loss": 0.8966,
+      "step": 10990
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.084275815715637e-05,
+      "loss": 0.8174,
+      "step": 10991
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.080905363550189e-05,
+      "loss": 0.9713,
+      "step": 10992
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.077535436242223e-05,
+      "loss": 0.8849,
+      "step": 10993
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.074166033940129e-05,
+      "loss": 0.877,
+      "step": 10994
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.070797156792291e-05,
+      "loss": 0.8916,
+      "step": 10995
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.067428804947042e-05,
+      "loss": 0.9097,
+      "step": 10996
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.064060978552713e-05,
+      "loss": 0.842,
+      "step": 10997
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.060693677757612e-05,
+      "loss": 0.876,
+      "step": 10998
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.057326902709998e-05,
+      "loss": 0.849,
+      "step": 10999
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.053960653558159e-05,
+      "loss": 0.8325,
+      "step": 11000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.050594930450304e-05,
+      "loss": 0.8274,
+      "step": 11001
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.047229733534649e-05,
+      "loss": 0.8986,
+      "step": 11002
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.043865062959388e-05,
+      "loss": 0.852,
+      "step": 11003
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.040500918872671e-05,
+      "loss": 0.8335,
+      "step": 11004
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.03713730142265e-05,
+      "loss": 0.9087,
+      "step": 11005
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.033774210757423e-05,
+      "loss": 0.8881,
+      "step": 11006
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.030411647025111e-05,
+      "loss": 0.8153,
+      "step": 11007
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.027049610373763e-05,
+      "loss": 0.9052,
+      "step": 11008
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.023688100951435e-05,
+      "loss": 0.9207,
+      "step": 11009
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.020327118906157e-05,
+      "loss": 0.9575,
+      "step": 11010
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.016966664385916e-05,
+      "loss": 0.8808,
+      "step": 11011
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.013606737538697e-05,
+      "loss": 0.8708,
+      "step": 11012
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.010247338512453e-05,
+      "loss": 0.9239,
+      "step": 11013
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.006888467455122e-05,
+      "loss": 0.863,
+      "step": 11014
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.003530124514602e-05,
+      "loss": 0.9261,
+      "step": 11015
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.00017230983878e-05,
+      "loss": 0.8997,
+      "step": 11016
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 7.996815023575528e-05,
+      "loss": 0.9765,
+      "step": 11017
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 7.99345826587266e-05,
+      "loss": 0.9614,
+      "step": 11018
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 7.990102036878022e-05,
+      "loss": 0.936,
+      "step": 11019
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 7.986746336739381e-05,
+      "loss": 0.8317,
+      "step": 11020
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 7.983391165604515e-05,
+      "loss": 0.9097,
+      "step": 11021
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 7.980036523621175e-05,
+      "loss": 0.9064,
+      "step": 11022
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 7.976682410937069e-05,
+      "loss": 0.8571,
+      "step": 11023
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 7.973328827699902e-05,
+      "loss": 0.9144,
+      "step": 11024
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 7.969975774057348e-05,
+      "loss": 0.9172,
+      "step": 11025
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 7.966623250157068e-05,
+      "loss": 0.925,
+      "step": 11026
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 7.963271256146677e-05,
+      "loss": 0.8632,
+      "step": 11027
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 7.959919792173785e-05,
+      "loss": 0.8928,
+      "step": 11028
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 7.95656885838598e-05,
+      "loss": 0.8647,
+      "step": 11029
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 7.953218454930808e-05,
+      "loss": 0.9067,
+      "step": 11030
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 7.94986858195581e-05,
+      "loss": 0.8949,
+      "step": 11031
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 7.946519239608501e-05,
+      "loss": 0.8399,
+      "step": 11032
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 7.943170428036373e-05,
+      "loss": 0.8936,
+      "step": 11033
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 7.939822147386877e-05,
+      "loss": 0.8883,
+      "step": 11034
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 7.936474397807463e-05,
+      "loss": 0.928,
+      "step": 11035
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 7.933127179445557e-05,
+      "loss": 0.9279,
+      "step": 11036
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 7.929780492448528e-05,
+      "loss": 0.923,
+      "step": 11037
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.926434336963782e-05,
+      "loss": 0.9105,
+      "step": 11038
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.923088713138642e-05,
+      "loss": 0.8763,
+      "step": 11039
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.919743621120441e-05,
+      "loss": 0.9188,
+      "step": 11040
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.916399061056485e-05,
+      "loss": 0.9277,
+      "step": 11041
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.913055033094039e-05,
+      "loss": 0.8924,
+      "step": 11042
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.909711537380367e-05,
+      "loss": 0.9153,
+      "step": 11043
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.906368574062695e-05,
+      "loss": 0.8366,
+      "step": 11044
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.903026143288242e-05,
+      "loss": 0.9567,
+      "step": 11045
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.899684245204174e-05,
+      "loss": 0.862,
+      "step": 11046
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.896342879957661e-05,
+      "loss": 0.9454,
+      "step": 11047
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.893002047695847e-05,
+      "loss": 0.8738,
+      "step": 11048
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.889661748565824e-05,
+      "loss": 0.8796,
+      "step": 11049
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.88632198271471e-05,
+      "loss": 0.9424,
+      "step": 11050
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.882982750289551e-05,
+      "loss": 0.8471,
+      "step": 11051
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.879644051437396e-05,
+      "loss": 0.8012,
+      "step": 11052
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.876305886305273e-05,
+      "loss": 0.938,
+      "step": 11053
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.872968255040165e-05,
+      "loss": 0.8512,
+      "step": 11054
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.86963115778905e-05,
+      "loss": 0.8659,
+      "step": 11055
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.866294594698878e-05,
+      "loss": 0.8701,
+      "step": 11056
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.862958565916579e-05,
+      "loss": 0.8569,
+      "step": 11057
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.859623071589043e-05,
+      "loss": 0.8158,
+      "step": 11058
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.856288111863159e-05,
+      "loss": 0.8522,
+      "step": 11059
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.852953686885782e-05,
+      "loss": 0.8636,
+      "step": 11060
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.849619796803735e-05,
+      "loss": 0.9251,
+      "step": 11061
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.846286441763832e-05,
+      "loss": 0.9299,
+      "step": 11062
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.842953621912854e-05,
+      "loss": 0.9174,
+      "step": 11063
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.839621337397574e-05,
+      "loss": 0.8582,
+      "step": 11064
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.836289588364709e-05,
+      "loss": 0.8992,
+      "step": 11065
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.832958374960986e-05,
+      "loss": 0.9083,
+      "step": 11066
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.829627697333097e-05,
+      "loss": 0.8596,
+      "step": 11067
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.82629755562769e-05,
+      "loss": 0.9093,
+      "step": 11068
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.822967949991436e-05,
+      "loss": 0.9576,
+      "step": 11069
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.819638880570932e-05,
+      "loss": 0.9738,
+      "step": 11070
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.816310347512779e-05,
+      "loss": 0.9333,
+      "step": 11071
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.81298235096356e-05,
+      "loss": 0.8496,
+      "step": 11072
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.809654891069807e-05,
+      "loss": 0.8246,
+      "step": 11073
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.806327967978053e-05,
+      "loss": 0.9046,
+      "step": 11074
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.803001581834797e-05,
+      "loss": 0.8155,
+      "step": 11075
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.799675732786524e-05,
+      "loss": 0.8904,
+      "step": 11076
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.796350420979677e-05,
+      "loss": 0.8861,
+      "step": 11077
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.79302564656069e-05,
+      "loss": 0.945,
+      "step": 11078
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.789701409675976e-05,
+      "loss": 0.9043,
+      "step": 11079
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.786377710471899e-05,
+      "loss": 0.9164,
+      "step": 11080
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.783054549094844e-05,
+      "loss": 0.8731,
+      "step": 11081
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.779731925691127e-05,
+      "loss": 0.9228,
+      "step": 11082
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.776409840407072e-05,
+      "loss": 0.845,
+      "step": 11083
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.773088293388956e-05,
+      "loss": 0.8462,
+      "step": 11084
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.76976728478305e-05,
+      "loss": 0.853,
+      "step": 11085
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.7664468147356e-05,
+      "loss": 0.8419,
+      "step": 11086
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.7631268833928e-05,
+      "loss": 0.833,
+      "step": 11087
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.759807490900875e-05,
+      "loss": 0.9079,
+      "step": 11088
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.756488637405973e-05,
+      "loss": 0.878,
+      "step": 11089
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.753170323054244e-05,
+      "loss": 0.8543,
+      "step": 11090
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.749852547991819e-05,
+      "loss": 0.8354,
+      "step": 11091
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.746535312364783e-05,
+      "loss": 0.8824,
+      "step": 11092
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.743218616319216e-05,
+      "loss": 0.8868,
+      "step": 11093
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.739902460001169e-05,
+      "loss": 0.8189,
+      "step": 11094
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.736586843556677e-05,
+      "loss": 0.9613,
+      "step": 11095
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.73327176713173e-05,
+      "loss": 0.8893,
+      "step": 11096
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.729957230872308e-05,
+      "loss": 0.8623,
+      "step": 11097
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.726643234924382e-05,
+      "loss": 0.8899,
+      "step": 11098
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.723329779433861e-05,
+      "loss": 0.8302,
+      "step": 11099
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.720016864546678e-05,
+      "loss": 0.8801,
+      "step": 11100
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.716704490408697e-05,
+      "loss": 0.8889,
+      "step": 11101
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.713392657165788e-05,
+      "loss": 0.8472,
+      "step": 11102
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.71008136496379e-05,
+      "loss": 0.8766,
+      "step": 11103
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.706770613948506e-05,
+      "loss": 0.9069,
+      "step": 11104
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.703460404265732e-05,
+      "loss": 0.8521,
+      "step": 11105
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.70015073606123e-05,
+      "loss": 0.8778,
+      "step": 11106
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.69684160948075e-05,
+      "loss": 0.8611,
+      "step": 11107
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.69353302466999e-05,
+      "loss": 0.9037,
+      "step": 11108
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.690224981774668e-05,
+      "loss": 0.8229,
+      "step": 11109
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.686917480940443e-05,
+      "loss": 0.8754,
+      "step": 11110
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.683610522312952e-05,
+      "loss": 0.9576,
+      "step": 11111
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.680304106037823e-05,
+      "loss": 0.8627,
+      "step": 11112
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.67699823226066e-05,
+      "loss": 0.8788,
+      "step": 11113
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.673692901127034e-05,
+      "loss": 0.9278,
+      "step": 11114
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.670388112782492e-05,
+      "loss": 0.8441,
+      "step": 11115
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.66708386737256e-05,
+      "loss": 0.8862,
+      "step": 11116
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.663780165042751e-05,
+      "loss": 0.8429,
+      "step": 11117
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.66047700593852e-05,
+      "loss": 0.8578,
+      "step": 11118
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.657174390205353e-05,
+      "loss": 0.8862,
+      "step": 11119
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.653872317988656e-05,
+      "loss": 0.8838,
+      "step": 11120
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.650570789433846e-05,
+      "loss": 0.8895,
+      "step": 11121
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.64726980468631e-05,
+      "loss": 0.8795,
+      "step": 11122
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.643969363891397e-05,
+      "loss": 0.8768,
+      "step": 11123
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.640669467194445e-05,
+      "loss": 0.9432,
+      "step": 11124
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.637370114740765e-05,
+      "loss": 0.9161,
+      "step": 11125
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.634071306675654e-05,
+      "loss": 0.8744,
+      "step": 11126
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.630773043144352e-05,
+      "loss": 0.9098,
+      "step": 11127
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.627475324292128e-05,
+      "loss": 0.9063,
+      "step": 11128
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.62417815026418e-05,
+      "loss": 0.935,
+      "step": 11129
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.620881521205688e-05,
+      "loss": 0.875,
+      "step": 11130
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.617585437261845e-05,
+      "loss": 0.8963,
+      "step": 11131
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.614289898577774e-05,
+      "loss": 0.9138,
+      "step": 11132
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.6109949052986e-05,
+      "loss": 0.8528,
+      "step": 11133
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.607700457569429e-05,
+      "loss": 0.8795,
+      "step": 11134
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.604406555535313e-05,
+      "loss": 0.9809,
+      "step": 11135
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.601113199341308e-05,
+      "loss": 0.8672,
+      "step": 11136
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.597820389132438e-05,
+      "loss": 0.8138,
+      "step": 11137
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.594528125053708e-05,
+      "loss": 0.9099,
+      "step": 11138
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.591236407250077e-05,
+      "loss": 0.8949,
+      "step": 11139
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.587945235866508e-05,
+      "loss": 0.8956,
+      "step": 11140
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.58465461104793e-05,
+      "loss": 0.8774,
+      "step": 11141
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.581364532939233e-05,
+      "loss": 0.9124,
+      "step": 11142
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.578075001685304e-05,
+      "loss": 0.8688,
+      "step": 11143
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.574786017430995e-05,
+      "loss": 0.8775,
+      "step": 11144
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.571497580321145e-05,
+      "loss": 0.8852,
+      "step": 11145
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.568209690500541e-05,
+      "loss": 0.8872,
+      "step": 11146
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.564922348113989e-05,
+      "loss": 0.8941,
+      "step": 11147
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.561635553306237e-05,
+      "loss": 1.0019,
+      "step": 11148
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.558349306222004e-05,
+      "loss": 0.8609,
+      "step": 11149
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.555063607006028e-05,
+      "loss": 0.888,
+      "step": 11150
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.551778455802974e-05,
+      "loss": 0.9355,
+      "step": 11151
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.548493852757507e-05,
+      "loss": 0.8673,
+      "step": 11152
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.54520979801427e-05,
+      "loss": 0.9452,
+      "step": 11153
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.541926291717882e-05,
+      "loss": 0.919,
+      "step": 11154
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.538643334012916e-05,
+      "loss": 0.9148,
+      "step": 11155
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.535360925043945e-05,
+      "loss": 0.9189,
+      "step": 11156
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.532079064955519e-05,
+      "loss": 0.9631,
+      "step": 11157
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.528797753892132e-05,
+      "loss": 0.8957,
+      "step": 11158
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.525516991998303e-05,
+      "loss": 0.9169,
+      "step": 11159
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.52223677941849e-05,
+      "loss": 0.9167,
+      "step": 11160
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.518957116297121e-05,
+      "loss": 0.8575,
+      "step": 11161
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.515678002778643e-05,
+      "loss": 0.8291,
+      "step": 11162
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.512399439007432e-05,
+      "loss": 0.8958,
+      "step": 11163
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.509121425127874e-05,
+      "loss": 0.8786,
+      "step": 11164
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.505843961284295e-05,
+      "loss": 0.8368,
+      "step": 11165
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.502567047621047e-05,
+      "loss": 0.9198,
+      "step": 11166
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.499290684282414e-05,
+      "loss": 0.9013,
+      "step": 11167
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.496014871412657e-05,
+      "loss": 0.8438,
+      "step": 11168
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.492739609156055e-05,
+      "loss": 0.8575,
+      "step": 11169
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.48946489765681e-05,
+      "loss": 0.8518,
+      "step": 11170
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.486190737059136e-05,
+      "loss": 0.8934,
+      "step": 11171
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.482917127507206e-05,
+      "loss": 0.9005,
+      "step": 11172
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.479644069145184e-05,
+      "loss": 0.9272,
+      "step": 11173
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.476371562117183e-05,
+      "loss": 0.8506,
+      "step": 11174
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.473099606567318e-05,
+      "loss": 0.8937,
+      "step": 11175
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.46982820263967e-05,
+      "loss": 0.9221,
+      "step": 11176
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.466557350478283e-05,
+      "loss": 0.8864,
+      "step": 11177
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.46328705022721e-05,
+      "loss": 0.9261,
+      "step": 11178
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.460017302030449e-05,
+      "loss": 0.8635,
+      "step": 11179
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.456748106031965e-05,
+      "loss": 0.9238,
+      "step": 11180
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.45347946237575e-05,
+      "loss": 0.8573,
+      "step": 11181
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.450211371205716e-05,
+      "loss": 0.8508,
+      "step": 11182
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.446943832665777e-05,
+      "loss": 0.9406,
+      "step": 11183
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.443676846899825e-05,
+      "loss": 0.901,
+      "step": 11184
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.440410414051726e-05,
+      "loss": 0.8245,
+      "step": 11185
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.437144534265301e-05,
+      "loss": 0.8732,
+      "step": 11186
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.433879207684375e-05,
+      "loss": 0.806,
+      "step": 11187
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.430614434452741e-05,
+      "loss": 0.8468,
+      "step": 11188
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.42735021471414e-05,
+      "loss": 0.8982,
+      "step": 11189
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.424086548612344e-05,
+      "loss": 0.9059,
+      "step": 11190
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.420823436291044e-05,
+      "loss": 0.9002,
+      "step": 11191
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.41756087789395e-05,
+      "loss": 0.8579,
+      "step": 11192
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.414298873564709e-05,
+      "loss": 0.8905,
+      "step": 11193
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.411037423446975e-05,
+      "loss": 0.8704,
+      "step": 11194
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.407776527684371e-05,
+      "loss": 0.8762,
+      "step": 11195
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.40451618642047e-05,
+      "loss": 0.8232,
+      "step": 11196
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.401256399798869e-05,
+      "loss": 0.883,
+      "step": 11197
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.397997167963098e-05,
+      "loss": 0.8862,
+      "step": 11198
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.394738491056665e-05,
+      "loss": 0.9068,
+      "step": 11199
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.391480369223091e-05,
+      "loss": 0.8722,
+      "step": 11200
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.38822280260583e-05,
+      "loss": 0.8278,
+      "step": 11201
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.384965791348335e-05,
+      "loss": 0.8797,
+      "step": 11202
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.381709335594029e-05,
+      "loss": 0.8952,
+      "step": 11203
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.378453435486314e-05,
+      "loss": 0.8832,
+      "step": 11204
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.375198091168552e-05,
+      "loss": 0.9127,
+      "step": 11205
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.371943302784099e-05,
+      "loss": 0.8763,
+      "step": 11206
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.368689070476285e-05,
+      "loss": 0.9476,
+      "step": 11207
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.365435394388391e-05,
+      "loss": 0.897,
+      "step": 11208
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.362182274663722e-05,
+      "loss": 0.8426,
+      "step": 11209
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.358929711445506e-05,
+      "loss": 0.8967,
+      "step": 11210
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.355677704876976e-05,
+      "loss": 0.8605,
+      "step": 11211
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.352426255101341e-05,
+      "loss": 0.8943,
+      "step": 11212
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.349175362261768e-05,
+      "loss": 0.8984,
+      "step": 11213
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.345925026501415e-05,
+      "loss": 0.8433,
+      "step": 11214
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.342675247963409e-05,
+      "loss": 0.9225,
+      "step": 11215
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.339426026790863e-05,
+      "loss": 0.9121,
+      "step": 11216
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.336177363126848e-05,
+      "loss": 0.9046,
+      "step": 11217
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.33292925711441e-05,
+      "loss": 0.8458,
+      "step": 11218
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.329681708896598e-05,
+      "loss": 0.9055,
+      "step": 11219
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.326434718616406e-05,
+      "loss": 0.8585,
+      "step": 11220
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.323188286416818e-05,
+      "loss": 0.92,
+      "step": 11221
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.31994241244079e-05,
+      "loss": 0.8713,
+      "step": 11222
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.316697096831263e-05,
+      "loss": 0.9651,
+      "step": 11223
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.313452339731129e-05,
+      "loss": 0.8627,
+      "step": 11224
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.31020814128328e-05,
+      "loss": 0.9145,
+      "step": 11225
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.306964501630578e-05,
+      "loss": 0.8862,
+      "step": 11226
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.303721420915837e-05,
+      "loss": 0.8399,
+      "step": 11227
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.300478899281898e-05,
+      "loss": 0.9187,
+      "step": 11228
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.297236936871518e-05,
+      "loss": 0.9428,
+      "step": 11229
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.293995533827469e-05,
+      "loss": 0.8943,
+      "step": 11230
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.290754690292492e-05,
+      "loss": 0.8033,
+      "step": 11231
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.287514406409282e-05,
+      "loss": 0.8879,
+      "step": 11232
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.284274682320533e-05,
+      "loss": 0.9085,
+      "step": 11233
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.281035518168906e-05,
+      "loss": 0.8905,
+      "step": 11234
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.277796914097044e-05,
+      "loss": 0.8643,
+      "step": 11235
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.274558870247548e-05,
+      "loss": 0.9058,
+      "step": 11236
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.271321386763009e-05,
+      "loss": 0.8286,
+      "step": 11237
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.268084463785995e-05,
+      "loss": 0.8428,
+      "step": 11238
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.26484810145903e-05,
+      "loss": 0.9218,
+      "step": 11239
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.261612299924647e-05,
+      "loss": 0.8786,
+      "step": 11240
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.258377059325321e-05,
+      "loss": 0.911,
+      "step": 11241
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.255142379803516e-05,
+      "loss": 0.8428,
+      "step": 11242
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.251908261501685e-05,
+      "loss": 0.9138,
+      "step": 11243
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.248674704562222e-05,
+      "loss": 0.8917,
+      "step": 11244
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.245441709127534e-05,
+      "loss": 0.8849,
+      "step": 11245
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.242209275339967e-05,
+      "loss": 0.8953,
+      "step": 11246
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.238977403341885e-05,
+      "loss": 0.9001,
+      "step": 11247
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.235746093275584e-05,
+      "loss": 0.8086,
+      "step": 11248
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.232515345283363e-05,
+      "loss": 0.846,
+      "step": 11249
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.229285159507495e-05,
+      "loss": 0.8935,
+      "step": 11250
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.226055536090204e-05,
+      "loss": 0.8961,
+      "step": 11251
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.222826475173719e-05,
+      "loss": 0.8545,
+      "step": 11252
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.219597976900228e-05,
+      "loss": 0.9026,
+      "step": 11253
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.216370041411904e-05,
+      "loss": 0.895,
+      "step": 11254
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.213142668850876e-05,
+      "loss": 0.8821,
+      "step": 11255
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.209915859359273e-05,
+      "loss": 0.8648,
+      "step": 11256
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.206689613079187e-05,
+      "loss": 0.8095,
+      "step": 11257
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.203463930152672e-05,
+      "loss": 0.9057,
+      "step": 11258
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.200238810721793e-05,
+      "loss": 0.8652,
+      "step": 11259
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.197014254928552e-05,
+      "loss": 0.8494,
+      "step": 11260
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.193790262914946e-05,
+      "loss": 0.9697,
+      "step": 11261
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.190566834822951e-05,
+      "loss": 0.9082,
+      "step": 11262
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.1873439707945e-05,
+      "loss": 0.856,
+      "step": 11263
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.184121670971513e-05,
+      "loss": 0.9168,
+      "step": 11264
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.180899935495891e-05,
+      "loss": 0.9125,
+      "step": 11265
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.177678764509503e-05,
+      "loss": 0.9012,
+      "step": 11266
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.174458158154185e-05,
+      "loss": 0.8459,
+      "step": 11267
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.17123811657176e-05,
+      "loss": 0.8553,
+      "step": 11268
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.16801863990403e-05,
+      "loss": 0.9302,
+      "step": 11269
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.164799728292746e-05,
+      "loss": 0.8825,
+      "step": 11270
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.161581381879679e-05,
+      "loss": 0.8717,
+      "step": 11271
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.158363600806526e-05,
+      "loss": 0.8843,
+      "step": 11272
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.155146385214999e-05,
+      "loss": 0.8805,
+      "step": 11273
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.151929735246756e-05,
+      "loss": 0.9578,
+      "step": 11274
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.148713651043444e-05,
+      "loss": 0.8955,
+      "step": 11275
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.145498132746693e-05,
+      "loss": 0.826,
+      "step": 11276
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.142283180498079e-05,
+      "loss": 0.8704,
+      "step": 11277
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.139068794439197e-05,
+      "loss": 0.8335,
+      "step": 11278
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.135854974711573e-05,
+      "loss": 0.7773,
+      "step": 11279
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.132641721456737e-05,
+      "loss": 0.9269,
+      "step": 11280
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.129429034816188e-05,
+      "loss": 0.854,
+      "step": 11281
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.126216914931386e-05,
+      "loss": 0.9303,
+      "step": 11282
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.123005361943778e-05,
+      "loss": 0.9118,
+      "step": 11283
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.11979437599479e-05,
+      "loss": 0.8191,
+      "step": 11284
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.116583957225826e-05,
+      "loss": 0.8276,
+      "step": 11285
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.113374105778238e-05,
+      "loss": 0.9328,
+      "step": 11286
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.110164821793382e-05,
+      "loss": 0.834,
+      "step": 11287
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.106956105412585e-05,
+      "loss": 0.9687,
+      "step": 11288
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.103747956777123e-05,
+      "loss": 0.8729,
+      "step": 11289
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.100540376028293e-05,
+      "loss": 0.9197,
+      "step": 11290
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.09733336330732e-05,
+      "loss": 0.8288,
+      "step": 11291
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.094126918755432e-05,
+      "loss": 0.8237,
+      "step": 11292
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.090921042513831e-05,
+      "loss": 0.8506,
+      "step": 11293
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.08771573472368e-05,
+      "loss": 0.9705,
+      "step": 11294
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.084510995526124e-05,
+      "loss": 0.8754,
+      "step": 11295
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.081306825062286e-05,
+      "loss": 0.8385,
+      "step": 11296
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.078103223473269e-05,
+      "loss": 0.8691,
+      "step": 11297
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.074900190900131e-05,
+      "loss": 0.9463,
+      "step": 11298
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.071697727483924e-05,
+      "loss": 0.874,
+      "step": 11299
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.068495833365674e-05,
+      "loss": 0.8808,
+      "step": 11300
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.065294508686361e-05,
+      "loss": 0.8746,
+      "step": 11301
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.06209375358697e-05,
+      "loss": 0.8567,
+      "step": 11302
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.058893568208438e-05,
+      "loss": 0.8858,
+      "step": 11303
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.055693952691695e-05,
+      "loss": 0.8725,
+      "step": 11304
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.052494907177625e-05,
+      "loss": 0.8745,
+      "step": 11305
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.0492964318071e-05,
+      "loss": 0.9086,
+      "step": 11306
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.046098526720975e-05,
+      "loss": 0.837,
+      "step": 11307
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.042901192060049e-05,
+      "loss": 0.8642,
+      "step": 11308
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.039704427965144e-05,
+      "loss": 0.9547,
+      "step": 11309
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.036508234577006e-05,
+      "loss": 0.8954,
+      "step": 11310
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.033312612036391e-05,
+      "loss": 0.8897,
+      "step": 11311
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.030117560484022e-05,
+      "loss": 0.8729,
+      "step": 11312
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.026923080060581e-05,
+      "loss": 0.8703,
+      "step": 11313
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.023729170906741e-05,
+      "loss": 0.8878,
+      "step": 11314
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.020535833163151e-05,
+      "loss": 0.8359,
+      "step": 11315
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.017343066970432e-05,
+      "loss": 0.907,
+      "step": 11316
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.014150872469165e-05,
+      "loss": 0.8606,
+      "step": 11317
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.010959249799926e-05,
+      "loss": 0.8937,
+      "step": 11318
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.007768199103264e-05,
+      "loss": 0.896,
+      "step": 11319
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.004577720519678e-05,
+      "loss": 0.9348,
+      "step": 11320
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.001387814189685e-05,
+      "loss": 0.8702,
+      "step": 11321
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.998198480253738e-05,
+      "loss": 0.8164,
+      "step": 11322
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.995009718852279e-05,
+      "loss": 0.8955,
+      "step": 11323
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.991821530125735e-05,
+      "loss": 0.8394,
+      "step": 11324
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.988633914214486e-05,
+      "loss": 0.8506,
+      "step": 11325
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.98544687125891e-05,
+      "loss": 0.8845,
+      "step": 11326
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.982260401399329e-05,
+      "loss": 0.8421,
+      "step": 11327
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.979074504776087e-05,
+      "loss": 0.8163,
+      "step": 11328
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.975889181529454e-05,
+      "loss": 0.8771,
+      "step": 11329
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.972704431799702e-05,
+      "loss": 0.9099,
+      "step": 11330
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.969520255727078e-05,
+      "loss": 0.9163,
+      "step": 11331
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.966336653451787e-05,
+      "loss": 0.9039,
+      "step": 11332
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.963153625114021e-05,
+      "loss": 0.8643,
+      "step": 11333
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.959971170853947e-05,
+      "loss": 0.8287,
+      "step": 11334
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.956789290811712e-05,
+      "loss": 0.8985,
+      "step": 11335
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.953607985127415e-05,
+      "loss": 0.8728,
+      "step": 11336
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.950427253941152e-05,
+      "loss": 0.9357,
+      "step": 11337
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.947247097392994e-05,
+      "loss": 0.8702,
+      "step": 11338
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.94406751562296e-05,
+      "loss": 0.9177,
+      "step": 11339
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.940888508771086e-05,
+      "loss": 0.8526,
+      "step": 11340
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.937710076977344e-05,
+      "loss": 0.9118,
+      "step": 11341
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.934532220381699e-05,
+      "loss": 0.8711,
+      "step": 11342
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.931354939124097e-05,
+      "loss": 0.9455,
+      "step": 11343
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.928178233344438e-05,
+      "loss": 0.8415,
+      "step": 11344
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.925002103182609e-05,
+      "loss": 0.8536,
+      "step": 11345
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.921826548778477e-05,
+      "loss": 0.8882,
+      "step": 11346
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.918651570271881e-05,
+      "loss": 0.9351,
+      "step": 11347
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.91547716780262e-05,
+      "loss": 0.86,
+      "step": 11348
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.912303341510483e-05,
+      "loss": 0.8752,
+      "step": 11349
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.909130091535239e-05,
+      "loss": 0.8477,
+      "step": 11350
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.905957418016604e-05,
+      "loss": 0.8755,
+      "step": 11351
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.9027853210943e-05,
+      "loss": 0.8871,
+      "step": 11352
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.899613800908005e-05,
+      "loss": 0.9187,
+      "step": 11353
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.896442857597388e-05,
+      "loss": 0.8189,
+      "step": 11354
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.893272491302063e-05,
+      "loss": 0.9219,
+      "step": 11355
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.890102702161647e-05,
+      "loss": 0.8899,
+      "step": 11356
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.886933490315728e-05,
+      "loss": 0.8915,
+      "step": 11357
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.883764855903843e-05,
+      "loss": 0.7973,
+      "step": 11358
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.880596799065546e-05,
+      "loss": 0.8598,
+      "step": 11359
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.877429319940328e-05,
+      "loss": 0.9032,
+      "step": 11360
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.874262418667672e-05,
+      "loss": 0.8339,
+      "step": 11361
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.871096095387039e-05,
+      "loss": 0.8356,
+      "step": 11362
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.867930350237847e-05,
+      "loss": 0.8561,
+      "step": 11363
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.864765183359506e-05,
+      "loss": 0.8287,
+      "step": 11364
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.861600594891391e-05,
+      "loss": 0.8941,
+      "step": 11365
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.858436584972862e-05,
+      "loss": 0.8872,
+      "step": 11366
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.855273153743239e-05,
+      "loss": 0.9331,
+      "step": 11367
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.852110301341823e-05,
+      "loss": 0.9314,
+      "step": 11368
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.8489480279079e-05,
+      "loss": 0.9463,
+      "step": 11369
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.845786333580699e-05,
+      "loss": 0.8956,
+      "step": 11370
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.842625218499477e-05,
+      "loss": 0.9318,
+      "step": 11371
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.839464682803407e-05,
+      "loss": 0.8214,
+      "step": 11372
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.836304726631675e-05,
+      "loss": 0.9368,
+      "step": 11373
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.833145350123433e-05,
+      "loss": 0.9266,
+      "step": 11374
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.829986553417793e-05,
+      "loss": 0.8182,
+      "step": 11375
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.826828336653858e-05,
+      "loss": 0.9663,
+      "step": 11376
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.823670699970702e-05,
+      "loss": 0.8887,
+      "step": 11377
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.820513643507373e-05,
+      "loss": 0.8799,
+      "step": 11378
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.817357167402887e-05,
+      "loss": 0.8907,
+      "step": 11379
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.814201271796241e-05,
+      "loss": 0.8641,
+      "step": 11380
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.81104595682641e-05,
+      "loss": 0.8963,
+      "step": 11381
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.80789122263233e-05,
+      "loss": 0.8933,
+      "step": 11382
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.804737069352922e-05,
+      "loss": 0.837,
+      "step": 11383
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.801583497127081e-05,
+      "loss": 0.907,
+      "step": 11384
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.798430506093683e-05,
+      "loss": 0.8785,
+      "step": 11385
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.795278096391553e-05,
+      "loss": 0.8779,
+      "step": 11386
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.792126268159516e-05,
+      "loss": 0.8729,
+      "step": 11387
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.788975021536368e-05,
+      "loss": 0.92,
+      "step": 11388
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.785824356660857e-05,
+      "loss": 0.8662,
+      "step": 11389
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.782674273671749e-05,
+      "loss": 0.9003,
+      "step": 11390
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.779524772707735e-05,
+      "loss": 0.8535,
+      "step": 11391
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.776375853907513e-05,
+      "loss": 0.8569,
+      "step": 11392
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.77322751740975e-05,
+      "loss": 0.8057,
+      "step": 11393
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.770079763353072e-05,
+      "loss": 0.9184,
+      "step": 11394
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.766932591876094e-05,
+      "loss": 0.9132,
+      "step": 11395
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.763786003117405e-05,
+      "loss": 0.8834,
+      "step": 11396
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.76063999721557e-05,
+      "loss": 0.8882,
+      "step": 11397
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.757494574309109e-05,
+      "loss": 0.8954,
+      "step": 11398
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.75434973453654e-05,
+      "loss": 0.8785,
+      "step": 11399
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.751205478036351e-05,
+      "loss": 0.8785,
+      "step": 11400
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.748061804946981e-05,
+      "loss": 0.8878,
+      "step": 11401
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.744918715406886e-05,
+      "loss": 0.9255,
+      "step": 11402
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.741776209554453e-05,
+      "loss": 0.8346,
+      "step": 11403
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.738634287528072e-05,
+      "loss": 0.8773,
+      "step": 11404
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.735492949466097e-05,
+      "loss": 0.807,
+      "step": 11405
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.73235219550685e-05,
+      "loss": 0.908,
+      "step": 11406
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.729212025788647e-05,
+      "loss": 0.8649,
+      "step": 11407
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.726072440449742e-05,
+      "loss": 0.8014,
+      "step": 11408
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.722933439628416e-05,
+      "loss": 0.9272,
+      "step": 11409
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.719795023462874e-05,
+      "loss": 0.8663,
+      "step": 11410
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.71665719209132e-05,
+      "loss": 0.8267,
+      "step": 11411
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.71351994565194e-05,
+      "loss": 0.9408,
+      "step": 11412
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.710383284282869e-05,
+      "loss": 0.8873,
+      "step": 11413
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.707247208122234e-05,
+      "loss": 0.9342,
+      "step": 11414
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.704111717308132e-05,
+      "loss": 0.8673,
+      "step": 11415
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.700976811978645e-05,
+      "loss": 0.8748,
+      "step": 11416
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.697842492271799e-05,
+      "loss": 0.8767,
+      "step": 11417
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.694708758325628e-05,
+      "loss": 0.8551,
+      "step": 11418
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.691575610278125e-05,
+      "loss": 0.8888,
+      "step": 11419
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.688443048267244e-05,
+      "loss": 0.8876,
+      "step": 11420
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.685311072430951e-05,
+      "loss": 0.8532,
+      "step": 11421
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.682179682907141e-05,
+      "loss": 0.7646,
+      "step": 11422
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.679048879833718e-05,
+      "loss": 0.8542,
+      "step": 11423
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.675918663348548e-05,
+      "loss": 0.8967,
+      "step": 11424
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.672789033589457e-05,
+      "loss": 0.8614,
+      "step": 11425
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.669659990694268e-05,
+      "loss": 0.9849,
+      "step": 11426
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.666531534800766e-05,
+      "loss": 0.8824,
+      "step": 11427
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.663403666046721e-05,
+      "loss": 0.8639,
+      "step": 11428
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.660276384569855e-05,
+      "loss": 0.8697,
+      "step": 11429
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.657149690507882e-05,
+      "loss": 0.7858,
+      "step": 11430
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.654023583998495e-05,
+      "loss": 0.844,
+      "step": 11431
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.650898065179338e-05,
+      "loss": 0.9385,
+      "step": 11432
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.647773134188051e-05,
+      "loss": 0.8826,
+      "step": 11433
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.644648791162241e-05,
+      "loss": 0.8786,
+      "step": 11434
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.641525036239493e-05,
+      "loss": 0.9269,
+      "step": 11435
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.63840186955735e-05,
+      "loss": 0.8688,
+      "step": 11436
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.635279291253345e-05,
+      "loss": 0.8436,
+      "step": 11437
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.632157301464988e-05,
+      "loss": 0.923,
+      "step": 11438
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.629035900329739e-05,
+      "loss": 0.8549,
+      "step": 11439
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.62591508798507e-05,
+      "loss": 0.881,
+      "step": 11440
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.622794864568394e-05,
+      "loss": 0.8908,
+      "step": 11441
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.619675230217108e-05,
+      "loss": 0.8466,
+      "step": 11442
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.616556185068599e-05,
+      "loss": 0.8283,
+      "step": 11443
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.613437729260197e-05,
+      "loss": 0.8045,
+      "step": 11444
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.610319862929231e-05,
+      "loss": 0.8609,
+      "step": 11445
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.607202586212994e-05,
+      "loss": 0.8991,
+      "step": 11446
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.604085899248767e-05,
+      "loss": 0.8896,
+      "step": 11447
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.600969802173775e-05,
+      "loss": 0.8981,
+      "step": 11448
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.597854295125245e-05,
+      "loss": 0.9024,
+      "step": 11449
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.594739378240374e-05,
+      "loss": 0.9382,
+      "step": 11450
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.591625051656307e-05,
+      "loss": 0.8341,
+      "step": 11451
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.58851131551021e-05,
+      "loss": 0.8971,
+      "step": 11452
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.585398169939179e-05,
+      "loss": 0.8724,
+      "step": 11453
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.582285615080304e-05,
+      "loss": 0.8958,
+      "step": 11454
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.579173651070656e-05,
+      "loss": 0.9041,
+      "step": 11455
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.576062278047254e-05,
+      "loss": 0.8534,
+      "step": 11456
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.572951496147119e-05,
+      "loss": 0.9193,
+      "step": 11457
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.569841305507232e-05,
+      "loss": 0.8669,
+      "step": 11458
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.566731706264555e-05,
+      "loss": 0.8838,
+      "step": 11459
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.563622698556006e-05,
+      "loss": 0.9208,
+      "step": 11460
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.5605142825185e-05,
+      "loss": 0.9191,
+      "step": 11461
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.557406458288921e-05,
+      "loss": 0.9467,
+      "step": 11462
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.554299226004108e-05,
+      "loss": 0.9082,
+      "step": 11463
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.551192585800896e-05,
+      "loss": 0.8674,
+      "step": 11464
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.548086537816084e-05,
+      "loss": 0.9195,
+      "step": 11465
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.544981082186458e-05,
+      "loss": 0.9295,
+      "step": 11466
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.541876219048749e-05,
+      "loss": 0.9036,
+      "step": 11467
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.538771948539688e-05,
+      "loss": 0.858,
+      "step": 11468
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.535668270795976e-05,
+      "loss": 0.8244,
+      "step": 11469
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.532565185954266e-05,
+      "loss": 0.8736,
+      "step": 11470
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.52946269415123e-05,
+      "loss": 0.8823,
+      "step": 11471
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.526360795523465e-05,
+      "loss": 0.9331,
+      "step": 11472
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.523259490207567e-05,
+      "loss": 0.8442,
+      "step": 11473
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.520158778340114e-05,
+      "loss": 0.8934,
+      "step": 11474
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.517058660057629e-05,
+      "loss": 0.8866,
+      "step": 11475
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.513959135496633e-05,
+      "loss": 0.8266,
+      "step": 11476
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.510860204793612e-05,
+      "loss": 0.8827,
+      "step": 11477
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.507761868085038e-05,
+      "loss": 0.913,
+      "step": 11478
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.504664125507332e-05,
+      "loss": 0.8516,
+      "step": 11479
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.501566977196907e-05,
+      "loss": 0.9009,
+      "step": 11480
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.498470423290153e-05,
+      "loss": 0.8437,
+      "step": 11481
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.495374463923412e-05,
+      "loss": 0.8504,
+      "step": 11482
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.492279099233034e-05,
+      "loss": 0.9049,
+      "step": 11483
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.489184329355307e-05,
+      "loss": 0.8508,
+      "step": 11484
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.486090154426523e-05,
+      "loss": 0.9052,
+      "step": 11485
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.48299657458292e-05,
+      "loss": 0.8962,
+      "step": 11486
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.479903589960732e-05,
+      "loss": 0.8797,
+      "step": 11487
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.476811200696162e-05,
+      "loss": 0.9085,
+      "step": 11488
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.473719406925367e-05,
+      "loss": 0.8257,
+      "step": 11489
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.470628208784519e-05,
+      "loss": 0.914,
+      "step": 11490
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.467537606409717e-05,
+      "loss": 0.8204,
+      "step": 11491
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.464447599937067e-05,
+      "loss": 0.892,
+      "step": 11492
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.46135818950264e-05,
+      "loss": 0.8827,
+      "step": 11493
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.458269375242467e-05,
+      "loss": 0.9136,
+      "step": 11494
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.45518115729257e-05,
+      "loss": 0.8607,
+      "step": 11495
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.45209353578894e-05,
+      "loss": 0.9078,
+      "step": 11496
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.449006510867544e-05,
+      "loss": 0.8583,
+      "step": 11497
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.445920082664308e-05,
+      "loss": 0.9071,
+      "step": 11498
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.442834251315148e-05,
+      "loss": 0.8839,
+      "step": 11499
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.43974901695596e-05,
+      "loss": 0.8919,
+      "step": 11500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.436664379722576e-05,
+      "loss": 0.8405,
+      "step": 11501
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.43358033975086e-05,
+      "loss": 0.9297,
+      "step": 11502
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.430496897176594e-05,
+      "loss": 0.8868,
+      "step": 11503
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.427414052135568e-05,
+      "loss": 0.842,
+      "step": 11504
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.424331804763538e-05,
+      "loss": 0.8143,
+      "step": 11505
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.421250155196217e-05,
+      "loss": 0.8767,
+      "step": 11506
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.418169103569316e-05,
+      "loss": 0.9362,
+      "step": 11507
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.415088650018506e-05,
+      "loss": 0.8718,
+      "step": 11508
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.412008794679442e-05,
+      "loss": 0.8725,
+      "step": 11509
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.408929537687734e-05,
+      "loss": 0.7869,
+      "step": 11510
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.40585087917898e-05,
+      "loss": 0.82,
+      "step": 11511
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.402772819288761e-05,
+      "loss": 0.9275,
+      "step": 11512
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.399695358152603e-05,
+      "loss": 0.8501,
+      "step": 11513
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.396618495906026e-05,
+      "loss": 0.8512,
+      "step": 11514
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.393542232684524e-05,
+      "loss": 0.8614,
+      "step": 11515
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.390466568623566e-05,
+      "loss": 0.8872,
+      "step": 11516
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.387391503858573e-05,
+      "loss": 0.9213,
+      "step": 11517
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.384317038524966e-05,
+      "loss": 0.8383,
+      "step": 11518
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.381243172758133e-05,
+      "loss": 0.8554,
+      "step": 11519
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.378169906693412e-05,
+      "loss": 0.8701,
+      "step": 11520
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.375097240466161e-05,
+      "loss": 0.8472,
+      "step": 11521
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.372025174211666e-05,
+      "loss": 0.8693,
+      "step": 11522
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.368953708065213e-05,
+      "loss": 0.8789,
+      "step": 11523
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.365882842162059e-05,
+      "loss": 0.9452,
+      "step": 11524
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.362812576637416e-05,
+      "loss": 0.8962,
+      "step": 11525
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.35974291162649e-05,
+      "loss": 0.8376,
+      "step": 11526
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.356673847264453e-05,
+      "loss": 0.8756,
+      "step": 11527
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.353605383686462e-05,
+      "loss": 0.8859,
+      "step": 11528
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.35053752102762e-05,
+      "loss": 0.9005,
+      "step": 11529
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.347470259423027e-05,
+      "loss": 0.8677,
+      "step": 11530
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.344403599007756e-05,
+      "loss": 0.8849,
+      "step": 11531
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.341337539916829e-05,
+      "loss": 0.8918,
+      "step": 11532
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.338272082285289e-05,
+      "loss": 0.91,
+      "step": 11533
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.335207226248099e-05,
+      "loss": 0.9057,
+      "step": 11534
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.332142971940227e-05,
+      "loss": 0.8381,
+      "step": 11535
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.329079319496616e-05,
+      "loss": 0.872,
+      "step": 11536
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.326016269052162e-05,
+      "loss": 0.8915,
+      "step": 11537
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.32295382074175e-05,
+      "loss": 0.9638,
+      "step": 11538
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.319891974700236e-05,
+      "loss": 0.8511,
+      "step": 11539
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.316830731062455e-05,
+      "loss": 0.7835,
+      "step": 11540
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.313770089963196e-05,
+      "loss": 0.9079,
+      "step": 11541
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.310710051537242e-05,
+      "loss": 0.8653,
+      "step": 11542
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.307650615919347e-05,
+      "loss": 0.8152,
+      "step": 11543
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.30459178324422e-05,
+      "loss": 0.8799,
+      "step": 11544
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.301533553646564e-05,
+      "loss": 0.8264,
+      "step": 11545
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.298475927261049e-05,
+      "loss": 0.9039,
+      "step": 11546
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.295418904222321e-05,
+      "loss": 0.8898,
+      "step": 11547
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.292362484664987e-05,
+      "loss": 0.9187,
+      "step": 11548
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.28930666872364e-05,
+      "loss": 0.8848,
+      "step": 11549
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.28625145653285e-05,
+      "loss": 0.9045,
+      "step": 11550
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.283196848227136e-05,
+      "loss": 0.875,
+      "step": 11551
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.28014284394103e-05,
+      "loss": 0.8635,
+      "step": 11552
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.277089443808998e-05,
+      "loss": 0.8599,
+      "step": 11553
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.274036647965504e-05,
+      "loss": 0.8434,
+      "step": 11554
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.270984456544984e-05,
+      "loss": 0.8686,
+      "step": 11555
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.267932869681825e-05,
+      "loss": 0.9332,
+      "step": 11556
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.264881887510415e-05,
+      "loss": 0.9384,
+      "step": 11557
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.261831510165099e-05,
+      "loss": 0.8717,
+      "step": 11558
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.258781737780211e-05,
+      "loss": 0.8621,
+      "step": 11559
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.255732570490031e-05,
+      "loss": 0.9907,
+      "step": 11560
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.252684008428839e-05,
+      "loss": 0.9026,
+      "step": 11561
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.249636051730886e-05,
+      "loss": 0.8909,
+      "step": 11562
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.246588700530364e-05,
+      "loss": 0.8689,
+      "step": 11563
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.243541954961493e-05,
+      "loss": 0.8477,
+      "step": 11564
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.240495815158418e-05,
+      "loss": 0.9417,
+      "step": 11565
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.237450281255283e-05,
+      "loss": 0.8844,
+      "step": 11566
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.234405353386195e-05,
+      "loss": 0.8674,
+      "step": 11567
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.231361031685234e-05,
+      "loss": 0.9006,
+      "step": 11568
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.228317316286468e-05,
+      "loss": 0.8786,
+      "step": 11569
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.225274207323908e-05,
+      "loss": 0.9581,
+      "step": 11570
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.22223170493158e-05,
+      "loss": 0.9038,
+      "step": 11571
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.219189809243445e-05,
+      "loss": 0.8825,
+      "step": 11572
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.216148520393457e-05,
+      "loss": 0.8437,
+      "step": 11573
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.213107838515548e-05,
+      "loss": 0.851,
+      "step": 11574
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.210067763743598e-05,
+      "loss": 0.8682,
+      "step": 11575
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.207028296211486e-05,
+      "loss": 0.8314,
+      "step": 11576
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.203989436053055e-05,
+      "loss": 0.9135,
+      "step": 11577
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.200951183402126e-05,
+      "loss": 0.9255,
+      "step": 11578
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.197913538392477e-05,
+      "loss": 0.8231,
+      "step": 11579
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.194876501157875e-05,
+      "loss": 0.8413,
+      "step": 11580
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.191840071832064e-05,
+      "loss": 0.8693,
+      "step": 11581
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.188804250548737e-05,
+      "loss": 0.8591,
+      "step": 11582
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.185769037441597e-05,
+      "loss": 0.8532,
+      "step": 11583
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.182734432644281e-05,
+      "loss": 0.8527,
+      "step": 11584
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.179700436290428e-05,
+      "loss": 0.9349,
+      "step": 11585
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.176667048513645e-05,
+      "loss": 0.9179,
+      "step": 11586
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.173634269447491e-05,
+      "loss": 0.8523,
+      "step": 11587
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.170602099225524e-05,
+      "loss": 0.8632,
+      "step": 11588
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.167570537981266e-05,
+      "loss": 0.8554,
+      "step": 11589
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.16453958584822e-05,
+      "loss": 0.9135,
+      "step": 11590
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.161509242959833e-05,
+      "loss": 0.9728,
+      "step": 11591
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.158479509449561e-05,
+      "loss": 0.8617,
+      "step": 11592
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.155450385450823e-05,
+      "loss": 0.925,
+      "step": 11593
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.152421871096991e-05,
+      "loss": 0.8927,
+      "step": 11594
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.149393966521435e-05,
+      "loss": 0.8781,
+      "step": 11595
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.146366671857488e-05,
+      "loss": 0.961,
+      "step": 11596
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.143339987238461e-05,
+      "loss": 0.8289,
+      "step": 11597
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.140313912797626e-05,
+      "loss": 0.8849,
+      "step": 11598
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.137288448668236e-05,
+      "loss": 0.8294,
+      "step": 11599
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.134263594983531e-05,
+      "loss": 0.8224,
+      "step": 11600
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.131239351876687e-05,
+      "loss": 0.8497,
+      "step": 11601
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.128215719480904e-05,
+      "loss": 0.8955,
+      "step": 11602
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.125192697929305e-05,
+      "loss": 0.903,
+      "step": 11603
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.122170287355018e-05,
+      "loss": 0.9574,
+      "step": 11604
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.119148487891143e-05,
+      "loss": 0.9759,
+      "step": 11605
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.116127299670729e-05,
+      "loss": 0.8826,
+      "step": 11606
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.113106722826822e-05,
+      "loss": 0.9413,
+      "step": 11607
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.110086757492433e-05,
+      "loss": 0.9514,
+      "step": 11608
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.10706740380055e-05,
+      "loss": 0.9103,
+      "step": 11609
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.104048661884124e-05,
+      "loss": 0.8974,
+      "step": 11610
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.101030531876084e-05,
+      "loss": 0.8901,
+      "step": 11611
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.0980130139093447e-05,
+      "loss": 0.8471,
+      "step": 11612
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.094996108116764e-05,
+      "loss": 0.8748,
+      "step": 11613
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.091979814631212e-05,
+      "loss": 0.855,
+      "step": 11614
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.088964133585497e-05,
+      "loss": 0.864,
+      "step": 11615
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.0859490651124176e-05,
+      "loss": 0.8718,
+      "step": 11616
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.0829346093447503e-05,
+      "loss": 0.9043,
+      "step": 11617
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.079920766415226e-05,
+      "loss": 0.8361,
+      "step": 11618
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.076907536456568e-05,
+      "loss": 0.8571,
+      "step": 11619
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.073894919601446e-05,
+      "loss": 0.8908,
+      "step": 11620
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.070882915982548e-05,
+      "loss": 0.9056,
+      "step": 11621
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.067871525732487e-05,
+      "loss": 0.8621,
+      "step": 11622
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.064860748983876e-05,
+      "loss": 0.8883,
+      "step": 11623
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.061850585869302e-05,
+      "loss": 0.8486,
+      "step": 11624
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.058841036521303e-05,
+      "loss": 0.8596,
+      "step": 11625
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.055832101072414e-05,
+      "loss": 0.8668,
+      "step": 11626
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.0528237796551304e-05,
+      "loss": 0.915,
+      "step": 11627
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.04981607240193e-05,
+      "loss": 0.8743,
+      "step": 11628
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.0468089794452463e-05,
+      "loss": 0.783,
+      "step": 11629
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.0438025009175015e-05,
+      "loss": 0.9162,
+      "step": 11630
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.040796636951096e-05,
+      "loss": 0.8967,
+      "step": 11631
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.0377913876783684e-05,
+      "loss": 0.8901,
+      "step": 11632
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.034786753231683e-05,
+      "loss": 0.9145,
+      "step": 11633
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.031782733743332e-05,
+      "loss": 0.8048,
+      "step": 11634
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.0287793293456e-05,
+      "loss": 0.857,
+      "step": 11635
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.025776540170749e-05,
+      "loss": 0.9065,
+      "step": 11636
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.022774366350996e-05,
+      "loss": 0.8509,
+      "step": 11637
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.019772808018549e-05,
+      "loss": 0.9066,
+      "step": 11638
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.0167718653055795e-05,
+      "loss": 0.8894,
+      "step": 11639
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.0137715383442396e-05,
+      "loss": 0.8674,
+      "step": 11640
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.010771827266637e-05,
+      "loss": 0.8917,
+      "step": 11641
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.0077727322048706e-05,
+      "loss": 0.8288,
+      "step": 11642
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.004774253291012e-05,
+      "loss": 0.8893,
+      "step": 11643
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.001776390657081e-05,
+      "loss": 0.9047,
+      "step": 11644
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.9987791444351116e-05,
+      "loss": 0.964,
+      "step": 11645
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.995782514757071e-05,
+      "loss": 0.853,
+      "step": 11646
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.992786501754928e-05,
+      "loss": 0.9026,
+      "step": 11647
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.989791105560598e-05,
+      "loss": 0.8544,
+      "step": 11648
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.98679632630599e-05,
+      "loss": 0.9424,
+      "step": 11649
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.983802164122987e-05,
+      "loss": 0.9165,
+      "step": 11650
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.9808086191434185e-05,
+      "loss": 0.8166,
+      "step": 11651
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.977815691499127e-05,
+      "loss": 0.811,
+      "step": 11652
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.97482338132189e-05,
+      "loss": 0.8624,
+      "step": 11653
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.971831688743479e-05,
+      "loss": 0.9739,
+      "step": 11654
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.9688406138956384e-05,
+      "loss": 0.8532,
+      "step": 11655
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.965850156910071e-05,
+      "loss": 0.9147,
+      "step": 11656
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.962860317918466e-05,
+      "loss": 0.8671,
+      "step": 11657
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.959871097052483e-05,
+      "loss": 0.9138,
+      "step": 11658
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.956882494443754e-05,
+      "loss": 0.8334,
+      "step": 11659
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.9538945102238716e-05,
+      "loss": 0.8163,
+      "step": 11660
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.950907144524423e-05,
+      "loss": 0.8464,
+      "step": 11661
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.947920397476954e-05,
+      "loss": 0.8508,
+      "step": 11662
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.944934269212976e-05,
+      "loss": 0.8871,
+      "step": 11663
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.941948759864002e-05,
+      "loss": 0.9342,
+      "step": 11664
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.9389638695614826e-05,
+      "loss": 0.8719,
+      "step": 11665
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.935979598436865e-05,
+      "loss": 0.9046,
+      "step": 11666
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.9329959466215666e-05,
+      "loss": 0.8543,
+      "step": 11667
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.9300129142469584e-05,
+      "loss": 0.8683,
+      "step": 11668
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.927030501444408e-05,
+      "loss": 0.8606,
+      "step": 11669
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.9240487083452424e-05,
+      "loss": 0.8701,
+      "step": 11670
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.921067535080773e-05,
+      "loss": 0.8363,
+      "step": 11671
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.9180869817822646e-05,
+      "loss": 0.91,
+      "step": 11672
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.9151070485809704e-05,
+      "loss": 0.8529,
+      "step": 11673
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.9121277356081175e-05,
+      "loss": 0.8675,
+      "step": 11674
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.9091490429948904e-05,
+      "loss": 0.8751,
+      "step": 11675
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.906170970872458e-05,
+      "loss": 0.9151,
+      "step": 11676
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.9031935193719615e-05,
+      "loss": 0.9374,
+      "step": 11677
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.90021668862452e-05,
+      "loss": 0.9067,
+      "step": 11678
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.897240478761206e-05,
+      "loss": 0.8678,
+      "step": 11679
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.894264889913081e-05,
+      "loss": 0.8429,
+      "step": 11680
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.891289922211185e-05,
+      "loss": 0.8675,
+      "step": 11681
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.8883155757864985e-05,
+      "loss": 0.8691,
+      "step": 11682
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.885341850770025e-05,
+      "loss": 0.8484,
+      "step": 11683
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.88236874729269e-05,
+      "loss": 0.9034,
+      "step": 11684
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.879396265485424e-05,
+      "loss": 0.9057,
+      "step": 11685
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.876424405479124e-05,
+      "loss": 0.8324,
+      "step": 11686
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.8734531674046434e-05,
+      "loss": 0.8823,
+      "step": 11687
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.8704825513928285e-05,
+      "loss": 0.9042,
+      "step": 11688
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.867512557574491e-05,
+      "loss": 0.89,
+      "step": 11689
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.8645431860804176e-05,
+      "loss": 0.9231,
+      "step": 11690
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.861574437041355e-05,
+      "loss": 0.8802,
+      "step": 11691
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.8586063105880396e-05,
+      "loss": 0.8841,
+      "step": 11692
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.855638806851174e-05,
+      "loss": 0.8647,
+      "step": 11693
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.852671925961419e-05,
+      "loss": 0.8735,
+      "step": 11694
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.849705668049443e-05,
+      "loss": 0.9098,
+      "step": 11695
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.8467400332458455e-05,
+      "loss": 0.8467,
+      "step": 11696
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.843775021681228e-05,
+      "loss": 0.8978,
+      "step": 11697
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.84081063348616e-05,
+      "loss": 0.8626,
+      "step": 11698
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.837846868791164e-05,
+      "loss": 0.8692,
+      "step": 11699
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.834883727726763e-05,
+      "loss": 0.9074,
+      "step": 11700
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.831921210423421e-05,
+      "loss": 0.8355,
+      "step": 11701
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.828959317011617e-05,
+      "loss": 0.9302,
+      "step": 11702
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.825998047621759e-05,
+      "loss": 0.8715,
+      "step": 11703
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.823037402384253e-05,
+      "loss": 0.9225,
+      "step": 11704
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.8200773814294764e-05,
+      "loss": 0.8978,
+      "step": 11705
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.817117984887761e-05,
+      "loss": 0.8672,
+      "step": 11706
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.814159212889434e-05,
+      "loss": 0.8304,
+      "step": 11707
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.81120106556478e-05,
+      "loss": 0.883,
+      "step": 11708
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.80824354304407e-05,
+      "loss": 0.8546,
+      "step": 11709
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.805286645457526e-05,
+      "loss": 0.9214,
+      "step": 11710
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.802330372935358e-05,
+      "loss": 0.9145,
+      "step": 11711
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.799374725607758e-05,
+      "loss": 0.8542,
+      "step": 11712
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.796419703604854e-05,
+      "loss": 0.8423,
+      "step": 11713
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.793465307056798e-05,
+      "loss": 0.8898,
+      "step": 11714
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.790511536093666e-05,
+      "loss": 0.8611,
+      "step": 11715
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.787558390845536e-05,
+      "loss": 0.9176,
+      "step": 11716
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.7846058714424546e-05,
+      "loss": 0.9425,
+      "step": 11717
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.781653978014423e-05,
+      "loss": 0.877,
+      "step": 11718
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.778702710691437e-05,
+      "loss": 0.8829,
+      "step": 11719
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.77575206960345e-05,
+      "loss": 0.8662,
+      "step": 11720
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.772802054880406e-05,
+      "loss": 0.8855,
+      "step": 11721
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.769852666652193e-05,
+      "loss": 0.9132,
+      "step": 11722
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.766903905048697e-05,
+      "loss": 0.897,
+      "step": 11723
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.763955770199767e-05,
+      "loss": 0.8714,
+      "step": 11724
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.76100826223521e-05,
+      "loss": 0.7967,
+      "step": 11725
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.758061381284845e-05,
+      "loss": 0.853,
+      "step": 11726
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.755115127478414e-05,
+      "loss": 0.8363,
+      "step": 11727
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.7521695009456745e-05,
+      "loss": 0.9011,
+      "step": 11728
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.7492245018163213e-05,
+      "loss": 0.8974,
+      "step": 11729
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.746280130220043e-05,
+      "loss": 0.8877,
+      "step": 11730
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.743336386286502e-05,
+      "loss": 0.8766,
+      "step": 11731
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.740393270145308e-05,
+      "loss": 0.8844,
+      "step": 11732
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.737450781926086e-05,
+      "loss": 0.8552,
+      "step": 11733
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.73450892175839e-05,
+      "loss": 0.8817,
+      "step": 11734
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.731567689771771e-05,
+      "loss": 0.9274,
+      "step": 11735
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.7286270860957503e-05,
+      "loss": 0.8872,
+      "step": 11736
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.725687110859808e-05,
+      "loss": 0.838,
+      "step": 11737
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.72274776419341e-05,
+      "loss": 0.9103,
+      "step": 11738
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.7198090462259926e-05,
+      "loss": 0.8937,
+      "step": 11739
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.7168709570869664e-05,
+      "loss": 0.8625,
+      "step": 11740
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.713933496905699e-05,
+      "loss": 0.8236,
+      "step": 11741
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.7109966658115476e-05,
+      "loss": 0.9105,
+      "step": 11742
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.708060463933842e-05,
+      "loss": 0.8719,
+      "step": 11743
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.705124891401856e-05,
+      "loss": 0.9316,
+      "step": 11744
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.7021899483448894e-05,
+      "loss": 0.8663,
+      "step": 11745
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.6992556348921555e-05,
+      "loss": 0.9034,
+      "step": 11746
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.6963219511728805e-05,
+      "loss": 0.826,
+      "step": 11747
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.69338889731625e-05,
+      "loss": 0.8605,
+      "step": 11748
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.690456473451413e-05,
+      "loss": 0.8828,
+      "step": 11749
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.687524679707501e-05,
+      "loss": 0.8427,
+      "step": 11750
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.6845935162136167e-05,
+      "loss": 0.9656,
+      "step": 11751
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.6816629830988436e-05,
+      "loss": 0.8676,
+      "step": 11752
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.678733080492209e-05,
+      "loss": 0.8683,
+      "step": 11753
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.675803808522744e-05,
+      "loss": 0.8008,
+      "step": 11754
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.67287516731944e-05,
+      "loss": 0.8889,
+      "step": 11755
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.669947157011251e-05,
+      "loss": 0.8583,
+      "step": 11756
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.667019777727116e-05,
+      "loss": 0.8547,
+      "step": 11757
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.664093029595943e-05,
+      "loss": 0.8042,
+      "step": 11758
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.661166912746618e-05,
+      "loss": 0.8724,
+      "step": 11759
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.65824142730798e-05,
+      "loss": 0.9129,
+      "step": 11760
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.6553165734088575e-05,
+      "loss": 0.7914,
+      "step": 11761
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.652392351178055e-05,
+      "loss": 0.8176,
+      "step": 11762
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.649468760744321e-05,
+      "loss": 0.8575,
+      "step": 11763
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.646545802236418e-05,
+      "loss": 0.8958,
+      "step": 11764
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.6436234757830464e-05,
+      "loss": 0.9206,
+      "step": 11765
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.640701781512891e-05,
+      "loss": 0.9422,
+      "step": 11766
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.6377807195546175e-05,
+      "loss": 0.8346,
+      "step": 11767
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.634860290036843e-05,
+      "loss": 0.9029,
+      "step": 11768
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.631940493088173e-05,
+      "loss": 0.8982,
+      "step": 11769
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.6290213288371805e-05,
+      "loss": 0.8154,
+      "step": 11770
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.6261027974124206e-05,
+      "loss": 0.8865,
+      "step": 11771
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.623184898942395e-05,
+      "loss": 0.7898,
+      "step": 11772
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.6202676335556e-05,
+      "loss": 0.879,
+      "step": 11773
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.6173510013805066e-05,
+      "loss": 0.8321,
+      "step": 11774
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.614435002545526e-05,
+      "loss": 0.8658,
+      "step": 11775
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.6115196371790926e-05,
+      "loss": 0.8524,
+      "step": 11776
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.608604905409562e-05,
+      "loss": 0.8926,
+      "step": 11777
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.605690807365296e-05,
+      "loss": 0.9142,
+      "step": 11778
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.602777343174617e-05,
+      "loss": 0.8549,
+      "step": 11779
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.599864512965811e-05,
+      "loss": 0.9541,
+      "step": 11780
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.5969523168671564e-05,
+      "loss": 0.8793,
+      "step": 11781
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.594040755006875e-05,
+      "loss": 0.8562,
+      "step": 11782
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.591129827513197e-05,
+      "loss": 0.8891,
+      "step": 11783
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.588219534514292e-05,
+      "loss": 0.8655,
+      "step": 11784
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.585309876138318e-05,
+      "loss": 0.9082,
+      "step": 11785
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.5824008525134076e-05,
+      "loss": 0.8597,
+      "step": 11786
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.5794924637676505e-05,
+      "loss": 0.8893,
+      "step": 11787
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.5765847100291204e-05,
+      "loss": 0.8634,
+      "step": 11788
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.573677591425861e-05,
+      "loss": 0.8307,
+      "step": 11789
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.570771108085895e-05,
+      "loss": 0.8878,
+      "step": 11790
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.567865260137195e-05,
+      "loss": 0.8483,
+      "step": 11791
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.564960047707728e-05,
+      "loss": 0.8212,
+      "step": 11792
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.562055470925431e-05,
+      "loss": 0.8039,
+      "step": 11793
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.5591515299181875e-05,
+      "loss": 0.8706,
+      "step": 11794
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.556248224813898e-05,
+      "loss": 0.892,
+      "step": 11795
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.5533455557403925e-05,
+      "loss": 0.8099,
+      "step": 11796
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.550443522825493e-05,
+      "loss": 0.9519,
+      "step": 11797
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.547542126196998e-05,
+      "loss": 0.8784,
+      "step": 11798
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.544641365982661e-05,
+      "loss": 0.8935,
+      "step": 11799
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.541741242310219e-05,
+      "loss": 0.8653,
+      "step": 11800
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.538841755307382e-05,
+      "loss": 0.8483,
+      "step": 11801
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.5359429051018316e-05,
+      "loss": 0.8753,
+      "step": 11802
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.533044691821212e-05,
+      "loss": 0.8159,
+      "step": 11803
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.5301471155931474e-05,
+      "loss": 0.8458,
+      "step": 11804
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.52725017654524e-05,
+      "loss": 0.8639,
+      "step": 11805
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.524353874805037e-05,
+      "loss": 0.8081,
+      "step": 11806
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.521458210500105e-05,
+      "loss": 0.8622,
+      "step": 11807
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.518563183757934e-05,
+      "loss": 0.9064,
+      "step": 11808
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.515668794706019e-05,
+      "loss": 0.8311,
+      "step": 11809
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.512775043471803e-05,
+      "loss": 0.8909,
+      "step": 11810
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.509881930182717e-05,
+      "loss": 0.9084,
+      "step": 11811
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.506989454966165e-05,
+      "loss": 0.9443,
+      "step": 11812
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.504097617949502e-05,
+      "loss": 0.8459,
+      "step": 11813
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.501206419260092e-05,
+      "loss": 0.8802,
+      "step": 11814
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.498315859025229e-05,
+      "loss": 0.8067,
+      "step": 11815
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.4954259373722095e-05,
+      "loss": 0.8331,
+      "step": 11816
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.492536654428295e-05,
+      "loss": 0.8791,
+      "step": 11817
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.4896480103207005e-05,
+      "loss": 0.8459,
+      "step": 11818
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.486760005176639e-05,
+      "loss": 0.9012,
+      "step": 11819
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.48387263912328e-05,
+      "loss": 0.9046,
+      "step": 11820
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.480985912287773e-05,
+      "loss": 0.8263,
+      "step": 11821
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.478099824797227e-05,
+      "loss": 0.8405,
+      "step": 11822
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.475214376778737e-05,
+      "loss": 0.8795,
+      "step": 11823
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.472329568359367e-05,
+      "loss": 0.8926,
+      "step": 11824
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.469445399666133e-05,
+      "loss": 0.8249,
+      "step": 11825
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.4665618708260615e-05,
+      "loss": 0.8881,
+      "step": 11826
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.463678981966114e-05,
+      "loss": 0.8964,
+      "step": 11827
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.4607967332132426e-05,
+      "loss": 0.9059,
+      "step": 11828
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.457915124694375e-05,
+      "loss": 0.8632,
+      "step": 11829
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.4550341565363874e-05,
+      "loss": 0.9083,
+      "step": 11830
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.452153828866151e-05,
+      "loss": 0.9023,
+      "step": 11831
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.449274141810501e-05,
+      "loss": 0.8837,
+      "step": 11832
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.446395095496253e-05,
+      "loss": 0.8505,
+      "step": 11833
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.44351669005017e-05,
+      "loss": 0.8774,
+      "step": 11834
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.440638925599011e-05,
+      "loss": 0.9198,
+      "step": 11835
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.437761802269503e-05,
+      "loss": 0.9112,
+      "step": 11836
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.434885320188328e-05,
+      "loss": 0.879,
+      "step": 11837
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.4320094794821605e-05,
+      "loss": 0.8314,
+      "step": 11838
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.4291342802776344e-05,
+      "loss": 0.8497,
+      "step": 11839
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.4262597227013676e-05,
+      "loss": 0.8304,
+      "step": 11840
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.4233858068799304e-05,
+      "loss": 0.898,
+      "step": 11841
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.420512532939879e-05,
+      "loss": 0.8631,
+      "step": 11842
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.417639901007745e-05,
+      "loss": 0.8799,
+      "step": 11843
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.414767911210006e-05,
+      "loss": 0.8507,
+      "step": 11844
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.411896563673158e-05,
+      "loss": 0.8297,
+      "step": 11845
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.4090258585236176e-05,
+      "loss": 0.9078,
+      "step": 11846
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.406155795887804e-05,
+      "loss": 0.8171,
+      "step": 11847
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.403286375892107e-05,
+      "loss": 0.9032,
+      "step": 11848
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.4004175986628705e-05,
+      "loss": 0.8635,
+      "step": 11849
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.397549464326428e-05,
+      "loss": 0.8085,
+      "step": 11850
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.394681973009073e-05,
+      "loss": 0.9362,
+      "step": 11851
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.391815124837085e-05,
+      "loss": 0.9033,
+      "step": 11852
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.388948919936696e-05,
+      "loss": 0.8466,
+      "step": 11853
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.38608335843412e-05,
+      "loss": 0.8903,
+      "step": 11854
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.383218440455551e-05,
+      "loss": 0.9193,
+      "step": 11855
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.380354166127128e-05,
+      "loss": 0.8777,
+      "step": 11856
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.377490535575003e-05,
+      "loss": 0.7976,
+      "step": 11857
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.3746275489252575e-05,
+      "loss": 0.9207,
+      "step": 11858
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.371765206303969e-05,
+      "loss": 0.857,
+      "step": 11859
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.368903507837186e-05,
+      "loss": 0.8887,
+      "step": 11860
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.366042453650915e-05,
+      "loss": 0.8503,
+      "step": 11861
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.363182043871151e-05,
+      "loss": 0.9008,
+      "step": 11862
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.360322278623835e-05,
+      "loss": 0.8852,
+      "step": 11863
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.357463158034921e-05,
+      "loss": 0.8423,
+      "step": 11864
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.354604682230293e-05,
+      "loss": 0.9273,
+      "step": 11865
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.351746851335831e-05,
+      "loss": 0.8316,
+      "step": 11866
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.348889665477381e-05,
+      "loss": 0.9248,
+      "step": 11867
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.346033124780754e-05,
+      "loss": 0.879,
+      "step": 11868
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.3431772293717364e-05,
+      "loss": 0.8643,
+      "step": 11869
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.340321979376095e-05,
+      "loss": 0.9083,
+      "step": 11870
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.33746737491956e-05,
+      "loss": 0.9322,
+      "step": 11871
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.334613416127827e-05,
+      "loss": 0.8464,
+      "step": 11872
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.331760103126575e-05,
+      "loss": 0.9286,
+      "step": 11873
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.3289074360414525e-05,
+      "loss": 0.9144,
+      "step": 11874
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.326055414998064e-05,
+      "loss": 0.8646,
+      "step": 11875
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.3232040401220165e-05,
+      "loss": 0.9293,
+      "step": 11876
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.320353311538857e-05,
+      "loss": 0.8463,
+      "step": 11877
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.317503229374121e-05,
+      "loss": 0.8871,
+      "step": 11878
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.314653793753319e-05,
+      "loss": 0.9256,
+      "step": 11879
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.311805004801913e-05,
+      "loss": 0.8242,
+      "step": 11880
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.308956862645355e-05,
+      "loss": 0.847,
+      "step": 11881
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.3061093674090646e-05,
+      "loss": 0.8691,
+      "step": 11882
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.303262519218437e-05,
+      "loss": 0.855,
+      "step": 11883
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.300416318198822e-05,
+      "loss": 0.8406,
+      "step": 11884
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.2975707644755544e-05,
+      "loss": 0.8287,
+      "step": 11885
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.2947258581739476e-05,
+      "loss": 0.836,
+      "step": 11886
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.2918815994192636e-05,
+      "loss": 0.9244,
+      "step": 11887
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.289037988336758e-05,
+      "loss": 0.8728,
+      "step": 11888
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.286195025051646e-05,
+      "loss": 0.872,
+      "step": 11889
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.283352709689122e-05,
+      "loss": 0.8587,
+      "step": 11890
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.280511042374341e-05,
+      "loss": 0.8928,
+      "step": 11891
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.277670023232439e-05,
+      "loss": 0.868,
+      "step": 11892
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.274829652388524e-05,
+      "loss": 0.8846,
+      "step": 11893
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.271989929967658e-05,
+      "loss": 0.86,
+      "step": 11894
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.269150856094911e-05,
+      "loss": 0.909,
+      "step": 11895
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.2663124308952814e-05,
+      "loss": 0.8106,
+      "step": 11896
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.263474654493765e-05,
+      "loss": 0.9201,
+      "step": 11897
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.260637527015333e-05,
+      "loss": 0.8116,
+      "step": 11898
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.257801048584905e-05,
+      "loss": 0.9475,
+      "step": 11899
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.25496521932739e-05,
+      "loss": 0.8981,
+      "step": 11900
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.252130039367664e-05,
+      "loss": 0.8963,
+      "step": 11901
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.249295508830583e-05,
+      "loss": 0.8614,
+      "step": 11902
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.246461627840949e-05,
+      "loss": 0.832,
+      "step": 11903
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.2436283965235614e-05,
+      "loss": 0.8738,
+      "step": 11904
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.240795815003186e-05,
+      "loss": 0.9623,
+      "step": 11905
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.237963883404538e-05,
+      "loss": 0.8628,
+      "step": 11906
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.235132601852346e-05,
+      "loss": 0.8601,
+      "step": 11907
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.2323019704712675e-05,
+      "loss": 0.8759,
+      "step": 11908
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.229471989385954e-05,
+      "loss": 0.8873,
+      "step": 11909
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.22664265872103e-05,
+      "loss": 0.7917,
+      "step": 11910
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.223813978601075e-05,
+      "loss": 0.8388,
+      "step": 11911
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.2209859491506566e-05,
+      "loss": 0.8807,
+      "step": 11912
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.218158570494303e-05,
+      "loss": 0.8406,
+      "step": 11913
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.2153318427565254e-05,
+      "loss": 0.869,
+      "step": 11914
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.2125057660617905e-05,
+      "loss": 0.8113,
+      "step": 11915
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.209680340534546e-05,
+      "loss": 0.8931,
+      "step": 11916
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.206855566299216e-05,
+      "loss": 0.8648,
+      "step": 11917
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.2040314434801794e-05,
+      "loss": 0.9157,
+      "step": 11918
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.2012079722018046e-05,
+      "loss": 0.8485,
+      "step": 11919
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.198385152588416e-05,
+      "loss": 0.8879,
+      "step": 11920
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.1955629847643306e-05,
+      "loss": 0.849,
+      "step": 11921
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.192741468853806e-05,
+      "loss": 0.8788,
+      "step": 11922
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.1899206049810934e-05,
+      "loss": 0.8872,
+      "step": 11923
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.1871003932704164e-05,
+      "loss": 0.9094,
+      "step": 11924
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.184280833845947e-05,
+      "loss": 0.8846,
+      "step": 11925
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.1814619268318674e-05,
+      "loss": 0.8708,
+      "step": 11926
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.178643672352288e-05,
+      "loss": 0.8541,
+      "step": 11927
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.17582607053132e-05,
+      "loss": 0.8315,
+      "step": 11928
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.17300912149304e-05,
+      "loss": 0.8612,
+      "step": 11929
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.1701928253614794e-05,
+      "loss": 0.8284,
+      "step": 11930
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.1673771822606645e-05,
+      "loss": 0.8742,
+      "step": 11931
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.164562192314579e-05,
+      "loss": 0.9242,
+      "step": 11932
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.161747855647188e-05,
+      "loss": 0.8797,
+      "step": 11933
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.158934172382406e-05,
+      "loss": 0.9059,
+      "step": 11934
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.1561211426441435e-05,
+      "loss": 0.8312,
+      "step": 11935
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.153308766556277e-05,
+      "loss": 0.8726,
+      "step": 11936
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.15049704424263e-05,
+      "loss": 0.8636,
+      "step": 11937
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.1476859758270436e-05,
+      "loss": 0.8082,
+      "step": 11938
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.144875561433282e-05,
+      "loss": 0.8043,
+      "step": 11939
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.142065801185116e-05,
+      "loss": 0.8747,
+      "step": 11940
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.139256695206263e-05,
+      "loss": 0.8285,
+      "step": 11941
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.136448243620422e-05,
+      "loss": 0.9041,
+      "step": 11942
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.1336404465512756e-05,
+      "loss": 0.8303,
+      "step": 11943
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.1308333041224445e-05,
+      "loss": 0.8174,
+      "step": 11944
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.128026816457563e-05,
+      "loss": 0.995,
+      "step": 11945
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.125220983680201e-05,
+      "loss": 0.8364,
+      "step": 11946
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.1224158059139185e-05,
+      "loss": 0.8833,
+      "step": 11947
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.119611283282246e-05,
+      "loss": 0.9125,
+      "step": 11948
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.116807415908669e-05,
+      "loss": 0.8706,
+      "step": 11949
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.1140042039166603e-05,
+      "loss": 0.8909,
+      "step": 11950
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.111201647429664e-05,
+      "loss": 0.819,
+      "step": 11951
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.108399746571093e-05,
+      "loss": 0.8504,
+      "step": 11952
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.105598501464317e-05,
+      "loss": 0.8869,
+      "step": 11953
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.1027979122326976e-05,
+      "loss": 0.8781,
+      "step": 11954
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.099997978999562e-05,
+      "loss": 0.8842,
+      "step": 11955
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.0971987018881886e-05,
+      "loss": 0.8433,
+      "step": 11956
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.0944000810218665e-05,
+      "loss": 0.933,
+      "step": 11957
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.0916021165238173e-05,
+      "loss": 0.8074,
+      "step": 11958
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.088804808517253e-05,
+      "loss": 0.8757,
+      "step": 11959
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.0860081571253595e-05,
+      "loss": 0.8663,
+      "step": 11960
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.083212162471276e-05,
+      "loss": 0.8731,
+      "step": 11961
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.080416824678131e-05,
+      "loss": 0.8879,
+      "step": 11962
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.077622143869016e-05,
+      "loss": 0.876,
+      "step": 11963
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.074828120167001e-05,
+      "loss": 0.9722,
+      "step": 11964
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.0720347536951077e-05,
+      "loss": 0.8266,
+      "step": 11965
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.06924204457635e-05,
+      "loss": 0.8482,
+      "step": 11966
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.0664499929337106e-05,
+      "loss": 0.9226,
+      "step": 11967
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.0636585988901264e-05,
+      "loss": 0.8355,
+      "step": 11968
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.060867862568519e-05,
+      "loss": 0.8408,
+      "step": 11969
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.0580777840917816e-05,
+      "loss": 0.8599,
+      "step": 11970
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.05528836358278e-05,
+      "loss": 0.9072,
+      "step": 11971
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.052499601164335e-05,
+      "loss": 0.848,
+      "step": 11972
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.049711496959255e-05,
+      "loss": 0.8556,
+      "step": 11973
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.046924051090321e-05,
+      "loss": 0.8577,
+      "step": 11974
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.04413726368026e-05,
+      "loss": 0.9196,
+      "step": 11975
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.041351134851813e-05,
+      "loss": 0.8641,
+      "step": 11976
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.038565664727648e-05,
+      "loss": 0.8542,
+      "step": 11977
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.035780853430429e-05,
+      "loss": 0.8468,
+      "step": 11978
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.0329967010827924e-05,
+      "loss": 0.902,
+      "step": 11979
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.030213207807326e-05,
+      "loss": 0.8459,
+      "step": 11980
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.027430373726607e-05,
+      "loss": 0.8901,
+      "step": 11981
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.024648198963175e-05,
+      "loss": 0.8596,
+      "step": 11982
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.0218666836395556e-05,
+      "loss": 0.9302,
+      "step": 11983
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.019085827878214e-05,
+      "loss": 0.8422,
+      "step": 11984
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.016305631801614e-05,
+      "loss": 0.8846,
+      "step": 11985
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.013526095532186e-05,
+      "loss": 0.8691,
+      "step": 11986
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.0107472191923135e-05,
+      "loss": 0.8735,
+      "step": 11987
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.0079690029043825e-05,
+      "loss": 0.9005,
+      "step": 11988
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.005191446790718e-05,
+      "loss": 0.8416,
+      "step": 11989
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.0024145509736334e-05,
+      "loss": 0.8228,
+      "step": 11990
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.999638315575417e-05,
+      "loss": 0.9322,
+      "step": 11991
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.9968627407183065e-05,
+      "loss": 0.8622,
+      "step": 11992
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.994087826524534e-05,
+      "loss": 0.8138,
+      "step": 11993
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.9913135731162876e-05,
+      "loss": 0.8773,
+      "step": 11994
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.9885399806157406e-05,
+      "loss": 0.9455,
+      "step": 11995
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.9857670491450156e-05,
+      "loss": 0.9531,
+      "step": 11996
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.982994778826227e-05,
+      "loss": 0.8983,
+      "step": 11997
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.980223169781453e-05,
+      "loss": 0.8772,
+      "step": 11998
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.977452222132735e-05,
+      "loss": 0.8788,
+      "step": 11999
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.974681936002093e-05,
+      "loss": 0.8776,
+      "step": 12000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.971912311511517e-05,
+      "loss": 0.9144,
+      "step": 12001
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.969143348782976e-05,
+      "loss": 0.834,
+      "step": 12002
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.9663750479383894e-05,
+      "loss": 0.8523,
+      "step": 12003
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.963607409099664e-05,
+      "loss": 0.8551,
+      "step": 12004
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.960840432388678e-05,
+      "loss": 0.8933,
+      "step": 12005
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.958074117927258e-05,
+      "loss": 0.8273,
+      "step": 12006
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.955308465837245e-05,
+      "loss": 0.8227,
+      "step": 12007
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.9525434762404013e-05,
+      "loss": 0.882,
+      "step": 12008
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.949779149258496e-05,
+      "loss": 0.9338,
+      "step": 12009
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.947015485013255e-05,
+      "loss": 0.8368,
+      "step": 12010
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.94425248362637e-05,
+      "loss": 0.8799,
+      "step": 12011
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.9414901452195115e-05,
+      "loss": 0.8545,
+      "step": 12012
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.938728469914324e-05,
+      "loss": 0.8999,
+      "step": 12013
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.93596745783242e-05,
+      "loss": 0.844,
+      "step": 12014
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.93320710909537e-05,
+      "loss": 0.8206,
+      "step": 12015
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.930447423824733e-05,
+      "loss": 0.9177,
+      "step": 12016
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.9276884021420344e-05,
+      "loss": 0.7538,
+      "step": 12017
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.924930044168754e-05,
+      "loss": 0.8879,
+      "step": 12018
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.922172350026377e-05,
+      "loss": 0.8334,
+      "step": 12019
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.9194153198363226e-05,
+      "loss": 0.9146,
+      "step": 12020
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.916658953720008e-05,
+      "loss": 0.9281,
+      "step": 12021
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.913903251798797e-05,
+      "loss": 0.8918,
+      "step": 12022
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.911148214194044e-05,
+      "loss": 0.8908,
+      "step": 12023
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.9083938410270725e-05,
+      "loss": 0.9027,
+      "step": 12024
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.9056401324191516e-05,
+      "loss": 0.897,
+      "step": 12025
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.90288708849157e-05,
+      "loss": 0.8789,
+      "step": 12026
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.900134709365534e-05,
+      "loss": 0.9652,
+      "step": 12027
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.8973829951622544e-05,
+      "loss": 0.8804,
+      "step": 12028
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.894631946002908e-05,
+      "loss": 0.9113,
+      "step": 12029
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.891881562008627e-05,
+      "loss": 0.8687,
+      "step": 12030
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.8891318433005275e-05,
+      "loss": 0.8676,
+      "step": 12031
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.886382789999695e-05,
+      "loss": 0.8713,
+      "step": 12032
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.883634402227188e-05,
+      "loss": 0.8593,
+      "step": 12033
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.8808866801040244e-05,
+      "loss": 0.7966,
+      "step": 12034
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.878139623751201e-05,
+      "loss": 0.9081,
+      "step": 12035
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.8753932332896954e-05,
+      "loss": 0.8917,
+      "step": 12036
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.8726475088404246e-05,
+      "loss": 0.8836,
+      "step": 12037
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.8699024505243174e-05,
+      "loss": 0.8593,
+      "step": 12038
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.86715805846224e-05,
+      "loss": 0.875,
+      "step": 12039
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.864414332775045e-05,
+      "loss": 0.9181,
+      "step": 12040
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.861671273583559e-05,
+      "loss": 0.8596,
+      "step": 12041
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.85892888100856e-05,
+      "loss": 0.8863,
+      "step": 12042
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.856187155170815e-05,
+      "loss": 0.8942,
+      "step": 12043
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.853446096191057e-05,
+      "loss": 0.8182,
+      "step": 12044
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.8507057041899926e-05,
+      "loss": 0.8118,
+      "step": 12045
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.847965979288287e-05,
+      "loss": 0.8626,
+      "step": 12046
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.845226921606587e-05,
+      "loss": 0.858,
+      "step": 12047
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.8424885312655124e-05,
+      "loss": 0.8773,
+      "step": 12048
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.839750808385639e-05,
+      "loss": 0.8662,
+      "step": 12049
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.837013753087525e-05,
+      "loss": 0.9016,
+      "step": 12050
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.8342773654917e-05,
+      "loss": 0.8679,
+      "step": 12051
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.831541645718665e-05,
+      "loss": 0.8475,
+      "step": 12052
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.828806593888875e-05,
+      "loss": 0.8761,
+      "step": 12053
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.826072210122776e-05,
+      "loss": 0.8539,
+      "step": 12054
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.823338494540781e-05,
+      "loss": 0.8529,
+      "step": 12055
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.820605447263253e-05,
+      "loss": 0.9171,
+      "step": 12056
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.8178730684105633e-05,
+      "loss": 0.8944,
+      "step": 12057
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.815141358103015e-05,
+      "loss": 0.8799,
+      "step": 12058
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.8124103164609066e-05,
+      "loss": 0.8106,
+      "step": 12059
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.809679943604504e-05,
+      "loss": 0.8382,
+      "step": 12060
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.8069502396540267e-05,
+      "loss": 0.8658,
+      "step": 12061
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.804221204729686e-05,
+      "loss": 0.9011,
+      "step": 12062
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.8014928389516513e-05,
+      "loss": 0.8831,
+      "step": 12063
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.7987651424400756e-05,
+      "loss": 0.8918,
+      "step": 12064
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.796038115315058e-05,
+      "loss": 0.9161,
+      "step": 12065
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.7933117576966924e-05,
+      "loss": 0.9043,
+      "step": 12066
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.7905860697050364e-05,
+      "loss": 0.8254,
+      "step": 12067
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.787861051460101e-05,
+      "loss": 0.8489,
+      "step": 12068
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.7851367030819026e-05,
+      "loss": 0.922,
+      "step": 12069
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.7824130246903955e-05,
+      "loss": 0.8887,
+      "step": 12070
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.779690016405518e-05,
+      "loss": 0.865,
+      "step": 12071
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.7769676783471865e-05,
+      "loss": 0.8911,
+      "step": 12072
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.774246010635264e-05,
+      "loss": 0.7778,
+      "step": 12073
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.771525013389615e-05,
+      "loss": 0.8395,
+      "step": 12074
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.768804686730039e-05,
+      "loss": 0.9473,
+      "step": 12075
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.7660850307763505e-05,
+      "loss": 0.8749,
+      "step": 12076
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.7633660456482896e-05,
+      "loss": 0.9261,
+      "step": 12077
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.760647731465593e-05,
+      "loss": 0.8636,
+      "step": 12078
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.7579300883479705e-05,
+      "loss": 0.8739,
+      "step": 12079
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.7552131164150784e-05,
+      "loss": 0.86,
+      "step": 12080
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.7524968157865666e-05,
+      "loss": 0.9435,
+      "step": 12081
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.749781186582047e-05,
+      "loss": 0.8524,
+      "step": 12082
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.747066228921109e-05,
+      "loss": 0.9961,
+      "step": 12083
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.744351942923291e-05,
+      "loss": 0.9654,
+      "step": 12084
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.741638328708125e-05,
+      "loss": 0.8594,
+      "step": 12085
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.73892538639511e-05,
+      "loss": 0.8988,
+      "step": 12086
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.736213116103694e-05,
+      "loss": 0.8608,
+      "step": 12087
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.733501517953334e-05,
+      "loss": 0.878,
+      "step": 12088
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.730790592063419e-05,
+      "loss": 0.8788,
+      "step": 12089
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.728080338553329e-05,
+      "loss": 0.836,
+      "step": 12090
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.725370757542418e-05,
+      "loss": 0.9084,
+      "step": 12091
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.7226618491499875e-05,
+      "loss": 0.8427,
+      "step": 12092
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.7199536134953334e-05,
+      "loss": 0.8674,
+      "step": 12093
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.7172460506977124e-05,
+      "loss": 0.881,
+      "step": 12094
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.714539160876355e-05,
+      "loss": 0.8098,
+      "step": 12095
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.711832944150452e-05,
+      "loss": 0.8992,
+      "step": 12096
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.709127400639173e-05,
+      "loss": 0.9322,
+      "step": 12097
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.706422530461665e-05,
+      "loss": 0.9291,
+      "step": 12098
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.703718333737017e-05,
+      "loss": 0.9006,
+      "step": 12099
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.7010148105843364e-05,
+      "loss": 0.8913,
+      "step": 12100
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.698311961122652e-05,
+      "loss": 0.7894,
+      "step": 12101
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.695609785470996e-05,
+      "loss": 0.8747,
+      "step": 12102
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.692908283748347e-05,
+      "loss": 0.8405,
+      "step": 12103
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.690207456073672e-05,
+      "loss": 0.9375,
+      "step": 12104
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.6875073025659075e-05,
+      "loss": 0.8496,
+      "step": 12105
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.6848078233439376e-05,
+      "loss": 0.9124,
+      "step": 12106
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.682109018526655e-05,
+      "loss": 0.9012,
+      "step": 12107
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.679410888232889e-05,
+      "loss": 0.8696,
+      "step": 12108
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.6767134325814524e-05,
+      "loss": 0.8679,
+      "step": 12109
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.6740166516911353e-05,
+      "loss": 0.9015,
+      "step": 12110
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.671320545680682e-05,
+      "loss": 0.8479,
+      "step": 12111
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.668625114668816e-05,
+      "loss": 0.8314,
+      "step": 12112
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.6659303587742356e-05,
+      "loss": 0.8375,
+      "step": 12113
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.663236278115606e-05,
+      "loss": 0.9103,
+      "step": 12114
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.660542872811553e-05,
+      "loss": 0.8954,
+      "step": 12115
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.657850142980684e-05,
+      "loss": 0.8869,
+      "step": 12116
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.65515808874158e-05,
+      "loss": 0.9106,
+      "step": 12117
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.6524667102127706e-05,
+      "loss": 0.8752,
+      "step": 12118
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.64977600751279e-05,
+      "loss": 0.8043,
+      "step": 12119
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.647085980760107e-05,
+      "loss": 0.8783,
+      "step": 12120
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.644396630073184e-05,
+      "loss": 0.8882,
+      "step": 12121
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.641707955570451e-05,
+      "loss": 0.8754,
+      "step": 12122
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.639019957370294e-05,
+      "loss": 0.8775,
+      "step": 12123
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.636332635591083e-05,
+      "loss": 0.8401,
+      "step": 12124
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.6336459903511566e-05,
+      "loss": 0.8857,
+      "step": 12125
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.630960021768824e-05,
+      "loss": 0.8538,
+      "step": 12126
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.628274729962352e-05,
+      "loss": 0.869,
+      "step": 12127
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.6255901150499935e-05,
+      "loss": 0.9109,
+      "step": 12128
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.622906177149971e-05,
+      "loss": 0.865,
+      "step": 12129
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.620222916380459e-05,
+      "loss": 0.8636,
+      "step": 12130
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.617540332859622e-05,
+      "loss": 0.9503,
+      "step": 12131
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.614858426705586e-05,
+      "loss": 0.865,
+      "step": 12132
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.612177198036458e-05,
+      "loss": 0.8684,
+      "step": 12133
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.609496646970293e-05,
+      "loss": 0.8819,
+      "step": 12134
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.6068167736251335e-05,
+      "loss": 0.8489,
+      "step": 12135
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.604137578118992e-05,
+      "loss": 0.8559,
+      "step": 12136
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.601459060569835e-05,
+      "loss": 0.8823,
+      "step": 12137
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.598781221095627e-05,
+      "loss": 0.8648,
+      "step": 12138
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.596104059814277e-05,
+      "loss": 0.8953,
+      "step": 12139
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.593427576843674e-05,
+      "loss": 0.8498,
+      "step": 12140
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.590751772301685e-05,
+      "loss": 0.887,
+      "step": 12141
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.5880766463061274e-05,
+      "loss": 0.8665,
+      "step": 12142
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.585402198974806e-05,
+      "loss": 0.8725,
+      "step": 12143
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.582728430425489e-05,
+      "loss": 0.8716,
+      "step": 12144
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.5800553407759236e-05,
+      "loss": 0.8443,
+      "step": 12145
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.577382930143808e-05,
+      "loss": 0.8886,
+      "step": 12146
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.5747111986468236e-05,
+      "loss": 0.8698,
+      "step": 12147
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.572040146402632e-05,
+      "loss": 0.8483,
+      "step": 12148
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.569369773528831e-05,
+      "loss": 0.8593,
+      "step": 12149
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.5667000801430335e-05,
+      "loss": 0.8734,
+      "step": 12150
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.564031066362786e-05,
+      "loss": 0.885,
+      "step": 12151
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.5613627323056205e-05,
+      "loss": 0.8025,
+      "step": 12152
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.558695078089046e-05,
+      "loss": 0.8956,
+      "step": 12153
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.556028103830519e-05,
+      "loss": 0.9285,
+      "step": 12154
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.5533618096474915e-05,
+      "loss": 0.8657,
+      "step": 12155
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.5506961956573555e-05,
+      "loss": 0.8868,
+      "step": 12156
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.548031261977519e-05,
+      "loss": 0.8862,
+      "step": 12157
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.54536700872531e-05,
+      "loss": 0.9102,
+      "step": 12158
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.5427034360180564e-05,
+      "loss": 0.9091,
+      "step": 12159
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.540040543973054e-05,
+      "loss": 0.8132,
+      "step": 12160
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.5373783327075534e-05,
+      "loss": 0.8949,
+      "step": 12161
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.53471680233879e-05,
+      "loss": 0.8683,
+      "step": 12162
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.532055952983964e-05,
+      "loss": 0.8725,
+      "step": 12163
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.5293957847602505e-05,
+      "loss": 0.8361,
+      "step": 12164
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.526736297784782e-05,
+      "loss": 0.8386,
+      "step": 12165
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.524077492174672e-05,
+      "loss": 0.8858,
+      "step": 12166
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.521419368047006e-05,
+      "loss": 0.8736,
+      "step": 12167
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.51876192551882e-05,
+      "loss": 0.8945,
+      "step": 12168
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.5161051647071565e-05,
+      "loss": 0.9315,
+      "step": 12169
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.513449085728987e-05,
+      "loss": 0.886,
+      "step": 12170
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.51079368870128e-05,
+      "loss": 0.8531,
+      "step": 12171
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.508138973740972e-05,
+      "loss": 0.838,
+      "step": 12172
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.505484940964948e-05,
+      "loss": 0.8799,
+      "step": 12173
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.502831590490091e-05,
+      "loss": 0.8911,
+      "step": 12174
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.5001789224332336e-05,
+      "loss": 0.8493,
+      "step": 12175
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.497526936911196e-05,
+      "loss": 0.8505,
+      "step": 12176
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.494875634040743e-05,
+      "loss": 0.8581,
+      "step": 12177
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.4922250139386426e-05,
+      "loss": 0.8044,
+      "step": 12178
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.4895750767216085e-05,
+      "loss": 0.8372,
+      "step": 12179
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.486925822506316e-05,
+      "loss": 0.8783,
+      "step": 12180
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.484277251409448e-05,
+      "loss": 0.8185,
+      "step": 12181
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.481629363547619e-05,
+      "loss": 0.8854,
+      "step": 12182
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.4789821590374395e-05,
+      "loss": 0.8314,
+      "step": 12183
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.476335637995466e-05,
+      "loss": 0.8355,
+      "step": 12184
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.473689800538246e-05,
+      "loss": 0.792,
+      "step": 12185
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.471044646782294e-05,
+      "loss": 0.8128,
+      "step": 12186
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.468400176844074e-05,
+      "loss": 0.8631,
+      "step": 12187
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.465756390840055e-05,
+      "loss": 0.9184,
+      "step": 12188
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.46311328888664e-05,
+      "loss": 0.8779,
+      "step": 12189
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.460470871100224e-05,
+      "loss": 0.8572,
+      "step": 12190
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.4578291375971717e-05,
+      "loss": 0.8897,
+      "step": 12191
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.4551880884938025e-05,
+      "loss": 0.8475,
+      "step": 12192
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.452547723906417e-05,
+      "loss": 0.8925,
+      "step": 12193
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.4499080439512854e-05,
+      "loss": 0.8902,
+      "step": 12194
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.4472690487446514e-05,
+      "loss": 0.8319,
+      "step": 12195
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.444630738402709e-05,
+      "loss": 0.7883,
+      "step": 12196
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.4419931130416534e-05,
+      "loss": 0.8514,
+      "step": 12197
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.4393561727776246e-05,
+      "loss": 0.9008,
+      "step": 12198
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.436719917726731e-05,
+      "loss": 0.9166,
+      "step": 12199
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.4340843480050764e-05,
+      "loss": 0.8294,
+      "step": 12200
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.431449463728707e-05,
+      "loss": 0.8295,
+      "step": 12201
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.428815265013655e-05,
+      "loss": 0.8592,
+      "step": 12202
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.42618175197592e-05,
+      "loss": 0.9388,
+      "step": 12203
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.4235489247314596e-05,
+      "loss": 0.9056,
+      "step": 12204
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.4209167833962165e-05,
+      "loss": 0.8941,
+      "step": 12205
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.418285328086096e-05,
+      "loss": 0.8838,
+      "step": 12206
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.415654558916977e-05,
+      "loss": 0.801,
+      "step": 12207
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.4130244760047014e-05,
+      "loss": 0.8841,
+      "step": 12208
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.410395079465084e-05,
+      "loss": 0.9062,
+      "step": 12209
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.4077663694139196e-05,
+      "loss": 0.8242,
+      "step": 12210
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.40513834596695e-05,
+      "loss": 0.8244,
+      "step": 12211
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.402511009239907e-05,
+      "loss": 0.9302,
+      "step": 12212
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.399884359348487e-05,
+      "loss": 0.815,
+      "step": 12213
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.397258396408357e-05,
+      "loss": 0.8567,
+      "step": 12214
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.394633120535136e-05,
+      "loss": 0.8194,
+      "step": 12215
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.3920085318444495e-05,
+      "loss": 0.8592,
+      "step": 12216
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.3893846304518626e-05,
+      "loss": 0.9136,
+      "step": 12217
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.386761416472904e-05,
+      "loss": 0.8461,
+      "step": 12218
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.3841388900231127e-05,
+      "loss": 0.7945,
+      "step": 12219
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.381517051217954e-05,
+      "loss": 0.817,
+      "step": 12220
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.378895900172886e-05,
+      "loss": 0.8762,
+      "step": 12221
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.3762754370033365e-05,
+      "loss": 0.8761,
+      "step": 12222
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.373655661824685e-05,
+      "loss": 0.9316,
+      "step": 12223
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.371036574752301e-05,
+      "loss": 0.8506,
+      "step": 12224
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.3684181759015164e-05,
+      "loss": 0.9181,
+      "step": 12225
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.365800465387635e-05,
+      "loss": 0.8562,
+      "step": 12226
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.363183443325915e-05,
+      "loss": 0.8718,
+      "step": 12227
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.3605671098316146e-05,
+      "loss": 0.8736,
+      "step": 12228
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.357951465019938e-05,
+      "loss": 0.8477,
+      "step": 12229
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.3553365090060514e-05,
+      "loss": 0.897,
+      "step": 12230
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.3527222419051274e-05,
+      "loss": 0.8554,
+      "step": 12231
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.3501086638322665e-05,
+      "loss": 0.8276,
+      "step": 12232
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.347495774902566e-05,
+      "loss": 0.9483,
+      "step": 12233
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.3448835752310824e-05,
+      "loss": 0.8326,
+      "step": 12234
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.3422720649328506e-05,
+      "loss": 0.9229,
+      "step": 12235
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.3396612441228634e-05,
+      "loss": 0.9028,
+      "step": 12236
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.337051112916075e-05,
+      "loss": 0.9001,
+      "step": 12237
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.334441671427447e-05,
+      "loss": 0.836,
+      "step": 12238
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.33183291977187e-05,
+      "loss": 0.8047,
+      "step": 12239
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.329224858064225e-05,
+      "loss": 0.8655,
+      "step": 12240
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.3266174864193555e-05,
+      "loss": 0.8851,
+      "step": 12241
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.3240108049520875e-05,
+      "loss": 0.8471,
+      "step": 12242
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.32140481377719e-05,
+      "loss": 0.8658,
+      "step": 12243
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.3187995130094286e-05,
+      "loss": 0.8034,
+      "step": 12244
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.316194902763531e-05,
+      "loss": 0.8934,
+      "step": 12245
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.3135909831541724e-05,
+      "loss": 0.8708,
+      "step": 12246
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.310987754296041e-05,
+      "loss": 0.8739,
+      "step": 12247
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.3083852163037584e-05,
+      "loss": 0.9179,
+      "step": 12248
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.305783369291918e-05,
+      "loss": 0.9258,
+      "step": 12249
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.303182213375112e-05,
+      "loss": 0.8773,
+      "step": 12250
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.300581748667867e-05,
+      "loss": 0.8438,
+      "step": 12251
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.297981975284699e-05,
+      "loss": 0.8093,
+      "step": 12252
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.29538289334009e-05,
+      "loss": 0.8527,
+      "step": 12253
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.292784502948495e-05,
+      "loss": 0.9196,
+      "step": 12254
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.290186804224323e-05,
+      "loss": 0.8489,
+      "step": 12255
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.2875897972819704e-05,
+      "loss": 0.9149,
+      "step": 12256
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.284993482235804e-05,
+      "loss": 0.8536,
+      "step": 12257
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.282397859200131e-05,
+      "loss": 0.8622,
+      "step": 12258
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.279802928289276e-05,
+      "loss": 0.8308,
+      "step": 12259
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.277208689617487e-05,
+      "loss": 0.8846,
+      "step": 12260
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.2746151432990104e-05,
+      "loss": 0.8292,
+      "step": 12261
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.2720222894480546e-05,
+      "loss": 0.9233,
+      "step": 12262
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.269430128178786e-05,
+      "loss": 0.8674,
+      "step": 12263
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.266838659605366e-05,
+      "loss": 0.8956,
+      "step": 12264
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.2642478838418877e-05,
+      "loss": 0.8516,
+      "step": 12265
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.261657801002461e-05,
+      "loss": 0.8703,
+      "step": 12266
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.259068411201128e-05,
+      "loss": 0.8207,
+      "step": 12267
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.256479714551902e-05,
+      "loss": 0.8761,
+      "step": 12268
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.253891711168798e-05,
+      "loss": 0.8262,
+      "step": 12269
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.251304401165761e-05,
+      "loss": 0.8627,
+      "step": 12270
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.248717784656733e-05,
+      "loss": 0.8513,
+      "step": 12271
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.246131861755611e-05,
+      "loss": 0.8666,
+      "step": 12272
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.243546632576274e-05,
+      "loss": 0.9031,
+      "step": 12273
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.24096209723255e-05,
+      "loss": 0.8092,
+      "step": 12274
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.2383782558382553e-05,
+      "loss": 0.8055,
+      "step": 12275
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.235795108507177e-05,
+      "loss": 0.8874,
+      "step": 12276
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.2332126553530446e-05,
+      "loss": 0.8922,
+      "step": 12277
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.2306308964896e-05,
+      "loss": 0.8391,
+      "step": 12278
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.228049832030514e-05,
+      "loss": 0.8291,
+      "step": 12279
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.225469462089451e-05,
+      "loss": 0.8707,
+      "step": 12280
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.2228897867800396e-05,
+      "loss": 0.9001,
+      "step": 12281
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.220310806215866e-05,
+      "loss": 0.877,
+      "step": 12282
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.2177325205105046e-05,
+      "loss": 0.8163,
+      "step": 12283
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.215154929777485e-05,
+      "loss": 0.8354,
+      "step": 12284
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.212578034130321e-05,
+      "loss": 0.9061,
+      "step": 12285
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.2100018336824735e-05,
+      "loss": 0.8335,
+      "step": 12286
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.2074263285473905e-05,
+      "loss": 0.893,
+      "step": 12287
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.204851518838491e-05,
+      "loss": 0.8705,
+      "step": 12288
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.202277404669144e-05,
+      "loss": 0.8731,
+      "step": 12289
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.19970398615271e-05,
+      "loss": 0.9341,
+      "step": 12290
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.197131263402505e-05,
+      "loss": 0.8366,
+      "step": 12291
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.1945592365318275e-05,
+      "loss": 0.8292,
+      "step": 12292
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.191987905653925e-05,
+      "loss": 0.8481,
+      "step": 12293
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.1894172708820304e-05,
+      "loss": 0.868,
+      "step": 12294
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.186847332329347e-05,
+      "loss": 0.895,
+      "step": 12295
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.184278090109028e-05,
+      "loss": 0.8231,
+      "step": 12296
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.181709544334231e-05,
+      "loss": 0.9051,
+      "step": 12297
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.179141695118045e-05,
+      "loss": 0.8845,
+      "step": 12298
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.1765745425735506e-05,
+      "loss": 0.8023,
+      "step": 12299
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.174008086813797e-05,
+      "loss": 0.8941,
+      "step": 12300
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.171442327951789e-05,
+      "loss": 0.8428,
+      "step": 12301
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.168877266100515e-05,
+      "loss": 0.897,
+      "step": 12302
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.166312901372926e-05,
+      "loss": 0.8343,
+      "step": 12303
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.163749233881951e-05,
+      "loss": 0.8368,
+      "step": 12304
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.161186263740468e-05,
+      "loss": 0.8391,
+      "step": 12305
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.1586239910613454e-05,
+      "loss": 0.8485,
+      "step": 12306
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.156062415957418e-05,
+      "loss": 0.7987,
+      "step": 12307
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.153501538541469e-05,
+      "loss": 0.9388,
+      "step": 12308
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.150941358926286e-05,
+      "loss": 0.9192,
+      "step": 12309
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.1483818772245894e-05,
+      "loss": 0.8608,
+      "step": 12310
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.1458230935490974e-05,
+      "loss": 0.8552,
+      "step": 12311
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.143265008012485e-05,
+      "loss": 0.8529,
+      "step": 12312
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.140707620727391e-05,
+      "loss": 0.9315,
+      "step": 12313
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.138150931806433e-05,
+      "loss": 0.8398,
+      "step": 12314
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.135594941362195e-05,
+      "loss": 0.8649,
+      "step": 12315
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.1330396495072356e-05,
+      "loss": 0.8002,
+      "step": 12316
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.130485056354068e-05,
+      "loss": 0.8769,
+      "step": 12317
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.1279311620151886e-05,
+      "loss": 0.8618,
+      "step": 12318
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.125377966603061e-05,
+      "loss": 0.8072,
+      "step": 12319
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.122825470230107e-05,
+      "loss": 0.9178,
+      "step": 12320
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.12027367300873e-05,
+      "loss": 0.8742,
+      "step": 12321
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.117722575051299e-05,
+      "loss": 0.8481,
+      "step": 12322
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.115172176470157e-05,
+      "loss": 0.8777,
+      "step": 12323
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.1126224773776014e-05,
+      "loss": 0.8577,
+      "step": 12324
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.11007347788591e-05,
+      "loss": 0.8975,
+      "step": 12325
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.107525178107336e-05,
+      "loss": 0.8238,
+      "step": 12326
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.1049775781540786e-05,
+      "loss": 0.8617,
+      "step": 12327
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.1024306781383404e-05,
+      "loss": 0.8953,
+      "step": 12328
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.09988447817226e-05,
+      "loss": 0.8524,
+      "step": 12329
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.097338978367964e-05,
+      "loss": 0.8732,
+      "step": 12330
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.094794178837549e-05,
+      "loss": 0.9174,
+      "step": 12331
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.092250079693063e-05,
+      "loss": 0.8426,
+      "step": 12332
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.089706681046541e-05,
+      "loss": 0.886,
+      "step": 12333
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.087163983009985e-05,
+      "loss": 0.8178,
+      "step": 12334
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.084621985695365e-05,
+      "loss": 0.8419,
+      "step": 12335
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.082080689214609e-05,
+      "loss": 0.8771,
+      "step": 12336
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.079540093679628e-05,
+      "loss": 0.8234,
+      "step": 12337
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.0770001992022986e-05,
+      "loss": 0.8472,
+      "step": 12338
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.0744610058944545e-05,
+      "loss": 0.8794,
+      "step": 12339
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.071922513867929e-05,
+      "loss": 0.9075,
+      "step": 12340
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.069384723234486e-05,
+      "loss": 0.8964,
+      "step": 12341
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.066847634105893e-05,
+      "loss": 0.8348,
+      "step": 12342
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.064311246593855e-05,
+      "loss": 0.8677,
+      "step": 12343
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.061775560810068e-05,
+      "loss": 0.8805,
+      "step": 12344
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.059240576866197e-05,
+      "loss": 0.8109,
+      "step": 12345
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.0567062948738554e-05,
+      "loss": 0.8779,
+      "step": 12346
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.0541727149446616e-05,
+      "loss": 0.9161,
+      "step": 12347
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.0516398371901645e-05,
+      "loss": 0.9263,
+      "step": 12348
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.049107661721903e-05,
+      "loss": 0.8456,
+      "step": 12349
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.04657618865139e-05,
+      "loss": 0.894,
+      "step": 12350
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.0440454180900875e-05,
+      "loss": 0.8407,
+      "step": 12351
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.041515350149443e-05,
+      "loss": 0.9273,
+      "step": 12352
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.0389859849408684e-05,
+      "loss": 0.8462,
+      "step": 12353
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.0364573225757485e-05,
+      "loss": 0.8243,
+      "step": 12354
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.033929363165425e-05,
+      "loss": 0.8746,
+      "step": 12355
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.0314021068212185e-05,
+      "loss": 0.8352,
+      "step": 12356
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.0288755536544255e-05,
+      "loss": 0.8632,
+      "step": 12357
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.026349703776282e-05,
+      "loss": 0.8766,
+      "step": 12358
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.023824557298039e-05,
+      "loss": 0.9262,
+      "step": 12359
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.021300114330877e-05,
+      "loss": 0.8477,
+      "step": 12360
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.018776374985962e-05,
+      "loss": 0.8737,
+      "step": 12361
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.016253339374432e-05,
+      "loss": 0.8368,
+      "step": 12362
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.013731007607378e-05,
+      "loss": 0.8726,
+      "step": 12363
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.011209379795879e-05,
+      "loss": 0.8559,
+      "step": 12364
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.008688456050973e-05,
+      "loss": 0.8836,
+      "step": 12365
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.006168236483676e-05,
+      "loss": 0.9431,
+      "step": 12366
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.0036487212049514e-05,
+      "loss": 0.8787,
+      "step": 12367
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.001129910325756e-05,
+      "loss": 0.8098,
+      "step": 12368
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.998611803957009e-05,
+      "loss": 0.8674,
+      "step": 12369
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.996094402209582e-05,
+      "loss": 0.8127,
+      "step": 12370
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.993577705194338e-05,
+      "loss": 0.8645,
+      "step": 12371
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.991061713022099e-05,
+      "loss": 0.9209,
+      "step": 12372
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.988546425803661e-05,
+      "loss": 0.8456,
+      "step": 12373
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.986031843649773e-05,
+      "loss": 0.8514,
+      "step": 12374
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.983517966671175e-05,
+      "loss": 0.8618,
+      "step": 12375
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.981004794978564e-05,
+      "loss": 0.875,
+      "step": 12376
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.978492328682597e-05,
+      "loss": 0.8512,
+      "step": 12377
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.975980567893929e-05,
+      "loss": 0.9106,
+      "step": 12378
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9734695127231515e-05,
+      "loss": 0.8456,
+      "step": 12379
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.970959163280841e-05,
+      "loss": 0.8697,
+      "step": 12380
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.968449519677549e-05,
+      "loss": 0.8191,
+      "step": 12381
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9659405820237747e-05,
+      "loss": 0.8574,
+      "step": 12382
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9634323504300077e-05,
+      "loss": 0.8263,
+      "step": 12383
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.960924825006695e-05,
+      "loss": 0.8732,
+      "step": 12384
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.958418005864264e-05,
+      "loss": 0.8917,
+      "step": 12385
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9559118931130865e-05,
+      "loss": 0.8794,
+      "step": 12386
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.953406486863531e-05,
+      "loss": 0.8435,
+      "step": 12387
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9509017872259225e-05,
+      "loss": 0.8559,
+      "step": 12388
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.948397794310546e-05,
+      "loss": 0.8725,
+      "step": 12389
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.945894508227679e-05,
+      "loss": 0.8631,
+      "step": 12390
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.943391929087543e-05,
+      "loss": 0.8759,
+      "step": 12391
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.940890057000342e-05,
+      "loss": 0.8819,
+      "step": 12392
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9383888920762525e-05,
+      "loss": 0.8463,
+      "step": 12393
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.935888434425401e-05,
+      "loss": 0.887,
+      "step": 12394
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.933388684157901e-05,
+      "loss": 0.9456,
+      "step": 12395
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9308896413838324e-05,
+      "loss": 0.9121,
+      "step": 12396
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9283913062132415e-05,
+      "loss": 0.8714,
+      "step": 12397
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.925893678756134e-05,
+      "loss": 0.9046,
+      "step": 12398
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.923396759122497e-05,
+      "loss": 0.8643,
+      "step": 12399
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9209005474222906e-05,
+      "loss": 0.8479,
+      "step": 12400
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9184050437654205e-05,
+      "loss": 0.9372,
+      "step": 12401
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9159102482617847e-05,
+      "loss": 0.8937,
+      "step": 12402
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.913416161021242e-05,
+      "loss": 0.8627,
+      "step": 12403
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9109227821536234e-05,
+      "loss": 0.8685,
+      "step": 12404
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.908430111768713e-05,
+      "loss": 0.8406,
+      "step": 12405
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9059381499762824e-05,
+      "loss": 0.8633,
+      "step": 12406
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.903446896886071e-05,
+      "loss": 0.8445,
+      "step": 12407
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9009563526077654e-05,
+      "loss": 0.8876,
+      "step": 12408
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.898466517251056e-05,
+      "loss": 0.8813,
+      "step": 12409
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.895977390925567e-05,
+      "loss": 0.8569,
+      "step": 12410
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.893488973740915e-05,
+      "loss": 0.889,
+      "step": 12411
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.891001265806682e-05,
+      "loss": 0.8733,
+      "step": 12412
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.888514267232401e-05,
+      "loss": 0.8886,
+      "step": 12413
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.886027978127596e-05,
+      "loss": 0.9293,
+      "step": 12414
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.8835423986017475e-05,
+      "loss": 0.8457,
+      "step": 12415
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.8810575287643155e-05,
+      "loss": 0.8993,
+      "step": 12416
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.8785733687247094e-05,
+      "loss": 0.7963,
+      "step": 12417
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.876089918592325e-05,
+      "loss": 0.8884,
+      "step": 12418
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.873607178476526e-05,
+      "loss": 0.868,
+      "step": 12419
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.871125148486627e-05,
+      "loss": 0.8596,
+      "step": 12420
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.868643828731942e-05,
+      "loss": 0.8829,
+      "step": 12421
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.866163219321719e-05,
+      "loss": 0.8309,
+      "step": 12422
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.8636833203652056e-05,
+      "loss": 0.896,
+      "step": 12423
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.8612041319715894e-05,
+      "loss": 0.8534,
+      "step": 12424
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.858725654250053e-05,
+      "loss": 0.7832,
+      "step": 12425
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.8562478873097366e-05,
+      "loss": 0.8398,
+      "step": 12426
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.853770831259735e-05,
+      "loss": 0.8293,
+      "step": 12427
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.851294486209145e-05,
+      "loss": 0.8881,
+      "step": 12428
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.848818852266998e-05,
+      "loss": 0.8621,
+      "step": 12429
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.846343929542313e-05,
+      "loss": 0.8803,
+      "step": 12430
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.843869718144077e-05,
+      "loss": 0.8867,
+      "step": 12431
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.841396218181237e-05,
+      "loss": 0.8702,
+      "step": 12432
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.838923429762713e-05,
+      "loss": 0.8975,
+      "step": 12433
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.8364513529973964e-05,
+      "loss": 0.8904,
+      "step": 12434
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.83397998799415e-05,
+      "loss": 0.8223,
+      "step": 12435
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.831509334861789e-05,
+      "loss": 0.9302,
+      "step": 12436
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.8290393937091173e-05,
+      "loss": 0.8534,
+      "step": 12437
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.8265701646449025e-05,
+      "loss": 0.8587,
+      "step": 12438
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.824101647777858e-05,
+      "loss": 0.7806,
+      "step": 12439
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.82163384321671e-05,
+      "loss": 0.8633,
+      "step": 12440
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.819166751070111e-05,
+      "loss": 0.8271,
+      "step": 12441
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.8167003714467064e-05,
+      "loss": 0.8534,
+      "step": 12442
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.8142347044551044e-05,
+      "loss": 0.9524,
+      "step": 12443
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.811769750203875e-05,
+      "loss": 0.8342,
+      "step": 12444
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.809305508801564e-05,
+      "loss": 0.8171,
+      "step": 12445
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.806841980356688e-05,
+      "loss": 0.8877,
+      "step": 12446
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.8043791649777295e-05,
+      "loss": 0.8946,
+      "step": 12447
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.801917062773131e-05,
+      "loss": 0.8708,
+      "step": 12448
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7994556738513156e-05,
+      "loss": 0.8869,
+      "step": 12449
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.796994998320675e-05,
+      "loss": 0.8622,
+      "step": 12450
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.794535036289557e-05,
+      "loss": 0.798,
+      "step": 12451
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.792075787866289e-05,
+      "loss": 0.8875,
+      "step": 12452
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.789617253159163e-05,
+      "loss": 0.8653,
+      "step": 12453
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.78715943227645e-05,
+      "loss": 0.9542,
+      "step": 12454
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.784702325326368e-05,
+      "loss": 0.9028,
+      "step": 12455
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.782245932417119e-05,
+      "loss": 0.8556,
+      "step": 12456
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.779790253656877e-05,
+      "loss": 0.8733,
+      "step": 12457
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7773352891537626e-05,
+      "loss": 0.8856,
+      "step": 12458
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.774881039015901e-05,
+      "loss": 0.8507,
+      "step": 12459
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7724275033513454e-05,
+      "loss": 0.8504,
+      "step": 12460
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.769974682268149e-05,
+      "loss": 0.8422,
+      "step": 12461
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.767522575874325e-05,
+      "loss": 0.8252,
+      "step": 12462
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7650711842778394e-05,
+      "loss": 0.8624,
+      "step": 12463
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.762620507586645e-05,
+      "loss": 0.8792,
+      "step": 12464
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.760170545908659e-05,
+      "loss": 0.8201,
+      "step": 12465
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.757721299351768e-05,
+      "loss": 0.8486,
+      "step": 12466
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7552727680238185e-05,
+      "loss": 0.8428,
+      "step": 12467
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.752824952032634e-05,
+      "loss": 0.822,
+      "step": 12468
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.750377851486009e-05,
+      "loss": 0.8885,
+      "step": 12469
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7479314664916856e-05,
+      "loss": 0.8556,
+      "step": 12470
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.745485797157413e-05,
+      "loss": 0.8158,
+      "step": 12471
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.74304084359087e-05,
+      "loss": 0.9305,
+      "step": 12472
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.740596605899724e-05,
+      "loss": 0.91,
+      "step": 12473
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.738153084191616e-05,
+      "loss": 0.9234,
+      "step": 12474
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.735710278574132e-05,
+      "loss": 0.903,
+      "step": 12475
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7332681891548503e-05,
+      "loss": 0.8593,
+      "step": 12476
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.730826816041299e-05,
+      "loss": 0.8668,
+      "step": 12477
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7283861593409996e-05,
+      "loss": 0.8835,
+      "step": 12478
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7259462191614134e-05,
+      "loss": 0.7896,
+      "step": 12479
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7235069956099865e-05,
+      "loss": 0.8309,
+      "step": 12480
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7210684887941375e-05,
+      "loss": 0.8505,
+      "step": 12481
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.718630698821233e-05,
+      "loss": 0.872,
+      "step": 12482
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.716193625798627e-05,
+      "loss": 0.8478,
+      "step": 12483
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.713757269833637e-05,
+      "loss": 0.9168,
+      "step": 12484
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.711321631033555e-05,
+      "loss": 0.955,
+      "step": 12485
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.708886709505621e-05,
+      "loss": 0.8606,
+      "step": 12486
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.706452505357063e-05,
+      "loss": 0.8147,
+      "step": 12487
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7040190186950753e-05,
+      "loss": 0.8684,
+      "step": 12488
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.701586249626803e-05,
+      "loss": 0.8111,
+      "step": 12489
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.6991541982593935e-05,
+      "loss": 0.8998,
+      "step": 12490
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.696722864699924e-05,
+      "loss": 0.9201,
+      "step": 12491
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.694292249055467e-05,
+      "loss": 0.8862,
+      "step": 12492
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.6918623514330575e-05,
+      "loss": 0.8499,
+      "step": 12493
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.689433171939687e-05,
+      "loss": 0.8796,
+      "step": 12494
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.687004710682329e-05,
+      "loss": 0.8809,
+      "step": 12495
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.6845769677679207e-05,
+      "loss": 0.8861,
+      "step": 12496
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.6821499433033746e-05,
+      "loss": 0.8915,
+      "step": 12497
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.679723637395551e-05,
+      "loss": 0.8027,
+      "step": 12498
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.677298050151301e-05,
+      "loss": 0.8871,
+      "step": 12499
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.674873181677438e-05,
+      "loss": 0.8784,
+      "step": 12500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.6724490320807295e-05,
+      "loss": 0.9253,
+      "step": 12501
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.670025601467939e-05,
+      "loss": 0.8915,
+      "step": 12502
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.667602889945767e-05,
+      "loss": 0.8473,
+      "step": 12503
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.6651808976209126e-05,
+      "loss": 0.8051,
+      "step": 12504
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.662759624600012e-05,
+      "loss": 0.8623,
+      "step": 12505
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.660339070989696e-05,
+      "loss": 0.8644,
+      "step": 12506
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.657919236896556e-05,
+      "loss": 0.8058,
+      "step": 12507
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.6555001224271354e-05,
+      "loss": 0.9059,
+      "step": 12508
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.653081727687979e-05,
+      "loss": 0.8715,
+      "step": 12509
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.650664052785568e-05,
+      "loss": 0.8886,
+      "step": 12510
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.648247097826367e-05,
+      "loss": 0.8592,
+      "step": 12511
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.645830862916815e-05,
+      "loss": 0.8081,
+      "step": 12512
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.643415348163297e-05,
+      "loss": 0.836,
+      "step": 12513
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.641000553672187e-05,
+      "loss": 0.8641,
+      "step": 12514
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.638586479549822e-05,
+      "loss": 0.8936,
+      "step": 12515
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.636173125902509e-05,
+      "loss": 0.8768,
+      "step": 12516
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.633760492836511e-05,
+      "loss": 0.8483,
+      "step": 12517
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.6313485804580716e-05,
+      "loss": 0.8083,
+      "step": 12518
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.628937388873406e-05,
+      "loss": 0.9281,
+      "step": 12519
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.6265269181886754e-05,
+      "loss": 0.8542,
+      "step": 12520
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.624117168510046e-05,
+      "loss": 0.8498,
+      "step": 12521
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.621708139943614e-05,
+      "loss": 0.8298,
+      "step": 12522
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.619299832595469e-05,
+      "loss": 0.8196,
+      "step": 12523
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.6168922465716616e-05,
+      "loss": 0.8628,
+      "step": 12524
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.614485381978205e-05,
+      "loss": 0.8166,
+      "step": 12525
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.6120792389210866e-05,
+      "loss": 0.8641,
+      "step": 12526
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.609673817506263e-05,
+      "loss": 0.8477,
+      "step": 12527
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.60726911783966e-05,
+      "loss": 0.8795,
+      "step": 12528
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.6048651400271625e-05,
+      "loss": 0.814,
+      "step": 12529
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.6024618841746306e-05,
+      "loss": 0.8598,
+      "step": 12530
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.6000593503878965e-05,
+      "loss": 0.9082,
+      "step": 12531
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.59765753877275e-05,
+      "loss": 0.9091,
+      "step": 12532
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.5952564494349536e-05,
+      "loss": 0.8407,
+      "step": 12533
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.592856082480247e-05,
+      "loss": 0.8835,
+      "step": 12534
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.590456438014329e-05,
+      "loss": 0.8761,
+      "step": 12535
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.588057516142862e-05,
+      "loss": 0.8324,
+      "step": 12536
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.585659316971486e-05,
+      "loss": 0.8555,
+      "step": 12537
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.58326184060581e-05,
+      "loss": 0.8378,
+      "step": 12538
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.5808650871513905e-05,
+      "loss": 0.8502,
+      "step": 12539
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.578469056713793e-05,
+      "loss": 0.8852,
+      "step": 12540
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.5760737493985094e-05,
+      "loss": 0.8382,
+      "step": 12541
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.573679165311021e-05,
+      "loss": 0.8545,
+      "step": 12542
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.571285304556782e-05,
+      "loss": 0.8248,
+      "step": 12543
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.5688921672411915e-05,
+      "loss": 0.9196,
+      "step": 12544
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.566499753469637e-05,
+      "loss": 0.8295,
+      "step": 12545
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.564108063347471e-05,
+      "loss": 0.8016,
+      "step": 12546
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.5617170969800174e-05,
+      "loss": 0.8624,
+      "step": 12547
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.559326854472547e-05,
+      "loss": 0.9071,
+      "step": 12548
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.5569373359303236e-05,
+      "loss": 0.9511,
+      "step": 12549
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.554548541458576e-05,
+      "loss": 0.8542,
+      "step": 12550
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.5521604711624756e-05,
+      "loss": 0.9721,
+      "step": 12551
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.549773125147202e-05,
+      "loss": 0.8929,
+      "step": 12552
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.547386503517869e-05,
+      "loss": 0.8698,
+      "step": 12553
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.545000606379576e-05,
+      "loss": 0.861,
+      "step": 12554
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.5426154338373905e-05,
+      "loss": 0.8869,
+      "step": 12555
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.5402309859963335e-05,
+      "loss": 0.9267,
+      "step": 12556
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.5378472629614135e-05,
+      "loss": 0.8605,
+      "step": 12557
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.535464264837585e-05,
+      "loss": 0.8123,
+      "step": 12558
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.5330819917298006e-05,
+      "loss": 0.8804,
+      "step": 12559
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.5307004437429495e-05,
+      "loss": 0.8368,
+      "step": 12560
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.5283196209819104e-05,
+      "loss": 0.8408,
+      "step": 12561
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.5259395235515244e-05,
+      "loss": 0.9111,
+      "step": 12562
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.523560151556593e-05,
+      "loss": 0.894,
+      "step": 12563
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.521181505101894e-05,
+      "loss": 0.7902,
+      "step": 12564
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.518803584292172e-05,
+      "loss": 0.8941,
+      "step": 12565
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.516426389232144e-05,
+      "loss": 0.8545,
+      "step": 12566
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.514049920026479e-05,
+      "loss": 0.8917,
+      "step": 12567
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.511674176779829e-05,
+      "loss": 0.8942,
+      "step": 12568
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.509299159596817e-05,
+      "loss": 0.8627,
+      "step": 12569
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.506924868582011e-05,
+      "loss": 0.8292,
+      "step": 12570
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.504551303839985e-05,
+      "loss": 0.9037,
+      "step": 12571
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.502178465475241e-05,
+      "loss": 0.8342,
+      "step": 12572
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.499806353592272e-05,
+      "loss": 0.9059,
+      "step": 12573
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.497434968295541e-05,
+      "loss": 0.8918,
+      "step": 12574
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.495064309689462e-05,
+      "loss": 0.8379,
+      "step": 12575
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.4926943778784294e-05,
+      "loss": 0.9116,
+      "step": 12576
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.490325172966804e-05,
+      "loss": 0.9006,
+      "step": 12577
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.487956695058921e-05,
+      "loss": 0.8676,
+      "step": 12578
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.485588944259064e-05,
+      "loss": 0.8866,
+      "step": 12579
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.4832219206715044e-05,
+      "loss": 0.8832,
+      "step": 12580
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.4808556244004744e-05,
+      "loss": 0.8535,
+      "step": 12581
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.4784900555501655e-05,
+      "loss": 0.8919,
+      "step": 12582
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.4761252142247593e-05,
+      "loss": 0.8319,
+      "step": 12583
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.473761100528381e-05,
+      "loss": 0.859,
+      "step": 12584
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.471397714565141e-05,
+      "loss": 0.867,
+      "step": 12585
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.469035056439105e-05,
+      "loss": 0.7884,
+      "step": 12586
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.466673126254314e-05,
+      "loss": 0.9434,
+      "step": 12587
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.4643119241147826e-05,
+      "loss": 0.8899,
+      "step": 12588
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.46195145012447e-05,
+      "loss": 0.8768,
+      "step": 12589
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.459591704387342e-05,
+      "loss": 0.8304,
+      "step": 12590
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.4572326870072927e-05,
+      "loss": 0.8964,
+      "step": 12591
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.454874398088206e-05,
+      "loss": 0.8687,
+      "step": 12592
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.4525168377339366e-05,
+      "loss": 0.8884,
+      "step": 12593
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.450160006048291e-05,
+      "loss": 0.9456,
+      "step": 12594
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.4478039031350517e-05,
+      "loss": 0.9828,
+      "step": 12595
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.4454485290979744e-05,
+      "loss": 0.868,
+      "step": 12596
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.44309388404078e-05,
+      "loss": 0.8292,
+      "step": 12597
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.4407399680671484e-05,
+      "loss": 0.7929,
+      "step": 12598
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.438386781280738e-05,
+      "loss": 0.8394,
+      "step": 12599
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.436034323785175e-05,
+      "loss": 0.8276,
+      "step": 12600
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.433682595684038e-05,
+      "loss": 0.8619,
+      "step": 12601
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.431331597080902e-05,
+      "loss": 0.8379,
+      "step": 12602
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.428981328079279e-05,
+      "loss": 0.899,
+      "step": 12603
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.4266317887826705e-05,
+      "loss": 0.9071,
+      "step": 12604
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.424282979294542e-05,
+      "loss": 0.8882,
+      "step": 12605
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.421934899718311e-05,
+      "loss": 0.9283,
+      "step": 12606
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.419587550157384e-05,
+      "loss": 0.8682,
+      "step": 12607
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.4172409307151246e-05,
+      "loss": 0.8944,
+      "step": 12608
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.414895041494872e-05,
+      "loss": 0.895,
+      "step": 12609
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.412549882599918e-05,
+      "loss": 0.9153,
+      "step": 12610
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.410205454133535e-05,
+      "loss": 0.8388,
+      "step": 12611
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.4078617561989646e-05,
+      "loss": 0.8674,
+      "step": 12612
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.405518788899402e-05,
+      "loss": 0.876,
+      "step": 12613
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.4031765523380275e-05,
+      "loss": 0.8541,
+      "step": 12614
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.40083504661798e-05,
+      "loss": 0.8797,
+      "step": 12615
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.3984942718423695e-05,
+      "loss": 0.8255,
+      "step": 12616
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.3961542281142675e-05,
+      "loss": 0.853,
+      "step": 12617
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.3938149155367195e-05,
+      "loss": 0.8127,
+      "step": 12618
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.391476334212742e-05,
+      "loss": 0.8704,
+      "step": 12619
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.3891384842453e-05,
+      "loss": 0.8992,
+      "step": 12620
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.386801365737362e-05,
+      "loss": 0.8969,
+      "step": 12621
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.384464978791826e-05,
+      "loss": 0.8507,
+      "step": 12622
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.382129323511582e-05,
+      "loss": 0.9131,
+      "step": 12623
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.379794399999483e-05,
+      "loss": 0.8991,
+      "step": 12624
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.377460208358341e-05,
+      "loss": 0.8225,
+      "step": 12625
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.375126748690942e-05,
+      "loss": 0.9386,
+      "step": 12626
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.372794021100045e-05,
+      "loss": 0.8637,
+      "step": 12627
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.3704620256883745e-05,
+      "loss": 0.8402,
+      "step": 12628
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.36813076255861e-05,
+      "loss": 0.909,
+      "step": 12629
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.3658002318134144e-05,
+      "loss": 0.8747,
+      "step": 12630
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.363470433555416e-05,
+      "loss": 0.8805,
+      "step": 12631
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.361141367887193e-05,
+      "loss": 0.8869,
+      "step": 12632
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.3588130349113276e-05,
+      "loss": 0.8915,
+      "step": 12633
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.35648543473033e-05,
+      "loss": 0.8571,
+      "step": 12634
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.354158567446706e-05,
+      "loss": 0.9149,
+      "step": 12635
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.351832433162918e-05,
+      "loss": 0.8161,
+      "step": 12636
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.349507031981391e-05,
+      "loss": 0.8646,
+      "step": 12637
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.347182364004535e-05,
+      "loss": 0.8643,
+      "step": 12638
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.3448584293346984e-05,
+      "loss": 0.8533,
+      "step": 12639
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.34253522807424e-05,
+      "loss": 0.8984,
+      "step": 12640
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.340212760325441e-05,
+      "loss": 0.8895,
+      "step": 12641
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.337891026190583e-05,
+      "loss": 0.9202,
+      "step": 12642
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.3355700257719034e-05,
+      "loss": 0.8496,
+      "step": 12643
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.333249759171599e-05,
+      "loss": 0.8504,
+      "step": 12644
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.330930226491851e-05,
+      "loss": 0.8251,
+      "step": 12645
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.328611427834796e-05,
+      "loss": 0.898,
+      "step": 12646
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.3262933633025485e-05,
+      "loss": 0.8783,
+      "step": 12647
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.323976032997174e-05,
+      "loss": 0.8389,
+      "step": 12648
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.3216594370207235e-05,
+      "loss": 0.8483,
+      "step": 12649
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.319343575475211e-05,
+      "loss": 0.8417,
+      "step": 12650
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.3170284484626004e-05,
+      "loss": 0.8051,
+      "step": 12651
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.31471405608486e-05,
+      "loss": 0.9274,
+      "step": 12652
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.312400398443887e-05,
+      "loss": 0.8711,
+      "step": 12653
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.31008747564157e-05,
+      "loss": 0.8693,
+      "step": 12654
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.3077752877797644e-05,
+      "loss": 0.8794,
+      "step": 12655
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.3054638349602763e-05,
+      "loss": 0.8566,
+      "step": 12656
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.303153117284894e-05,
+      "loss": 0.8778,
+      "step": 12657
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.300843134855372e-05,
+      "loss": 0.8692,
+      "step": 12658
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.298533887773434e-05,
+      "loss": 0.8564,
+      "step": 12659
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.296225376140758e-05,
+      "loss": 0.8719,
+      "step": 12660
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.293917600059007e-05,
+      "loss": 0.813,
+      "step": 12661
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.291610559629805e-05,
+      "loss": 0.837,
+      "step": 12662
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.2893042549547305e-05,
+      "loss": 0.8575,
+      "step": 12663
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.286998686135361e-05,
+      "loss": 0.8792,
+      "step": 12664
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.2846938532732044e-05,
+      "loss": 0.9068,
+      "step": 12665
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.2823897564697683e-05,
+      "loss": 0.8902,
+      "step": 12666
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.280086395826501e-05,
+      "loss": 0.8606,
+      "step": 12667
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.2777837714448375e-05,
+      "loss": 0.8602,
+      "step": 12668
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.275481883426177e-05,
+      "loss": 0.8557,
+      "step": 12669
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.27318073187187e-05,
+      "loss": 0.8862,
+      "step": 12670
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.270880316883267e-05,
+      "loss": 0.8381,
+      "step": 12671
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.2685806385616534e-05,
+      "loss": 0.8937,
+      "step": 12672
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.2662816970082974e-05,
+      "loss": 0.8309,
+      "step": 12673
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.263983492324438e-05,
+      "loss": 0.9081,
+      "step": 12674
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.26168602461127e-05,
+      "loss": 0.8392,
+      "step": 12675
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.259389293969966e-05,
+      "loss": 0.9309,
+      "step": 12676
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.257093300501661e-05,
+      "loss": 0.8093,
+      "step": 12677
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.2547980443074674e-05,
+      "loss": 0.8682,
+      "step": 12678
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.252503525488444e-05,
+      "loss": 0.8537,
+      "step": 12679
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.2502097441456333e-05,
+      "loss": 0.8626,
+      "step": 12680
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.247916700380052e-05,
+      "loss": 0.9109,
+      "step": 12681
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.245624394292657e-05,
+      "loss": 0.874,
+      "step": 12682
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.243332825984406e-05,
+      "loss": 0.8345,
+      "step": 12683
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.2410419955562e-05,
+      "loss": 0.894,
+      "step": 12684
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.2387519031089185e-05,
+      "loss": 0.8814,
+      "step": 12685
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.236462548743409e-05,
+      "loss": 0.8278,
+      "step": 12686
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.234173932560474e-05,
+      "loss": 0.8081,
+      "step": 12687
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.231886054660895e-05,
+      "loss": 0.9193,
+      "step": 12688
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.229598915145426e-05,
+      "loss": 0.9318,
+      "step": 12689
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.22731251411478e-05,
+      "loss": 0.8074,
+      "step": 12690
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.22502685166963e-05,
+      "loss": 0.9014,
+      "step": 12691
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.222741927910633e-05,
+      "loss": 0.9146,
+      "step": 12692
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.220457742938408e-05,
+      "loss": 0.8349,
+      "step": 12693
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.218174296853527e-05,
+      "loss": 0.9153,
+      "step": 12694
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.21589158975655e-05,
+      "loss": 0.8775,
+      "step": 12695
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.213609621747997e-05,
+      "loss": 0.8597,
+      "step": 12696
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.211328392928356e-05,
+      "loss": 0.8138,
+      "step": 12697
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.209047903398072e-05,
+      "loss": 0.845,
+      "step": 12698
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.206768153257573e-05,
+      "loss": 0.8801,
+      "step": 12699
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.204489142607252e-05,
+      "loss": 0.8861,
+      "step": 12700
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.202210871547451e-05,
+      "loss": 0.8764,
+      "step": 12701
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.199933340178511e-05,
+      "loss": 0.8457,
+      "step": 12702
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.1976565486007094e-05,
+      "loss": 0.8476,
+      "step": 12703
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.195380496914313e-05,
+      "loss": 0.8653,
+      "step": 12704
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.193105185219547e-05,
+      "loss": 0.9367,
+      "step": 12705
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.1908306136166e-05,
+      "loss": 0.8325,
+      "step": 12706
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.188556782205634e-05,
+      "loss": 0.9123,
+      "step": 12707
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.186283691086778e-05,
+      "loss": 0.8766,
+      "step": 12708
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.1840113403601356e-05,
+      "loss": 0.9414,
+      "step": 12709
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.1817397301257544e-05,
+      "loss": 0.8732,
+      "step": 12710
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.179468860483677e-05,
+      "loss": 0.9009,
+      "step": 12711
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.1771987315338993e-05,
+      "loss": 0.8877,
+      "step": 12712
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.174929343376374e-05,
+      "loss": 0.8287,
+      "step": 12713
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.172660696111054e-05,
+      "loss": 0.8866,
+      "step": 12714
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.170392789837826e-05,
+      "loss": 0.8164,
+      "step": 12715
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.1681256246565596e-05,
+      "loss": 0.843,
+      "step": 12716
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.165859200667094e-05,
+      "loss": 0.8924,
+      "step": 12717
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.1635935179692235e-05,
+      "loss": 0.8512,
+      "step": 12718
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.161328576662726e-05,
+      "loss": 0.9501,
+      "step": 12719
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.159064376847323e-05,
+      "loss": 0.7606,
+      "step": 12720
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.15680091862274e-05,
+      "loss": 0.8748,
+      "step": 12721
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.154538202088633e-05,
+      "loss": 0.9019,
+      "step": 12722
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.152276227344644e-05,
+      "loss": 0.854,
+      "step": 12723
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.1500149944903844e-05,
+      "loss": 0.9473,
+      "step": 12724
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.147754503625422e-05,
+      "loss": 0.8465,
+      "step": 12725
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.1454947548492964e-05,
+      "loss": 0.8689,
+      "step": 12726
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.143235748261519e-05,
+      "loss": 0.7834,
+      "step": 12727
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.140977483961571e-05,
+      "loss": 0.8776,
+      "step": 12728
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.1387199620488816e-05,
+      "loss": 0.8721,
+      "step": 12729
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.1364631826228706e-05,
+      "loss": 0.8548,
+      "step": 12730
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.134207145782915e-05,
+      "loss": 0.8141,
+      "step": 12731
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.1319518516283495e-05,
+      "loss": 0.8848,
+      "step": 12732
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.1296973002585004e-05,
+      "loss": 0.8894,
+      "step": 12733
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.127443491772639e-05,
+      "loss": 0.9372,
+      "step": 12734
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.12519042627001e-05,
+      "loss": 0.9349,
+      "step": 12735
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.122938103849835e-05,
+      "loss": 0.8869,
+      "step": 12736
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.120686524611287e-05,
+      "loss": 0.8614,
+      "step": 12737
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.118435688653516e-05,
+      "loss": 0.8022,
+      "step": 12738
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.116185596075638e-05,
+      "loss": 0.8842,
+      "step": 12739
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.113936246976743e-05,
+      "loss": 0.9124,
+      "step": 12740
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.1116876414558695e-05,
+      "loss": 0.8276,
+      "step": 12741
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.10943977961204e-05,
+      "loss": 0.8769,
+      "step": 12742
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.107192661544245e-05,
+      "loss": 0.8996,
+      "step": 12743
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.104946287351427e-05,
+      "loss": 0.857,
+      "step": 12744
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.102700657132507e-05,
+      "loss": 0.8322,
+      "step": 12745
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.100455770986375e-05,
+      "loss": 0.9055,
+      "step": 12746
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.098211629011885e-05,
+      "loss": 0.859,
+      "step": 12747
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.095968231307853e-05,
+      "loss": 0.8078,
+      "step": 12748
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.09372557797307e-05,
+      "loss": 0.8549,
+      "step": 12749
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.0914836691062946e-05,
+      "loss": 0.893,
+      "step": 12750
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.0892425048062355e-05,
+      "loss": 0.8418,
+      "step": 12751
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.087002085171604e-05,
+      "loss": 0.8208,
+      "step": 12752
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.0847624103010385e-05,
+      "loss": 0.855,
+      "step": 12753
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.082523480293171e-05,
+      "loss": 0.8695,
+      "step": 12754
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.080285295246599e-05,
+      "loss": 0.8714,
+      "step": 12755
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.078047855259867e-05,
+      "loss": 0.8023,
+      "step": 12756
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.0758111604315076e-05,
+      "loss": 0.8375,
+      "step": 12757
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.073575210860016e-05,
+      "loss": 0.874,
+      "step": 12758
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.071340006643852e-05,
+      "loss": 0.865,
+      "step": 12759
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.069105547881437e-05,
+      "loss": 0.8705,
+      "step": 12760
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.06687183467117e-05,
+      "loss": 0.8638,
+      "step": 12761
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.064638867111416e-05,
+      "loss": 0.8674,
+      "step": 12762
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.062406645300491e-05,
+      "loss": 0.8657,
+      "step": 12763
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.060175169336708e-05,
+      "loss": 0.9095,
+      "step": 12764
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.057944439318319e-05,
+      "loss": 0.8451,
+      "step": 12765
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.0557144553435545e-05,
+      "loss": 0.8346,
+      "step": 12766
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.0534852175106186e-05,
+      "loss": 0.9637,
+      "step": 12767
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.0512567259176682e-05,
+      "loss": 0.8832,
+      "step": 12768
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.049028980662838e-05,
+      "loss": 0.807,
+      "step": 12769
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.0468019818442274e-05,
+      "loss": 0.8675,
+      "step": 12770
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.0445757295599064e-05,
+      "loss": 0.8234,
+      "step": 12771
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.0423502239078973e-05,
+      "loss": 0.8201,
+      "step": 12772
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.0401254649862075e-05,
+      "loss": 0.8756,
+      "step": 12773
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.0379014528928084e-05,
+      "loss": 0.86,
+      "step": 12774
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.0356781877256257e-05,
+      "loss": 0.9014,
+      "step": 12775
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.0334556695825632e-05,
+      "loss": 0.8535,
+      "step": 12776
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.0312338985614895e-05,
+      "loss": 0.9056,
+      "step": 12777
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.0290128747602487e-05,
+      "loss": 0.8945,
+      "step": 12778
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.0267925982766308e-05,
+      "loss": 0.8823,
+      "step": 12779
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.024573069208412e-05,
+      "loss": 0.9016,
+      "step": 12780
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.0223542876533327e-05,
+      "loss": 0.9069,
+      "step": 12781
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.020136253709085e-05,
+      "loss": 0.8396,
+      "step": 12782
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.0179189674733565e-05,
+      "loss": 0.8427,
+      "step": 12783
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.015702429043772e-05,
+      "loss": 0.9147,
+      "step": 12784
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.0134866385179416e-05,
+      "loss": 0.8971,
+      "step": 12785
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.0112715959934435e-05,
+      "loss": 0.8874,
+      "step": 12786
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.0090573015678057e-05,
+      "loss": 0.9003,
+      "step": 12787
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.006843755338542e-05,
+      "loss": 0.887,
+      "step": 12788
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.004630957403123e-05,
+      "loss": 0.8947,
+      "step": 12789
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.002418907858997e-05,
+      "loss": 0.8559,
+      "step": 12790
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.00020760680356e-05,
+      "loss": 0.9716,
+      "step": 12791
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.9979970543341917e-05,
+      "loss": 0.8695,
+      "step": 12792
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.9957872505482388e-05,
+      "loss": 0.8994,
+      "step": 12793
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.9935781955429943e-05,
+      "loss": 0.8493,
+      "step": 12794
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.991369889415756e-05,
+      "loss": 0.8692,
+      "step": 12795
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.989162332263751e-05,
+      "loss": 0.8628,
+      "step": 12796
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.9869555241841984e-05,
+      "loss": 0.8656,
+      "step": 12797
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.984749465274266e-05,
+      "loss": 0.8832,
+      "step": 12798
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.982544155631102e-05,
+      "loss": 0.8804,
+      "step": 12799
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.980339595351822e-05,
+      "loss": 0.8768,
+      "step": 12800
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.9781357845334912e-05,
+      "loss": 0.8967,
+      "step": 12801
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.9759327232731693e-05,
+      "loss": 0.9307,
+      "step": 12802
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.9737304116678567e-05,
+      "loss": 0.8276,
+      "step": 12803
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.9715288498145376e-05,
+      "loss": 0.8799,
+      "step": 12804
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.9693280378101617e-05,
+      "loss": 0.8291,
+      "step": 12805
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.9671279757516314e-05,
+      "loss": 0.83,
+      "step": 12806
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.9649286637358333e-05,
+      "loss": 0.834,
+      "step": 12807
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.962730101859612e-05,
+      "loss": 0.8633,
+      "step": 12808
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.960532290219786e-05,
+      "loss": 0.8566,
+      "step": 12809
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.9583352289131296e-05,
+      "loss": 0.7843,
+      "step": 12810
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.956138918036393e-05,
+      "loss": 0.8668,
+      "step": 12811
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.953943357686293e-05,
+      "loss": 0.8941,
+      "step": 12812
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.9517485479595007e-05,
+      "loss": 0.8992,
+      "step": 12813
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.9495544889526818e-05,
+      "loss": 0.8098,
+      "step": 12814
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.947361180762438e-05,
+      "loss": 0.8757,
+      "step": 12815
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.945168623485355e-05,
+      "loss": 0.8503,
+      "step": 12816
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.9429768172179907e-05,
+      "loss": 0.9443,
+      "step": 12817
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.940785762056846e-05,
+      "loss": 0.908,
+      "step": 12818
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.9385954580984144e-05,
+      "loss": 0.799,
+      "step": 12819
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.9364059054391435e-05,
+      "loss": 0.9823,
+      "step": 12820
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.934217104175454e-05,
+      "loss": 0.9169,
+      "step": 12821
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.932029054403722e-05,
+      "loss": 0.8823,
+      "step": 12822
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.9298417562203018e-05,
+      "loss": 0.8749,
+      "step": 12823
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.927655209721516e-05,
+      "loss": 0.8966,
+      "step": 12824
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.925469415003641e-05,
+      "loss": 0.8552,
+      "step": 12825
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.923284372162931e-05,
+      "loss": 0.8747,
+      "step": 12826
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.921100081295607e-05,
+      "loss": 0.8298,
+      "step": 12827
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.918916542497858e-05,
+      "loss": 0.8621,
+      "step": 12828
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.916733755865826e-05,
+      "loss": 0.8417,
+      "step": 12829
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.9145517214956353e-05,
+      "loss": 0.877,
+      "step": 12830
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.912370439483374e-05,
+      "loss": 0.8587,
+      "step": 12831
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.9101899099250852e-05,
+      "loss": 0.8148,
+      "step": 12832
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.908010132916805e-05,
+      "loss": 0.8735,
+      "step": 12833
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.905831108554504e-05,
+      "loss": 0.8921,
+      "step": 12834
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.9036528369341455e-05,
+      "loss": 0.8802,
+      "step": 12835
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.9014753181516498e-05,
+      "loss": 0.8301,
+      "step": 12836
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.8992985523028958e-05,
+      "loss": 0.8199,
+      "step": 12837
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.8971225394837408e-05,
+      "loss": 0.9091,
+      "step": 12838
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.89494727979001e-05,
+      "loss": 0.8679,
+      "step": 12839
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.8927727733174915e-05,
+      "loss": 0.8423,
+      "step": 12840
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.8905990201619303e-05,
+      "loss": 0.8269,
+      "step": 12841
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.888426020419055e-05,
+      "loss": 0.9106,
+      "step": 12842
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.886253774184555e-05,
+      "loss": 0.891,
+      "step": 12843
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.8840822815540747e-05,
+      "loss": 0.8341,
+      "step": 12844
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.8819115426232523e-05,
+      "loss": 0.8702,
+      "step": 12845
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.879741557487665e-05,
+      "loss": 0.9202,
+      "step": 12846
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.877572326242868e-05,
+      "loss": 0.8918,
+      "step": 12847
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.8754038489843927e-05,
+      "loss": 0.8315,
+      "step": 12848
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.8732361258077145e-05,
+      "loss": 0.8788,
+      "step": 12849
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.871069156808297e-05,
+      "loss": 0.873,
+      "step": 12850
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.8689029420815616e-05,
+      "loss": 0.8492,
+      "step": 12851
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.866737481722903e-05,
+      "loss": 0.9035,
+      "step": 12852
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.8645727758276674e-05,
+      "loss": 0.823,
+      "step": 12853
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.8624088244911828e-05,
+      "loss": 0.8451,
+      "step": 12854
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.860245627808742e-05,
+      "loss": 0.8698,
+      "step": 12855
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.858083185875593e-05,
+      "loss": 0.8472,
+      "step": 12856
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.855921498786964e-05,
+      "loss": 0.9096,
+      "step": 12857
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.853760566638044e-05,
+      "loss": 0.9311,
+      "step": 12858
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.851600389523996e-05,
+      "loss": 0.8707,
+      "step": 12859
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.8494409675399312e-05,
+      "loss": 0.9179,
+      "step": 12860
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.847282300780949e-05,
+      "loss": 0.8692,
+      "step": 12861
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.8451243893421063e-05,
+      "loss": 0.8596,
+      "step": 12862
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.8429672333184143e-05,
+      "loss": 0.8779,
+      "step": 12863
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.8408108328048832e-05,
+      "loss": 0.9377,
+      "step": 12864
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.838655187896455e-05,
+      "loss": 0.8507,
+      "step": 12865
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.8365002986880583e-05,
+      "loss": 0.775,
+      "step": 12866
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.8343461652745907e-05,
+      "loss": 0.8993,
+      "step": 12867
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.8321927877508957e-05,
+      "loss": 0.8056,
+      "step": 12868
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.830040166211805e-05,
+      "loss": 0.8335,
+      "step": 12869
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.8278883007521085e-05,
+      "loss": 0.8434,
+      "step": 12870
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.8257371914665688e-05,
+      "loss": 0.9007,
+      "step": 12871
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.823586838449901e-05,
+      "loss": 0.8376,
+      "step": 12872
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.8214372417967983e-05,
+      "loss": 0.8561,
+      "step": 12873
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.819288401601925e-05,
+      "loss": 0.8664,
+      "step": 12874
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.817140317959892e-05,
+      "loss": 0.8584,
+      "step": 12875
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.8149929909653062e-05,
+      "loss": 0.8574,
+      "step": 12876
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.8128464207127137e-05,
+      "loss": 0.8936,
+      "step": 12877
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.8107006072966457e-05,
+      "loss": 0.9225,
+      "step": 12878
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.808555550811587e-05,
+      "loss": 0.8723,
+      "step": 12879
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.806411251351997e-05,
+      "loss": 0.839,
+      "step": 12880
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.8042677090123047e-05,
+      "loss": 0.9489,
+      "step": 12881
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.8021249238868908e-05,
+      "loss": 0.8875,
+      "step": 12882
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.7999828960701258e-05,
+      "loss": 0.8028,
+      "step": 12883
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.7978416256563234e-05,
+      "loss": 0.881,
+      "step": 12884
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.795701112739779e-05,
+      "loss": 0.8725,
+      "step": 12885
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.7935613574147534e-05,
+      "loss": 0.8957,
+      "step": 12886
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.791422359775464e-05,
+      "loss": 0.8554,
+      "step": 12887
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.7892841199161024e-05,
+      "loss": 0.9405,
+      "step": 12888
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.7871466379308285e-05,
+      "loss": 0.9061,
+      "step": 12889
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.7850099139137718e-05,
+      "loss": 0.8382,
+      "step": 12890
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.7828739479590126e-05,
+      "loss": 0.8564,
+      "step": 12891
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.780738740160611e-05,
+      "loss": 0.8502,
+      "step": 12892
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.7786042906125987e-05,
+      "loss": 0.9026,
+      "step": 12893
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.7764705994089514e-05,
+      "loss": 0.7762,
+      "step": 12894
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.7743376666436426e-05,
+      "loss": 0.8225,
+      "step": 12895
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.7722054924105845e-05,
+      "loss": 0.7697,
+      "step": 12896
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.7700740768036727e-05,
+      "loss": 0.851,
+      "step": 12897
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.7679434199167652e-05,
+      "loss": 0.8696,
+      "step": 12898
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.7658135218436808e-05,
+      "loss": 0.8281,
+      "step": 12899
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.763684382678211e-05,
+      "loss": 0.9112,
+      "step": 12900
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.7615560025141118e-05,
+      "loss": 0.8408,
+      "step": 12901
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.7594283814451128e-05,
+      "loss": 0.8605,
+      "step": 12902
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.7573015195648967e-05,
+      "loss": 0.8551,
+      "step": 12903
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.7551754169671196e-05,
+      "loss": 0.8785,
+      "step": 12904
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.7530500737454135e-05,
+      "loss": 0.8644,
+      "step": 12905
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.7509254899933545e-05,
+      "loss": 0.8981,
+      "step": 12906
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.748801665804508e-05,
+      "loss": 0.8794,
+      "step": 12907
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.7466786012723944e-05,
+      "loss": 0.9363,
+      "step": 12908
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.744556296490506e-05,
+      "loss": 0.8969,
+      "step": 12909
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.7424347515522896e-05,
+      "loss": 0.9045,
+      "step": 12910
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.7403139665511756e-05,
+      "loss": 0.8971,
+      "step": 12911
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.7381939415805536e-05,
+      "loss": 0.8534,
+      "step": 12912
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.7360746767337663e-05,
+      "loss": 0.9099,
+      "step": 12913
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.7339561721041528e-05,
+      "loss": 0.8234,
+      "step": 12914
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.731838427784992e-05,
+      "loss": 0.932,
+      "step": 12915
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.7297214438695374e-05,
+      "loss": 0.8229,
+      "step": 12916
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.7276052204510193e-05,
+      "loss": 0.8653,
+      "step": 12917
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.7254897576226147e-05,
+      "loss": 0.861,
+      "step": 12918
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.723375055477484e-05,
+      "loss": 0.8439,
+      "step": 12919
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.7212611141087464e-05,
+      "loss": 0.8764,
+      "step": 12920
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.719147933609494e-05,
+      "loss": 0.8814,
+      "step": 12921
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.717035514072772e-05,
+      "loss": 0.8694,
+      "step": 12922
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.7149238555916066e-05,
+      "loss": 0.8638,
+      "step": 12923
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.712812958258988e-05,
+      "loss": 0.9075,
+      "step": 12924
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.7107028221678564e-05,
+      "loss": 0.8401,
+      "step": 12925
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.7085934474111497e-05,
+      "loss": 0.9196,
+      "step": 12926
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.7064848340817418e-05,
+      "loss": 0.8568,
+      "step": 12927
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.7043769822724872e-05,
+      "loss": 0.8676,
+      "step": 12928
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.7022698920762123e-05,
+      "loss": 0.8624,
+      "step": 12929
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.7001635635856915e-05,
+      "loss": 0.8108,
+      "step": 12930
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6980579968936882e-05,
+      "loss": 0.8972,
+      "step": 12931
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.695953192092906e-05,
+      "loss": 0.8524,
+      "step": 12932
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.693849149276051e-05,
+      "loss": 0.843,
+      "step": 12933
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6917458685357556e-05,
+      "loss": 0.8442,
+      "step": 12934
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.689643349964648e-05,
+      "loss": 0.8122,
+      "step": 12935
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.687541593655314e-05,
+      "loss": 0.848,
+      "step": 12936
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6854405997002975e-05,
+      "loss": 0.8886,
+      "step": 12937
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6833403681921175e-05,
+      "loss": 0.9014,
+      "step": 12938
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6812408992232585e-05,
+      "loss": 0.8314,
+      "step": 12939
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.679142192886179e-05,
+      "loss": 0.8639,
+      "step": 12940
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.6770442492732816e-05,
+      "loss": 0.8149,
+      "step": 12941
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.6749470684769563e-05,
+      "loss": 0.8384,
+      "step": 12942
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.6728506505895557e-05,
+      "loss": 0.8944,
+      "step": 12943
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.6707549957033818e-05,
+      "loss": 0.8565,
+      "step": 12944
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.668660103910736e-05,
+      "loss": 0.9317,
+      "step": 12945
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.666565975303854e-05,
+      "loss": 0.8988,
+      "step": 12946
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.664472609974955e-05,
+      "loss": 0.8922,
+      "step": 12947
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.662380008016223e-05,
+      "loss": 0.8905,
+      "step": 12948
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.6602881695198002e-05,
+      "loss": 0.9118,
+      "step": 12949
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.658197094577801e-05,
+      "loss": 0.8434,
+      "step": 12950
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.6561067832823084e-05,
+      "loss": 0.8804,
+      "step": 12951
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.6540172357253724e-05,
+      "loss": 0.8706,
+      "step": 12952
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.6519284519989994e-05,
+      "loss": 0.9247,
+      "step": 12953
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.649840432195172e-05,
+      "loss": 0.8483,
+      "step": 12954
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.6477531764058406e-05,
+      "loss": 0.9226,
+      "step": 12955
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.645666684722905e-05,
+      "loss": 0.8384,
+      "step": 12956
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.643580957238261e-05,
+      "loss": 0.8618,
+      "step": 12957
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.6414959940437413e-05,
+      "loss": 0.9317,
+      "step": 12958
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.6394117952311636e-05,
+      "loss": 0.884,
+      "step": 12959
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.6373283608922994e-05,
+      "loss": 0.8866,
+      "step": 12960
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.6352456911188973e-05,
+      "loss": 0.8455,
+      "step": 12961
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.6331637860026702e-05,
+      "loss": 0.8205,
+      "step": 12962
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.631082645635281e-05,
+      "loss": 0.8875,
+      "step": 12963
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.629002270108394e-05,
+      "loss": 0.874,
+      "step": 12964
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.626922659513602e-05,
+      "loss": 0.9409,
+      "step": 12965
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.624843813942488e-05,
+      "loss": 0.8747,
+      "step": 12966
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.622765733486594e-05,
+      "loss": 0.8454,
+      "step": 12967
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.6206884182374248e-05,
+      "loss": 0.8518,
+      "step": 12968
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.6186118682864558e-05,
+      "loss": 0.8667,
+      "step": 12969
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.6165360837251274e-05,
+      "loss": 0.8982,
+      "step": 12970
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.6144610646448553e-05,
+      "loss": 0.9257,
+      "step": 12971
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.6123868111369997e-05,
+      "loss": 0.8583,
+      "step": 12972
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.6103133232929078e-05,
+      "loss": 0.8817,
+      "step": 12973
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.6082406012038885e-05,
+      "loss": 0.8872,
+      "step": 12974
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.6061686449612e-05,
+      "loss": 0.8648,
+      "step": 12975
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.6040974546561002e-05,
+      "loss": 0.8582,
+      "step": 12976
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.6020270303797833e-05,
+      "loss": 0.9088,
+      "step": 12977
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5999573722234183e-05,
+      "loss": 0.8731,
+      "step": 12978
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5978884802781543e-05,
+      "loss": 0.8299,
+      "step": 12979
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.595820354635079e-05,
+      "loss": 0.8873,
+      "step": 12980
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5937529953852722e-05,
+      "loss": 0.9055,
+      "step": 12981
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5916864026197686e-05,
+      "loss": 0.891,
+      "step": 12982
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5896205764295743e-05,
+      "loss": 0.8412,
+      "step": 12983
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5875555169056508e-05,
+      "loss": 0.8553,
+      "step": 12984
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5854912241389338e-05,
+      "loss": 0.7973,
+      "step": 12985
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.583427698220333e-05,
+      "loss": 0.8675,
+      "step": 12986
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5813649392407048e-05,
+      "loss": 0.8436,
+      "step": 12987
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.579302947290887e-05,
+      "loss": 0.8508,
+      "step": 12988
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5772417224616807e-05,
+      "loss": 0.8413,
+      "step": 12989
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5751812648438557e-05,
+      "loss": 0.8322,
+      "step": 12990
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5731215745281366e-05,
+      "loss": 0.8589,
+      "step": 12991
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.571062651605225e-05,
+      "loss": 0.8156,
+      "step": 12992
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5690044961657877e-05,
+      "loss": 0.8152,
+      "step": 12993
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.566947108300448e-05,
+      "loss": 0.8724,
+      "step": 12994
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.564890488099816e-05,
+      "loss": 0.8612,
+      "step": 12995
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.562834635654443e-05,
+      "loss": 0.9317,
+      "step": 12996
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5607795510548637e-05,
+      "loss": 0.8713,
+      "step": 12997
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.558725234391577e-05,
+      "loss": 0.8846,
+      "step": 12998
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5566716857550387e-05,
+      "loss": 0.8729,
+      "step": 12999
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5546189052356773e-05,
+      "loss": 0.8306,
+      "step": 13000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.552566892923889e-05,
+      "loss": 0.8577,
+      "step": 13001
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.550515648910039e-05,
+      "loss": 0.869,
+      "step": 13002
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5484651732844445e-05,
+      "loss": 0.9337,
+      "step": 13003
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.546415466137404e-05,
+      "loss": 0.8643,
+      "step": 13004
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5443665275591767e-05,
+      "loss": 0.9514,
+      "step": 13005
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.542318357639979e-05,
+      "loss": 0.8677,
+      "step": 13006
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5402709564700167e-05,
+      "loss": 0.8327,
+      "step": 13007
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5382243241394378e-05,
+      "loss": 0.8412,
+      "step": 13008
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5361784607383655e-05,
+      "loss": 0.8186,
+      "step": 13009
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5341333663568946e-05,
+      "loss": 0.8807,
+      "step": 13010
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5320890410850752e-05,
+      "loss": 0.8491,
+      "step": 13011
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.530045485012935e-05,
+      "loss": 0.8918,
+      "step": 13012
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5280026982304495e-05,
+      "loss": 0.8637,
+      "step": 13013
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5259606808275926e-05,
+      "loss": 0.8279,
+      "step": 13014
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5239194328942684e-05,
+      "loss": 0.8847,
+      "step": 13015
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5218789545203693e-05,
+      "loss": 0.9362,
+      "step": 13016
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.519839245795752e-05,
+      "loss": 0.8487,
+      "step": 13017
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5178003068102252e-05,
+      "loss": 0.8328,
+      "step": 13018
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.515762137653579e-05,
+      "loss": 0.9278,
+      "step": 13019
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5137247384155638e-05,
+      "loss": 0.9101,
+      "step": 13020
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5116881091859014e-05,
+      "loss": 0.9128,
+      "step": 13021
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5096522500542664e-05,
+      "loss": 0.835,
+      "step": 13022
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5076171611103093e-05,
+      "loss": 0.8148,
+      "step": 13023
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5055828424436546e-05,
+      "loss": 0.8067,
+      "step": 13024
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5035492941438675e-05,
+      "loss": 0.7682,
+      "step": 13025
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5015165163005126e-05,
+      "loss": 0.8379,
+      "step": 13026
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.4994845090030895e-05,
+      "loss": 0.9022,
+      "step": 13027
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.497453272341086e-05,
+      "loss": 0.8906,
+      "step": 13028
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.4954228064039464e-05,
+      "loss": 0.8293,
+      "step": 13029
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.4933931112810772e-05,
+      "loss": 0.787,
+      "step": 13030
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.4913641870618597e-05,
+      "loss": 0.8854,
+      "step": 13031
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.4893360338356386e-05,
+      "loss": 0.8659,
+      "step": 13032
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.4873086516917264e-05,
+      "loss": 0.877,
+      "step": 13033
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.48528204071939e-05,
+      "loss": 0.8931,
+      "step": 13034
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.4832562010078773e-05,
+      "loss": 0.8494,
+      "step": 13035
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.4812311326463977e-05,
+      "loss": 0.9338,
+      "step": 13036
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.4792068357241148e-05,
+      "loss": 0.9107,
+      "step": 13037
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.4771833103301867e-05,
+      "loss": 0.8689,
+      "step": 13038
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.4751605565537016e-05,
+      "loss": 0.8811,
+      "step": 13039
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.4731385744837464e-05,
+      "loss": 0.8312,
+      "step": 13040
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.4711173642093455e-05,
+      "loss": 0.8274,
+      "step": 13041
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.4690969258195073e-05,
+      "loss": 0.863,
+      "step": 13042
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.4670772594032098e-05,
+      "loss": 0.8015,
+      "step": 13043
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4650583650493752e-05,
+      "loss": 0.9583,
+      "step": 13044
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4630402428469214e-05,
+      "loss": 0.8947,
+      "step": 13045
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4610228928847013e-05,
+      "loss": 0.941,
+      "step": 13046
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4590063152515576e-05,
+      "loss": 0.8307,
+      "step": 13047
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4569905100362922e-05,
+      "loss": 0.8474,
+      "step": 13048
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4549754773276656e-05,
+      "loss": 0.8241,
+      "step": 13049
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.452961217214409e-05,
+      "loss": 0.8738,
+      "step": 13050
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4509477297852224e-05,
+      "loss": 0.911,
+      "step": 13051
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.448935015128777e-05,
+      "loss": 0.8634,
+      "step": 13052
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.446923073333689e-05,
+      "loss": 0.9068,
+      "step": 13053
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4449119044885626e-05,
+      "loss": 0.8631,
+      "step": 13054
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4429015086819627e-05,
+      "loss": 0.8406,
+      "step": 13055
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4408918860024056e-05,
+      "loss": 0.8624,
+      "step": 13056
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4388830365383997e-05,
+      "loss": 0.8631,
+      "step": 13057
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4368749603783926e-05,
+      "loss": 0.9189,
+      "step": 13058
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.434867657610813e-05,
+      "loss": 0.8628,
+      "step": 13059
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.43286112832406e-05,
+      "loss": 0.9046,
+      "step": 13060
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.430855372606482e-05,
+      "loss": 0.7753,
+      "step": 13061
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4288503905464042e-05,
+      "loss": 0.9123,
+      "step": 13062
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4268461822321188e-05,
+      "loss": 0.8574,
+      "step": 13063
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4248427477518832e-05,
+      "loss": 0.851,
+      "step": 13064
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4228400871939094e-05,
+      "loss": 0.8446,
+      "step": 13065
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4208382006463936e-05,
+      "loss": 0.851,
+      "step": 13066
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4188370881974874e-05,
+      "loss": 0.8385,
+      "step": 13067
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.416836749935305e-05,
+      "loss": 0.8171,
+      "step": 13068
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4148371859479356e-05,
+      "loss": 0.832,
+      "step": 13069
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4128383963234268e-05,
+      "loss": 0.8511,
+      "step": 13070
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4108403811498037e-05,
+      "loss": 0.8368,
+      "step": 13071
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.408843140515038e-05,
+      "loss": 0.8412,
+      "step": 13072
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4068466745070816e-05,
+      "loss": 0.8105,
+      "step": 13073
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4048509832138554e-05,
+      "loss": 0.8664,
+      "step": 13074
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.402856066723227e-05,
+      "loss": 0.91,
+      "step": 13075
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.400861925123057e-05,
+      "loss": 0.8669,
+      "step": 13076
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.3988685585011482e-05,
+      "loss": 0.8331,
+      "step": 13077
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.39687596694528e-05,
+      "loss": 0.8748,
+      "step": 13078
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.3948841505431996e-05,
+      "loss": 0.8736,
+      "step": 13079
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.3928931093826123e-05,
+      "loss": 0.8539,
+      "step": 13080
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.3909028435511928e-05,
+      "loss": 0.871,
+      "step": 13081
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.388913353136588e-05,
+      "loss": 0.8169,
+      "step": 13082
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.3869246382264042e-05,
+      "loss": 0.8433,
+      "step": 13083
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.384936698908209e-05,
+      "loss": 0.8875,
+      "step": 13084
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.382949535269545e-05,
+      "loss": 0.8999,
+      "step": 13085
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.380963147397921e-05,
+      "loss": 0.8301,
+      "step": 13086
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.378977535380795e-05,
+      "loss": 0.912,
+      "step": 13087
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.37699269930562e-05,
+      "loss": 0.7914,
+      "step": 13088
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.375008639259788e-05,
+      "loss": 0.8998,
+      "step": 13089
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.3730253553306668e-05,
+      "loss": 0.7971,
+      "step": 13090
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.3710428476055978e-05,
+      "loss": 0.9522,
+      "step": 13091
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.369061116171871e-05,
+      "loss": 0.8932,
+      "step": 13092
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.3670801611167638e-05,
+      "loss": 0.8459,
+      "step": 13093
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.36509998252749e-05,
+      "loss": 0.8211,
+      "step": 13094
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.3631205804912647e-05,
+      "loss": 0.8156,
+      "step": 13095
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.361141955095241e-05,
+      "loss": 0.9309,
+      "step": 13096
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.359164106426548e-05,
+      "loss": 0.8486,
+      "step": 13097
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.357187034572288e-05,
+      "loss": 0.8442,
+      "step": 13098
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.3552107396195135e-05,
+      "loss": 0.8537,
+      "step": 13099
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.35323522165525e-05,
+      "loss": 0.8771,
+      "step": 13100
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.351260480766495e-05,
+      "loss": 0.8843,
+      "step": 13101
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.3492865170402057e-05,
+      "loss": 0.802,
+      "step": 13102
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.3473133305633012e-05,
+      "loss": 0.8767,
+      "step": 13103
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.345340921422674e-05,
+      "loss": 0.801,
+      "step": 13104
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.3433692897051795e-05,
+      "loss": 0.8351,
+      "step": 13105
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.3413984354976325e-05,
+      "loss": 0.8308,
+      "step": 13106
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.3394283588868304e-05,
+      "loss": 0.8174,
+      "step": 13107
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.337459059959517e-05,
+      "loss": 0.9063,
+      "step": 13108
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.335490538802412e-05,
+      "loss": 0.8646,
+      "step": 13109
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.3335227955022056e-05,
+      "loss": 0.8152,
+      "step": 13110
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.3315558301455376e-05,
+      "loss": 0.8087,
+      "step": 13111
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.3295896428190277e-05,
+      "loss": 0.8644,
+      "step": 13112
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.3276242336092557e-05,
+      "loss": 0.8572,
+      "step": 13113
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.3256596026027767e-05,
+      "loss": 0.9048,
+      "step": 13114
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.32369574988609e-05,
+      "loss": 0.8918,
+      "step": 13115
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.3217326755456824e-05,
+      "loss": 0.8557,
+      "step": 13116
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.319770379667998e-05,
+      "loss": 0.9006,
+      "step": 13117
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.3178088623394367e-05,
+      "loss": 0.8076,
+      "step": 13118
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.3158481236463914e-05,
+      "loss": 0.8561,
+      "step": 13119
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.3138881636751886e-05,
+      "loss": 0.8426,
+      "step": 13120
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.311928982512144e-05,
+      "loss": 0.8755,
+      "step": 13121
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.309970580243521e-05,
+      "loss": 0.8587,
+      "step": 13122
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.3080129569555632e-05,
+      "loss": 0.9003,
+      "step": 13123
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.306056112734478e-05,
+      "loss": 0.8446,
+      "step": 13124
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.3041000476664242e-05,
+      "loss": 0.8774,
+      "step": 13125
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.3021447618375524e-05,
+      "loss": 0.8471,
+      "step": 13126
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.3001902553339516e-05,
+      "loss": 0.9074,
+      "step": 13127
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.298236528241693e-05,
+      "loss": 0.9628,
+      "step": 13128
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.2962835806468118e-05,
+      "loss": 0.8552,
+      "step": 13129
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.2943314126353e-05,
+      "loss": 0.8687,
+      "step": 13130
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.2923800242931237e-05,
+      "loss": 0.8807,
+      "step": 13131
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.290429415706212e-05,
+      "loss": 0.8286,
+      "step": 13132
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.288479586960466e-05,
+      "loss": 0.8273,
+      "step": 13133
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.2865305381417357e-05,
+      "loss": 0.8645,
+      "step": 13134
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.2845822693358552e-05,
+      "loss": 0.851,
+      "step": 13135
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.2826347806286164e-05,
+      "loss": 0.8812,
+      "step": 13136
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.2806880721057676e-05,
+      "loss": 0.9356,
+      "step": 13137
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.2787421438530477e-05,
+      "loss": 0.8997,
+      "step": 13138
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.276796995956132e-05,
+      "loss": 0.9203,
+      "step": 13139
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.274852628500681e-05,
+      "loss": 0.8739,
+      "step": 13140
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.2729090415723196e-05,
+      "loss": 0.8622,
+      "step": 13141
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.2709662352566243e-05,
+      "loss": 0.8334,
+      "step": 13142
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.2690242096391502e-05,
+      "loss": 0.9004,
+      "step": 13143
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.267082964805416e-05,
+      "loss": 0.871,
+      "step": 13144
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.2651425008409065e-05,
+      "loss": 0.871,
+      "step": 13145
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.2632028178310626e-05,
+      "loss": 0.8112,
+      "step": 13146
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.2612639158613046e-05,
+      "loss": 0.8921,
+      "step": 13147
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.2593257950170134e-05,
+      "loss": 0.8285,
+      "step": 13148
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.2573884553835244e-05,
+      "loss": 0.8794,
+      "step": 13149
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.255451897046159e-05,
+      "loss": 0.9051,
+      "step": 13150
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.2535161200901867e-05,
+      "loss": 0.8465,
+      "step": 13151
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.2515811246008565e-05,
+      "loss": 0.8939,
+      "step": 13152
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.24964691066337e-05,
+      "loss": 0.8506,
+      "step": 13153
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.2477134783629006e-05,
+      "loss": 0.8475,
+      "step": 13154
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.2457808277845915e-05,
+      "loss": 0.7969,
+      "step": 13155
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.243848959013537e-05,
+      "loss": 0.9167,
+      "step": 13156
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.2419178721348224e-05,
+      "loss": 0.8356,
+      "step": 13157
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.2399875672334703e-05,
+      "loss": 0.8314,
+      "step": 13158
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.2380580443944864e-05,
+      "loss": 0.9086,
+      "step": 13159
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.2361293037028408e-05,
+      "loss": 0.8288,
+      "step": 13160
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.2342013452434564e-05,
+      "loss": 0.8642,
+      "step": 13161
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.2322741691012383e-05,
+      "loss": 0.8777,
+      "step": 13162
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.230347775361046e-05,
+      "loss": 0.9121,
+      "step": 13163
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.228422164107715e-05,
+      "loss": 0.9252,
+      "step": 13164
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.22649733542603e-05,
+      "loss": 0.9017,
+      "step": 13165
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.2245732894007555e-05,
+      "loss": 0.8292,
+      "step": 13166
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.2226500261166216e-05,
+      "loss": 0.8198,
+      "step": 13167
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.2207275456583056e-05,
+      "loss": 0.896,
+      "step": 13168
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.218805848110481e-05,
+      "loss": 0.8484,
+      "step": 13169
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.2168849335577568e-05,
+      "loss": 0.8476,
+      "step": 13170
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.2149648020847246e-05,
+      "loss": 0.8462,
+      "step": 13171
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.2130454537759438e-05,
+      "loss": 0.8678,
+      "step": 13172
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.211126888715924e-05,
+      "loss": 0.8704,
+      "step": 13173
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.209209106989154e-05,
+      "loss": 0.8252,
+      "step": 13174
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.2072921086800746e-05,
+      "loss": 0.843,
+      "step": 13175
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.2053758938731163e-05,
+      "loss": 0.8768,
+      "step": 13176
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.2034604626526468e-05,
+      "loss": 0.8626,
+      "step": 13177
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.2015458151030187e-05,
+      "loss": 0.8364,
+      "step": 13178
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.1996319513085427e-05,
+      "loss": 0.8751,
+      "step": 13179
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.1977188713534914e-05,
+      "loss": 0.8836,
+      "step": 13180
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.1958065753221124e-05,
+      "loss": 0.8691,
+      "step": 13181
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.19389506329861e-05,
+      "loss": 0.8889,
+      "step": 13182
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.191984335367163e-05,
+      "loss": 0.8725,
+      "step": 13183
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.1900743916119048e-05,
+      "loss": 0.9186,
+      "step": 13184
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.188165232116941e-05,
+      "loss": 0.8543,
+      "step": 13185
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.1862568569663446e-05,
+      "loss": 0.8986,
+      "step": 13186
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.1843492662441412e-05,
+      "loss": 0.847,
+      "step": 13187
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.1824424600343485e-05,
+      "loss": 0.8561,
+      "step": 13188
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.1805364384209193e-05,
+      "loss": 0.8999,
+      "step": 13189
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.1786312014877864e-05,
+      "loss": 0.8686,
+      "step": 13190
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.176726749318856e-05,
+      "loss": 0.8703,
+      "step": 13191
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.1748230819979808e-05,
+      "loss": 0.805,
+      "step": 13192
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.17292019960899e-05,
+      "loss": 0.8431,
+      "step": 13193
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.1710181022356824e-05,
+      "loss": 0.874,
+      "step": 13194
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.169116789961816e-05,
+      "loss": 0.836,
+      "step": 13195
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.1672162628711078e-05,
+      "loss": 0.845,
+      "step": 13196
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.1653165210472558e-05,
+      "loss": 0.8697,
+      "step": 13197
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.163417564573915e-05,
+      "loss": 0.9102,
+      "step": 13198
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.1615193935346988e-05,
+      "loss": 0.8329,
+      "step": 13199
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.159622008013198e-05,
+      "loss": 0.8767,
+      "step": 13200
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.157725408092961e-05,
+      "loss": 0.8681,
+      "step": 13201
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.1558295938575145e-05,
+      "loss": 0.9053,
+      "step": 13202
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.1539345653903274e-05,
+      "loss": 0.9154,
+      "step": 13203
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.152040322774853e-05,
+      "loss": 0.8418,
+      "step": 13204
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.150146866094509e-05,
+      "loss": 0.8335,
+      "step": 13205
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.148254195432662e-05,
+      "loss": 0.9136,
+      "step": 13206
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.1463623108726693e-05,
+      "loss": 0.9116,
+      "step": 13207
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.1444712124978316e-05,
+      "loss": 0.8962,
+      "step": 13208
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.1425809003914243e-05,
+      "loss": 0.8796,
+      "step": 13209
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.1406913746366943e-05,
+      "loss": 0.9558,
+      "step": 13210
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.138802635316837e-05,
+      "loss": 0.8643,
+      "step": 13211
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.1369146825150255e-05,
+      "loss": 0.8222,
+      "step": 13212
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.1350275163143983e-05,
+      "loss": 0.8674,
+      "step": 13213
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.1331411367980603e-05,
+      "loss": 0.9016,
+      "step": 13214
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.1312555440490712e-05,
+      "loss": 0.8704,
+      "step": 13215
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.1293707381504647e-05,
+      "loss": 0.8248,
+      "step": 13216
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.1274867191852432e-05,
+      "loss": 0.8562,
+      "step": 13217
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.1256034872363584e-05,
+      "loss": 0.8415,
+      "step": 13218
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.123721042386755e-05,
+      "loss": 0.9154,
+      "step": 13219
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.121839384719311e-05,
+      "loss": 0.8851,
+      "step": 13220
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.1199585143168933e-05,
+      "loss": 0.8758,
+      "step": 13221
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.118078431262329e-05,
+      "loss": 0.8756,
+      "step": 13222
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.116199135638397e-05,
+      "loss": 0.8854,
+      "step": 13223
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.1143206275278594e-05,
+      "loss": 0.8723,
+      "step": 13224
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.1124429070134345e-05,
+      "loss": 0.8329,
+      "step": 13225
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.1105659741778117e-05,
+      "loss": 0.8532,
+      "step": 13226
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.1086898291036337e-05,
+      "loss": 0.8023,
+      "step": 13227
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.1068144718735218e-05,
+      "loss": 0.905,
+      "step": 13228
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.10493990257006e-05,
+      "loss": 0.765,
+      "step": 13229
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.1030661212757897e-05,
+      "loss": 0.8753,
+      "step": 13230
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.101193128073222e-05,
+      "loss": 0.8966,
+      "step": 13231
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.0993209230448384e-05,
+      "loss": 0.781,
+      "step": 13232
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.097449506273086e-05,
+      "loss": 0.8447,
+      "step": 13233
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.0955788778403608e-05,
+      "loss": 0.8771,
+      "step": 13234
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.0937090378290412e-05,
+      "loss": 0.9043,
+      "step": 13235
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.091839986321471e-05,
+      "loss": 0.8572,
+      "step": 13236
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.0899717233999417e-05,
+      "loss": 0.8912,
+      "step": 13237
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.0881042491467385e-05,
+      "loss": 0.8064,
+      "step": 13238
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.0862375636440824e-05,
+      "loss": 0.8981,
+      "step": 13239
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.0843716669741764e-05,
+      "loss": 0.8595,
+      "step": 13240
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.0825065592191907e-05,
+      "loss": 0.8574,
+      "step": 13241
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.080642240461248e-05,
+      "loss": 0.8311,
+      "step": 13242
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.0787787107824474e-05,
+      "loss": 0.8692,
+      "step": 13243
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.0769159702648455e-05,
+      "loss": 0.8632,
+      "step": 13244
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.0750540189904787e-05,
+      "loss": 0.8776,
+      "step": 13245
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.0731928570413193e-05,
+      "loss": 0.7811,
+      "step": 13246
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.0713324844993464e-05,
+      "loss": 0.8369,
+      "step": 13247
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.0694729014464676e-05,
+      "loss": 0.8565,
+      "step": 13248
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.067614107964566e-05,
+      "loss": 0.9005,
+      "step": 13249
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.0657561041355056e-05,
+      "loss": 0.8767,
+      "step": 13250
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.0638988900410938e-05,
+      "loss": 0.8988,
+      "step": 13251
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.062042465763119e-05,
+      "loss": 0.8339,
+      "step": 13252
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.0601868313833282e-05,
+      "loss": 0.8566,
+      "step": 13253
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.0583319869834307e-05,
+      "loss": 0.9425,
+      "step": 13254
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.0564779326451112e-05,
+      "loss": 0.8736,
+      "step": 13255
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.0546246684500002e-05,
+      "loss": 0.8228,
+      "step": 13256
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.0527721944797216e-05,
+      "loss": 0.8986,
+      "step": 13257
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.0509205108158393e-05,
+      "loss": 0.8943,
+      "step": 13258
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.0490696175398937e-05,
+      "loss": 0.843,
+      "step": 13259
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.047219514733396e-05,
+      "loss": 0.8416,
+      "step": 13260
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.0453702024778054e-05,
+      "loss": 0.8683,
+      "step": 13261
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.0435216808545587e-05,
+      "loss": 0.9166,
+      "step": 13262
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.04167394994506e-05,
+      "loss": 0.9156,
+      "step": 13263
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.0398270098306748e-05,
+      "loss": 0.8652,
+      "step": 13264
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.037980860592723e-05,
+      "loss": 0.8335,
+      "step": 13265
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.036135502312513e-05,
+      "loss": 0.8904,
+      "step": 13266
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.0342909350713012e-05,
+      "loss": 0.85,
+      "step": 13267
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.0324471589503014e-05,
+      "loss": 0.9153,
+      "step": 13268
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.0306041740307237e-05,
+      "loss": 0.8688,
+      "step": 13269
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.0287619803937074e-05,
+      "loss": 0.8543,
+      "step": 13270
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.0269205781203836e-05,
+      "loss": 0.8324,
+      "step": 13271
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.0250799672918362e-05,
+      "loss": 0.9058,
+      "step": 13272
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.023240147989114e-05,
+      "loss": 0.8952,
+      "step": 13273
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.0214011202932336e-05,
+      "loss": 0.8592,
+      "step": 13274
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.0195628842851776e-05,
+      "loss": 0.9216,
+      "step": 13275
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.0177254400458988e-05,
+      "loss": 0.8675,
+      "step": 13276
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.0158887876562947e-05,
+      "loss": 0.852,
+      "step": 13277
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.0140529271972587e-05,
+      "loss": 0.8195,
+      "step": 13278
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.0122178587496253e-05,
+      "loss": 0.8909,
+      "step": 13279
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.0103835823941973e-05,
+      "loss": 0.8388,
+      "step": 13280
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.008550098211752e-05,
+      "loss": 0.8609,
+      "step": 13281
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.0067174062830253e-05,
+      "loss": 0.8828,
+      "step": 13282
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.0048855066887272e-05,
+      "loss": 0.8689,
+      "step": 13283
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.0030543995095098e-05,
+      "loss": 0.808,
+      "step": 13284
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.0012240848260234e-05,
+      "loss": 0.8125,
+      "step": 13285
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.999394562718857e-05,
+      "loss": 0.8737,
+      "step": 13286
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9975658332685666e-05,
+      "loss": 0.9154,
+      "step": 13287
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.995737896555696e-05,
+      "loss": 0.8609,
+      "step": 13288
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.993910752660728e-05,
+      "loss": 0.8497,
+      "step": 13289
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9920844016641227e-05,
+      "loss": 0.8588,
+      "step": 13290
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9902588436463086e-05,
+      "loss": 0.9134,
+      "step": 13291
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9884340786876642e-05,
+      "loss": 0.8215,
+      "step": 13292
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9866101068685494e-05,
+      "loss": 0.8719,
+      "step": 13293
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9847869282692844e-05,
+      "loss": 0.8835,
+      "step": 13294
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9829645429701515e-05,
+      "loss": 0.8747,
+      "step": 13295
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9811429510513914e-05,
+      "loss": 0.8401,
+      "step": 13296
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.979322152593235e-05,
+      "loss": 0.8814,
+      "step": 13297
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9775021476758492e-05,
+      "loss": 0.8564,
+      "step": 13298
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.975682936379373e-05,
+      "loss": 0.8552,
+      "step": 13299
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.97386451878393e-05,
+      "loss": 0.791,
+      "step": 13300
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9720468949695815e-05,
+      "loss": 0.8253,
+      "step": 13301
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.970230065016372e-05,
+      "loss": 0.881,
+      "step": 13302
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9684140290043064e-05,
+      "loss": 0.885,
+      "step": 13303
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.966598787013356e-05,
+      "loss": 0.8511,
+      "step": 13304
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.964784339123449e-05,
+      "loss": 0.8756,
+      "step": 13305
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.962970685414487e-05,
+      "loss": 0.8948,
+      "step": 13306
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9611578259663378e-05,
+      "loss": 0.7679,
+      "step": 13307
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.959345760858824e-05,
+      "loss": 0.89,
+      "step": 13308
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.957534490171744e-05,
+      "loss": 0.8886,
+      "step": 13309
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9557240139848608e-05,
+      "loss": 0.7979,
+      "step": 13310
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9539143323778884e-05,
+      "loss": 0.8811,
+      "step": 13311
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9521054454305253e-05,
+      "loss": 0.922,
+      "step": 13312
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.950297353222421e-05,
+      "loss": 0.8221,
+      "step": 13313
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.948490055833201e-05,
+      "loss": 0.8481,
+      "step": 13314
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.946683553342439e-05,
+      "loss": 0.8387,
+      "step": 13315
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9448778458296957e-05,
+      "loss": 0.9342,
+      "step": 13316
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9430729333744835e-05,
+      "loss": 0.8618,
+      "step": 13317
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9412688160562697e-05,
+      "loss": 0.8608,
+      "step": 13318
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9394654939545154e-05,
+      "loss": 0.9242,
+      "step": 13319
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.937662967148619e-05,
+      "loss": 0.8594,
+      "step": 13320
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9358612357179575e-05,
+      "loss": 0.8558,
+      "step": 13321
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9340602997418688e-05,
+      "loss": 0.859,
+      "step": 13322
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.932260159299666e-05,
+      "loss": 0.942,
+      "step": 13323
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9304608144706048e-05,
+      "loss": 0.8352,
+      "step": 13324
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.928662265333927e-05,
+      "loss": 0.8449,
+      "step": 13325
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9268645119688354e-05,
+      "loss": 0.8625,
+      "step": 13326
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9250675544544804e-05,
+      "loss": 0.798,
+      "step": 13327
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9232713928700075e-05,
+      "loss": 0.8633,
+      "step": 13328
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9214760272945e-05,
+      "loss": 0.7751,
+      "step": 13329
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.919681457807021e-05,
+      "loss": 0.8324,
+      "step": 13330
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9178876844865944e-05,
+      "loss": 0.9156,
+      "step": 13331
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.916094707412206e-05,
+      "loss": 0.8471,
+      "step": 13332
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.914302526662817e-05,
+      "loss": 0.8475,
+      "step": 13333
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9125111423173304e-05,
+      "loss": 0.9086,
+      "step": 13334
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.910720554454648e-05,
+      "loss": 0.8511,
+      "step": 13335
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9089307631536113e-05,
+      "loss": 0.8288,
+      "step": 13336
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.907141768493026e-05,
+      "loss": 0.8122,
+      "step": 13337
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9053535705516844e-05,
+      "loss": 0.8082,
+      "step": 13338
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9035661694083173e-05,
+      "loss": 0.8067,
+      "step": 13339
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9017795651416413e-05,
+      "loss": 0.8857,
+      "step": 13340
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.899993757830325e-05,
+      "loss": 0.8292,
+      "step": 13341
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.8982087475530118e-05,
+      "loss": 0.9053,
+      "step": 13342
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.896424534388299e-05,
+      "loss": 0.8726,
+      "step": 13343
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.894641118414755e-05,
+      "loss": 0.8626,
+      "step": 13344
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.8928584997109188e-05,
+      "loss": 0.8523,
+      "step": 13345
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.8910766783552746e-05,
+      "loss": 0.8261,
+      "step": 13346
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.889295654426302e-05,
+      "loss": 0.8673,
+      "step": 13347
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.8875154280024155e-05,
+      "loss": 0.915,
+      "step": 13348
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.8857359991620128e-05,
+      "loss": 0.8355,
+      "step": 13349
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.8839573679834533e-05,
+      "loss": 0.8643,
+      "step": 13350
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.8821795345450522e-05,
+      "loss": 0.9086,
+      "step": 13351
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.8804024989251003e-05,
+      "loss": 0.8604,
+      "step": 13352
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.8786262612018502e-05,
+      "loss": 0.8165,
+      "step": 13353
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.8768508214535197e-05,
+      "loss": 0.8453,
+      "step": 13354
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.875076179758284e-05,
+      "loss": 0.8146,
+      "step": 13355
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.873302336194296e-05,
+      "loss": 0.919,
+      "step": 13356
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.8715292908396664e-05,
+      "loss": 0.8149,
+      "step": 13357
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.869757043772462e-05,
+      "loss": 0.8321,
+      "step": 13358
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.8679855950707402e-05,
+      "loss": 0.8935,
+      "step": 13359
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.866214944812492e-05,
+      "loss": 0.8643,
+      "step": 13360
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.8644450930756995e-05,
+      "loss": 0.9425,
+      "step": 13361
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.862676039938287e-05,
+      "loss": 0.8732,
+      "step": 13362
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.860907785478161e-05,
+      "loss": 0.8609,
+      "step": 13363
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.8591403297731903e-05,
+      "loss": 0.8893,
+      "step": 13364
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.8573736729011903e-05,
+      "loss": 0.9164,
+      "step": 13365
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.8556078149399746e-05,
+      "loss": 0.8114,
+      "step": 13366
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.8538427559672923e-05,
+      "loss": 0.8824,
+      "step": 13367
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.852078496060867e-05,
+      "loss": 0.7811,
+      "step": 13368
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.8503150352983934e-05,
+      "loss": 0.8741,
+      "step": 13369
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.8485523737575217e-05,
+      "loss": 0.8456,
+      "step": 13370
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.846790511515868e-05,
+      "loss": 0.8425,
+      "step": 13371
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.845029448651021e-05,
+      "loss": 0.8299,
+      "step": 13372
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.8432691852405328e-05,
+      "loss": 0.8559,
+      "step": 13373
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.8415097213619047e-05,
+      "loss": 0.8121,
+      "step": 13374
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.8397510570926246e-05,
+      "loss": 0.8447,
+      "step": 13375
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.8379931925101323e-05,
+      "loss": 0.818,
+      "step": 13376
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.8362361276918305e-05,
+      "loss": 0.8975,
+      "step": 13377
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.8344798627151016e-05,
+      "loss": 0.8832,
+      "step": 13378
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.832724397657275e-05,
+      "loss": 0.8768,
+      "step": 13379
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.8309697325956554e-05,
+      "loss": 0.8217,
+      "step": 13380
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.829215867607512e-05,
+      "loss": 0.8755,
+      "step": 13381
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.82746280277007e-05,
+      "loss": 0.8202,
+      "step": 13382
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.825710538160528e-05,
+      "loss": 0.9062,
+      "step": 13383
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.82395907385605e-05,
+      "loss": 0.8442,
+      "step": 13384
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.822208409933761e-05,
+      "loss": 0.8063,
+      "step": 13385
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.820458546470749e-05,
+      "loss": 0.8651,
+      "step": 13386
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.8187094835440676e-05,
+      "loss": 0.8278,
+      "step": 13387
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.8169612212307462e-05,
+      "loss": 0.8512,
+      "step": 13388
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.815213759607759e-05,
+      "loss": 0.857,
+      "step": 13389
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.8134670987520574e-05,
+      "loss": 0.8305,
+      "step": 13390
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.8117212387405602e-05,
+      "loss": 0.9066,
+      "step": 13391
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.8099761796501458e-05,
+      "loss": 0.9297,
+      "step": 13392
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.808231921557655e-05,
+      "loss": 0.8615,
+      "step": 13393
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.806488464539895e-05,
+      "loss": 0.8297,
+      "step": 13394
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.8047458086736446e-05,
+      "loss": 0.8412,
+      "step": 13395
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.8030039540356315e-05,
+      "loss": 0.887,
+      "step": 13396
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.801262900702574e-05,
+      "loss": 0.8866,
+      "step": 13397
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.799522648751124e-05,
+      "loss": 0.8597,
+      "step": 13398
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.797783198257923e-05,
+      "loss": 0.8643,
+      "step": 13399
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.796044549299567e-05,
+      "loss": 0.8797,
+      "step": 13400
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.7943067019526106e-05,
+      "loss": 0.861,
+      "step": 13401
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.7925696562935856e-05,
+      "loss": 0.8781,
+      "step": 13402
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.7908334123989824e-05,
+      "loss": 0.8366,
+      "step": 13403
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.7890979703452593e-05,
+      "loss": 0.8566,
+      "step": 13404
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7873633302088288e-05,
+      "loss": 0.8765,
+      "step": 13405
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.785629492066079e-05,
+      "loss": 0.8517,
+      "step": 13406
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.783896455993366e-05,
+      "loss": 0.9141,
+      "step": 13407
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7821642220669908e-05,
+      "loss": 0.8483,
+      "step": 13408
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7804327903632467e-05,
+      "loss": 0.944,
+      "step": 13409
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7787021609583678e-05,
+      "loss": 0.9109,
+      "step": 13410
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.776972333928566e-05,
+      "loss": 0.8281,
+      "step": 13411
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7752433093500165e-05,
+      "loss": 0.8309,
+      "step": 13412
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7735150872988516e-05,
+      "loss": 0.9348,
+      "step": 13413
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7717876678511768e-05,
+      "loss": 0.8722,
+      "step": 13414
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.770061051083054e-05,
+      "loss": 0.8455,
+      "step": 13415
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7683352370705243e-05,
+      "loss": 0.817,
+      "step": 13416
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7666102258895755e-05,
+      "loss": 0.8262,
+      "step": 13417
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7648860176161718e-05,
+      "loss": 0.8097,
+      "step": 13418
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7631626123262434e-05,
+      "loss": 0.9057,
+      "step": 13419
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7614400100956717e-05,
+      "loss": 0.8388,
+      "step": 13420
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.759718211000314e-05,
+      "loss": 0.8692,
+      "step": 13421
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.757997215115992e-05,
+      "loss": 0.8235,
+      "step": 13422
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7562770225184934e-05,
+      "loss": 0.8658,
+      "step": 13423
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7545576332835577e-05,
+      "loss": 0.8808,
+      "step": 13424
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7528390474869028e-05,
+      "loss": 0.9208,
+      "step": 13425
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.751121265204212e-05,
+      "loss": 0.8604,
+      "step": 13426
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.749404286511116e-05,
+      "loss": 0.8938,
+      "step": 13427
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7476881114832345e-05,
+      "loss": 0.8543,
+      "step": 13428
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.745972740196129e-05,
+      "loss": 0.8955,
+      "step": 13429
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7442581727253414e-05,
+      "loss": 0.9054,
+      "step": 13430
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.742544409146376e-05,
+      "loss": 0.8861,
+      "step": 13431
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.74083144953469e-05,
+      "loss": 0.8035,
+      "step": 13432
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7391192939657184e-05,
+      "loss": 0.8536,
+      "step": 13433
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.737407942514855e-05,
+      "loss": 0.9025,
+      "step": 13434
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7356973952574628e-05,
+      "loss": 0.8978,
+      "step": 13435
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.733987652268858e-05,
+      "loss": 0.8919,
+      "step": 13436
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7322787136243336e-05,
+      "loss": 0.8564,
+      "step": 13437
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.730570579399149e-05,
+      "loss": 0.8173,
+      "step": 13438
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7288632496685063e-05,
+      "loss": 0.8658,
+      "step": 13439
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7271567245076036e-05,
+      "loss": 0.8678,
+      "step": 13440
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7254510039915783e-05,
+      "loss": 0.8436,
+      "step": 13441
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7237460881955502e-05,
+      "loss": 0.8333,
+      "step": 13442
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.722041977194584e-05,
+      "loss": 0.8794,
+      "step": 13443
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7203386710637282e-05,
+      "loss": 0.882,
+      "step": 13444
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7186361698779873e-05,
+      "loss": 0.8353,
+      "step": 13445
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7169344737123238e-05,
+      "loss": 0.7914,
+      "step": 13446
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7152335826416844e-05,
+      "loss": 0.8281,
+      "step": 13447
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7135334967409578e-05,
+      "loss": 0.8296,
+      "step": 13448
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.711834216085011e-05,
+      "loss": 0.8274,
+      "step": 13449
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7101357407486752e-05,
+      "loss": 0.874,
+      "step": 13450
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7084380708067327e-05,
+      "loss": 0.8724,
+      "step": 13451
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7067412063339506e-05,
+      "loss": 0.8372,
+      "step": 13452
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7050451474050444e-05,
+      "loss": 0.9035,
+      "step": 13453
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7033498940947058e-05,
+      "loss": 0.8715,
+      "step": 13454
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.701655446477577e-05,
+      "loss": 0.8663,
+      "step": 13455
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6999618046282807e-05,
+      "loss": 0.9275,
+      "step": 13456
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.698268968621395e-05,
+      "loss": 0.9238,
+      "step": 13457
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6965769385314557e-05,
+      "loss": 0.8426,
+      "step": 13458
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6948857144329832e-05,
+      "loss": 0.8507,
+      "step": 13459
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.693195296400445e-05,
+      "loss": 0.8308,
+      "step": 13460
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6915056845082765e-05,
+      "loss": 0.8994,
+      "step": 13461
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6898168788308877e-05,
+      "loss": 0.8481,
+      "step": 13462
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6881288794426364e-05,
+      "loss": 0.8768,
+      "step": 13463
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6864416864178566e-05,
+      "loss": 0.831,
+      "step": 13464
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6847552998308447e-05,
+      "loss": 0.8308,
+      "step": 13465
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.683069719755863e-05,
+      "loss": 0.8192,
+      "step": 13466
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6813849462671306e-05,
+      "loss": 0.9053,
+      "step": 13467
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6797009794388406e-05,
+      "loss": 0.8757,
+      "step": 13468
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.678017819345148e-05,
+      "loss": 0.8351,
+      "step": 13469
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6763354660601636e-05,
+      "loss": 0.8976,
+      "step": 13470
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6746539196579735e-05,
+      "loss": 0.8288,
+      "step": 13471
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.672973180212629e-05,
+      "loss": 0.8169,
+      "step": 13472
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.671293247798138e-05,
+      "loss": 0.9356,
+      "step": 13473
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6696141224884743e-05,
+      "loss": 0.8576,
+      "step": 13474
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.667935804357581e-05,
+      "loss": 0.8428,
+      "step": 13475
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6662582934793637e-05,
+      "loss": 0.8964,
+      "step": 13476
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6645815899276852e-05,
+      "loss": 0.8495,
+      "step": 13477
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6629056937763887e-05,
+      "loss": 0.8769,
+      "step": 13478
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6612306050992664e-05,
+      "loss": 0.8895,
+      "step": 13479
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6595563239700796e-05,
+      "loss": 0.8141,
+      "step": 13480
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6578828504625622e-05,
+      "loss": 0.8891,
+      "step": 13481
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6562101846503973e-05,
+      "loss": 0.8262,
+      "step": 13482
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6545383266072445e-05,
+      "loss": 0.8675,
+      "step": 13483
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6528672764067266e-05,
+      "loss": 0.7717,
+      "step": 13484
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.651197034122427e-05,
+      "loss": 0.9105,
+      "step": 13485
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6495275998278892e-05,
+      "loss": 0.9178,
+      "step": 13486
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.647858973596632e-05,
+      "loss": 0.8584,
+      "step": 13487
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.646191155502137e-05,
+      "loss": 0.8792,
+      "step": 13488
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.644524145617836e-05,
+      "loss": 0.9152,
+      "step": 13489
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.642857944017149e-05,
+      "loss": 0.8583,
+      "step": 13490
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6411925507734337e-05,
+      "loss": 0.8826,
+      "step": 13491
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.639527965960035e-05,
+      "loss": 0.8064,
+      "step": 13492
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6378641896502532e-05,
+      "loss": 0.869,
+      "step": 13493
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.636201221917346e-05,
+      "loss": 0.8362,
+      "step": 13494
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6345390628345504e-05,
+      "loss": 0.8672,
+      "step": 13495
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.632877712475045e-05,
+      "loss": 0.8809,
+      "step": 13496
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6312171709120073e-05,
+      "loss": 0.8912,
+      "step": 13497
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6295574382185453e-05,
+      "loss": 0.8725,
+      "step": 13498
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6278985144677493e-05,
+      "loss": 0.8494,
+      "step": 13499
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6262403997326726e-05,
+      "loss": 0.8858,
+      "step": 13500
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.624583094086327e-05,
+      "loss": 0.8544,
+      "step": 13501
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6229265976016907e-05,
+      "loss": 0.933,
+      "step": 13502
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.621270910351711e-05,
+      "loss": 0.8702,
+      "step": 13503
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.619616032409297e-05,
+      "loss": 0.8592,
+      "step": 13504
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6179619638473163e-05,
+      "loss": 0.8713,
+      "step": 13505
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6163087047386095e-05,
+      "loss": 0.8566,
+      "step": 13506
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.6146562551559796e-05,
+      "loss": 0.8235,
+      "step": 13507
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.613004615172182e-05,
+      "loss": 0.8362,
+      "step": 13508
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.6113537848599635e-05,
+      "loss": 0.7989,
+      "step": 13509
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.6097037642920033e-05,
+      "loss": 0.8972,
+      "step": 13510
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.6080545535409674e-05,
+      "loss": 0.8879,
+      "step": 13511
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.6064061526794805e-05,
+      "loss": 0.8686,
+      "step": 13512
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.604758561780124e-05,
+      "loss": 0.8432,
+      "step": 13513
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.6031117809154518e-05,
+      "loss": 0.8745,
+      "step": 13514
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.6014658101579826e-05,
+      "loss": 0.8345,
+      "step": 13515
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.599820649580197e-05,
+      "loss": 0.9325,
+      "step": 13516
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.5981762992545323e-05,
+      "loss": 0.7696,
+      "step": 13517
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.596532759253404e-05,
+      "loss": 0.8676,
+      "step": 13518
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.5948900296491875e-05,
+      "loss": 0.855,
+      "step": 13519
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.5932481105142093e-05,
+      "loss": 0.8213,
+      "step": 13520
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.5916070019207874e-05,
+      "loss": 0.931,
+      "step": 13521
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.5899667039411748e-05,
+      "loss": 0.9164,
+      "step": 13522
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.5883272166476114e-05,
+      "loss": 0.8142,
+      "step": 13523
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.5866885401122845e-05,
+      "loss": 0.8899,
+      "step": 13524
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.5850506744073558e-05,
+      "loss": 0.8005,
+      "step": 13525
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.583413619604952e-05,
+      "loss": 0.9118,
+      "step": 13526
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.5817773757771513e-05,
+      "loss": 0.8412,
+      "step": 13527
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.5801419429960185e-05,
+      "loss": 0.8772,
+      "step": 13528
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.57850732133356e-05,
+      "loss": 0.8455,
+      "step": 13529
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.5768735108617626e-05,
+      "loss": 0.8298,
+      "step": 13530
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.5752405116525715e-05,
+      "loss": 0.8504,
+      "step": 13531
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.5736083237778886e-05,
+      "loss": 0.816,
+      "step": 13532
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.5719769473095924e-05,
+      "loss": 0.9062,
+      "step": 13533
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.570346382319521e-05,
+      "loss": 0.859,
+      "step": 13534
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.5687166288794763e-05,
+      "loss": 0.8227,
+      "step": 13535
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.5670876870612217e-05,
+      "loss": 0.8009,
+      "step": 13536
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.5654595569364906e-05,
+      "loss": 0.9135,
+      "step": 13537
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.5638322385769788e-05,
+      "loss": 0.8549,
+      "step": 13538
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.562205732054336e-05,
+      "loss": 0.867,
+      "step": 13539
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.5605800374402002e-05,
+      "loss": 0.9272,
+      "step": 13540
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.5589551548061477e-05,
+      "loss": 0.8442,
+      "step": 13541
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.5573310842237344e-05,
+      "loss": 0.8888,
+      "step": 13542
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.5557078257644763e-05,
+      "loss": 0.864,
+      "step": 13543
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.554085379499852e-05,
+      "loss": 0.7518,
+      "step": 13544
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.552463745501307e-05,
+      "loss": 0.8043,
+      "step": 13545
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.55084292384025e-05,
+      "loss": 0.781,
+      "step": 13546
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.5492229145880578e-05,
+      "loss": 0.8434,
+      "step": 13547
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.5476037178160573e-05,
+      "loss": 0.8605,
+      "step": 13548
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.5459853335955587e-05,
+      "loss": 0.8733,
+      "step": 13549
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.544367761997827e-05,
+      "loss": 0.8199,
+      "step": 13550
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.5427510030940872e-05,
+      "loss": 0.9081,
+      "step": 13551
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.541135056955536e-05,
+      "loss": 0.8053,
+      "step": 13552
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.539519923653332e-05,
+      "loss": 0.9098,
+      "step": 13553
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.5379056032586004e-05,
+      "loss": 0.8746,
+      "step": 13554
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.53629209584242e-05,
+      "loss": 0.8858,
+      "step": 13555
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.534679401475847e-05,
+      "loss": 0.9001,
+      "step": 13556
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.5330675202299004e-05,
+      "loss": 0.8168,
+      "step": 13557
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.531456452175546e-05,
+      "loss": 0.8543,
+      "step": 13558
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.529846197383744e-05,
+      "loss": 0.8136,
+      "step": 13559
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5282367559253897e-05,
+      "loss": 0.8723,
+      "step": 13560
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.526628127871359e-05,
+      "loss": 0.8607,
+      "step": 13561
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5250203132924935e-05,
+      "loss": 0.874,
+      "step": 13562
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5234133122595828e-05,
+      "loss": 0.8755,
+      "step": 13563
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5218071248433952e-05,
+      "loss": 0.8929,
+      "step": 13564
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5202017511146605e-05,
+      "loss": 0.8277,
+      "step": 13565
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5185971911440756e-05,
+      "loss": 0.8919,
+      "step": 13566
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5169934450022882e-05,
+      "loss": 0.865,
+      "step": 13567
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5153905127599243e-05,
+      "loss": 0.9064,
+      "step": 13568
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5137883944875742e-05,
+      "loss": 0.8416,
+      "step": 13569
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5121870902557722e-05,
+      "loss": 0.9178,
+      "step": 13570
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5105866001350488e-05,
+      "loss": 0.8364,
+      "step": 13571
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5089869241958698e-05,
+      "loss": 0.87,
+      "step": 13572
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5073880625086812e-05,
+      "loss": 0.8637,
+      "step": 13573
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5057900151438908e-05,
+      "loss": 0.8566,
+      "step": 13574
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5041927821718649e-05,
+      "loss": 0.8434,
+      "step": 13575
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5025963636629425e-05,
+      "loss": 0.8107,
+      "step": 13576
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.501000759687412e-05,
+      "loss": 0.9309,
+      "step": 13577
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4994059703155504e-05,
+      "loss": 0.8258,
+      "step": 13578
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.497811995617573e-05,
+      "loss": 0.8366,
+      "step": 13579
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.496218835663672e-05,
+      "loss": 0.8337,
+      "step": 13580
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4946264905240092e-05,
+      "loss": 0.8573,
+      "step": 13581
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4930349602686933e-05,
+      "loss": 0.8238,
+      "step": 13582
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4914442449678146e-05,
+      "loss": 0.8552,
+      "step": 13583
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.489854344691417e-05,
+      "loss": 0.8231,
+      "step": 13584
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4882652595095182e-05,
+      "loss": 0.9399,
+      "step": 13585
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.486676989492084e-05,
+      "loss": 0.9381,
+      "step": 13586
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4850895347090588e-05,
+      "loss": 0.8238,
+      "step": 13587
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4835028952303465e-05,
+      "loss": 0.8501,
+      "step": 13588
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.481917071125809e-05,
+      "loss": 0.817,
+      "step": 13589
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4803320624652906e-05,
+      "loss": 0.926,
+      "step": 13590
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4787478693185752e-05,
+      "loss": 0.9305,
+      "step": 13591
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4771644917554294e-05,
+      "loss": 0.8917,
+      "step": 13592
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4755819298455753e-05,
+      "loss": 0.8488,
+      "step": 13593
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4740001836586992e-05,
+      "loss": 0.8574,
+      "step": 13594
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4724192532644543e-05,
+      "loss": 0.8735,
+      "step": 13595
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4708391387324582e-05,
+      "loss": 0.8212,
+      "step": 13596
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4692598401322931e-05,
+      "loss": 0.8629,
+      "step": 13597
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4676813575334992e-05,
+      "loss": 0.8855,
+      "step": 13598
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4661036910055847e-05,
+      "loss": 0.906,
+      "step": 13599
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4645268406180302e-05,
+      "loss": 0.85,
+      "step": 13600
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4629508064402598e-05,
+      "loss": 0.876,
+      "step": 13601
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4613755885416824e-05,
+      "loss": 0.8509,
+      "step": 13602
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4598011869916606e-05,
+      "loss": 0.8817,
+      "step": 13603
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4582276018595275e-05,
+      "loss": 0.8525,
+      "step": 13604
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.456654833214568e-05,
+      "loss": 0.8438,
+      "step": 13605
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.455082881126042e-05,
+      "loss": 0.9156,
+      "step": 13606
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4535117456631764e-05,
+      "loss": 0.8804,
+      "step": 13607
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4519414268951448e-05,
+      "loss": 0.8616,
+      "step": 13608
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4503719248911074e-05,
+      "loss": 0.8455,
+      "step": 13609
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4488032397201711e-05,
+      "loss": 0.8703,
+      "step": 13610
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4472353714514142e-05,
+      "loss": 0.8353,
+      "step": 13611
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4456683201538812e-05,
+      "loss": 0.9025,
+      "step": 13612
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4441020858965704e-05,
+      "loss": 0.8713,
+      "step": 13613
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4425366687484532e-05,
+      "loss": 0.8842,
+      "step": 13614
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4409720687784656e-05,
+      "loss": 0.8762,
+      "step": 13615
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4394082860555058e-05,
+      "loss": 0.8553,
+      "step": 13616
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4378453206484299e-05,
+      "loss": 0.8426,
+      "step": 13617
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.436283172626065e-05,
+      "loss": 0.884,
+      "step": 13618
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4347218420572051e-05,
+      "loss": 0.9032,
+      "step": 13619
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4331613290105927e-05,
+      "loss": 0.8792,
+      "step": 13620
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4316016335549576e-05,
+      "loss": 0.8417,
+      "step": 13621
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4300427557589712e-05,
+      "loss": 0.825,
+      "step": 13622
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4284846956912834e-05,
+      "loss": 0.8308,
+      "step": 13623
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4269274534205057e-05,
+      "loss": 0.8212,
+      "step": 13624
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4253710290152033e-05,
+      "loss": 0.883,
+      "step": 13625
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4238154225439216e-05,
+      "loss": 0.8279,
+      "step": 13626
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4222606340751542e-05,
+      "loss": 0.7829,
+      "step": 13627
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4207066636773757e-05,
+      "loss": 0.8561,
+      "step": 13628
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4191535114190069e-05,
+      "loss": 0.8502,
+      "step": 13629
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4176011773684439e-05,
+      "loss": 0.8723,
+      "step": 13630
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4160496615940455e-05,
+      "loss": 0.8848,
+      "step": 13631
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4144989641641282e-05,
+      "loss": 0.8473,
+      "step": 13632
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4129490851469796e-05,
+      "loss": 0.8471,
+      "step": 13633
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4114000246108495e-05,
+      "loss": 0.8989,
+      "step": 13634
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.409851782623952e-05,
+      "loss": 0.8478,
+      "step": 13635
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4083043592544598e-05,
+      "loss": 0.8937,
+      "step": 13636
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4067577545705157e-05,
+      "loss": 0.8894,
+      "step": 13637
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4052119686402276e-05,
+      "loss": 0.8687,
+      "step": 13638
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4036670015316545e-05,
+      "loss": 0.8729,
+      "step": 13639
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.402122853312844e-05,
+      "loss": 0.8301,
+      "step": 13640
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4005795240517794e-05,
+      "loss": 0.8726,
+      "step": 13641
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.3990370138164266e-05,
+      "loss": 0.9177,
+      "step": 13642
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.3974953226747134e-05,
+      "loss": 0.8759,
+      "step": 13643
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.395954450694521e-05,
+      "loss": 0.8662,
+      "step": 13644
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.3944143979437041e-05,
+      "loss": 0.831,
+      "step": 13645
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.3928751644900817e-05,
+      "loss": 0.84,
+      "step": 13646
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.3913367504014352e-05,
+      "loss": 0.8638,
+      "step": 13647
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.3897991557455014e-05,
+      "loss": 0.859,
+      "step": 13648
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.388262380589993e-05,
+      "loss": 0.8662,
+      "step": 13649
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.3867264250025869e-05,
+      "loss": 0.8305,
+      "step": 13650
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.3851912890509044e-05,
+      "loss": 0.8683,
+      "step": 13651
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.3836569728025627e-05,
+      "loss": 0.8591,
+      "step": 13652
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.3821234763251145e-05,
+      "loss": 0.8627,
+      "step": 13653
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.38059079968609e-05,
+      "loss": 0.9147,
+      "step": 13654
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.3790589429529821e-05,
+      "loss": 0.8594,
+      "step": 13655
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.3775279061932433e-05,
+      "loss": 0.7701,
+      "step": 13656
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.3759976894742976e-05,
+      "loss": 0.9266,
+      "step": 13657
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.3744682928635156e-05,
+      "loss": 0.8149,
+      "step": 13658
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.372939716428263e-05,
+      "loss": 0.8399,
+      "step": 13659
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.3714119602358355e-05,
+      "loss": 0.8463,
+      "step": 13660
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.3698850243535166e-05,
+      "loss": 0.8695,
+      "step": 13661
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.3683589088485416e-05,
+      "loss": 0.8279,
+      "step": 13662
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.3668336137881122e-05,
+      "loss": 0.8231,
+      "step": 13663
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.3653091392393946e-05,
+      "loss": 0.9139,
+      "step": 13664
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.363785485269522e-05,
+      "loss": 0.7566,
+      "step": 13665
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.3622626519455894e-05,
+      "loss": 0.8479,
+      "step": 13666
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.3607406393346477e-05,
+      "loss": 0.8817,
+      "step": 13667
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.3592194475037256e-05,
+      "loss": 0.8546,
+      "step": 13668
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.3576990765198072e-05,
+      "loss": 0.9302,
+      "step": 13669
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.3561795264498366e-05,
+      "loss": 0.8772,
+      "step": 13670
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.3546607973607384e-05,
+      "loss": 0.8804,
+      "step": 13671
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.3531428893193787e-05,
+      "loss": 0.829,
+      "step": 13672
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.3516258023926043e-05,
+      "loss": 0.8935,
+      "step": 13673
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.3501095366472216e-05,
+      "loss": 0.8462,
+      "step": 13674
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.348594092149995e-05,
+      "loss": 0.8916,
+      "step": 13675
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.347079468967658e-05,
+      "loss": 0.8203,
+      "step": 13676
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.345565667166908e-05,
+      "loss": 0.8178,
+      "step": 13677
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.3440526868144099e-05,
+      "loss": 0.9239,
+      "step": 13678
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.3425405279767788e-05,
+      "loss": 0.8834,
+      "step": 13679
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.3410291907206063e-05,
+      "loss": 0.8633,
+      "step": 13680
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.33951867511245e-05,
+      "loss": 0.8324,
+      "step": 13681
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.3380089812188145e-05,
+      "loss": 0.8533,
+      "step": 13682
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.3365001091061869e-05,
+      "loss": 0.8686,
+      "step": 13683
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.3349920588410092e-05,
+      "loss": 0.8365,
+      "step": 13684
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.3334848304896907e-05,
+      "loss": 0.8742,
+      "step": 13685
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.3319784241185963e-05,
+      "loss": 0.8185,
+      "step": 13686
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.3304728397940635e-05,
+      "loss": 0.8362,
+      "step": 13687
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.3289680775823932e-05,
+      "loss": 0.8754,
+      "step": 13688
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.3274641375498386e-05,
+      "loss": 0.8139,
+      "step": 13689
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.3259610197626403e-05,
+      "loss": 0.8875,
+      "step": 13690
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.3244587242869767e-05,
+      "loss": 0.8784,
+      "step": 13691
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.3229572511890053e-05,
+      "loss": 0.8961,
+      "step": 13692
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.321456600534845e-05,
+      "loss": 0.8365,
+      "step": 13693
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.3199567723905737e-05,
+      "loss": 0.8541,
+      "step": 13694
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.3184577668222364e-05,
+      "loss": 0.8171,
+      "step": 13695
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.3169595838958426e-05,
+      "loss": 0.8354,
+      "step": 13696
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.3154622236773706e-05,
+      "loss": 0.8707,
+      "step": 13697
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.3139656862327476e-05,
+      "loss": 0.9095,
+      "step": 13698
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.3124699716278765e-05,
+      "loss": 0.8666,
+      "step": 13699
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.3109750799286246e-05,
+      "loss": 0.8286,
+      "step": 13700
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.3094810112008126e-05,
+      "loss": 0.9002,
+      "step": 13701
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.3079877655102412e-05,
+      "loss": 0.8942,
+      "step": 13702
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.3064953429226578e-05,
+      "loss": 0.8784,
+      "step": 13703
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.3050037435037855e-05,
+      "loss": 0.8673,
+      "step": 13704
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.303512967319307e-05,
+      "loss": 0.8524,
+      "step": 13705
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.3020230144348632e-05,
+      "loss": 0.8491,
+      "step": 13706
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.3005338849160687e-05,
+      "loss": 0.8574,
+      "step": 13707
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.2990455788284972e-05,
+      "loss": 0.8638,
+      "step": 13708
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.2975580962376876e-05,
+      "loss": 0.9157,
+      "step": 13709
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.2960714372091387e-05,
+      "loss": 0.8025,
+      "step": 13710
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.2945856018083136e-05,
+      "loss": 0.8902,
+      "step": 13711
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.2931005901006465e-05,
+      "loss": 0.8436,
+      "step": 13712
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2916164021515253e-05,
+      "loss": 0.7928,
+      "step": 13713
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2901330380263089e-05,
+      "loss": 0.813,
+      "step": 13714
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2886504977903158e-05,
+      "loss": 0.8919,
+      "step": 13715
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2871687815088318e-05,
+      "loss": 0.8803,
+      "step": 13716
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2856878892471003e-05,
+      "loss": 0.8091,
+      "step": 13717
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2842078210703357e-05,
+      "loss": 0.9182,
+      "step": 13718
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2827285770437148e-05,
+      "loss": 0.8639,
+      "step": 13719
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2812501572323655e-05,
+      "loss": 0.865,
+      "step": 13720
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2797725617014067e-05,
+      "loss": 0.865,
+      "step": 13721
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.278295790515891e-05,
+      "loss": 0.8916,
+      "step": 13722
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2768198437408507e-05,
+      "loss": 0.9411,
+      "step": 13723
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2753447214412873e-05,
+      "loss": 0.8575,
+      "step": 13724
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2738704236821464e-05,
+      "loss": 0.8027,
+      "step": 13725
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.272396950528354e-05,
+      "loss": 0.9396,
+      "step": 13726
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2709243020447936e-05,
+      "loss": 0.8416,
+      "step": 13727
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.269452478296318e-05,
+      "loss": 0.8994,
+      "step": 13728
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.267981479347733e-05,
+      "loss": 0.9574,
+      "step": 13729
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2665113052638155e-05,
+      "loss": 0.8179,
+      "step": 13730
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2650419561093075e-05,
+      "loss": 0.8957,
+      "step": 13731
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2635734319489035e-05,
+      "loss": 0.8681,
+      "step": 13732
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2621057328472807e-05,
+      "loss": 0.8486,
+      "step": 13733
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2606388588690632e-05,
+      "loss": 0.8089,
+      "step": 13734
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2591728100788481e-05,
+      "loss": 0.8486,
+      "step": 13735
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2577075865411902e-05,
+      "loss": 0.8908,
+      "step": 13736
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2562431883206094e-05,
+      "loss": 0.8328,
+      "step": 13737
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2547796154815938e-05,
+      "loss": 0.834,
+      "step": 13738
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2533168680885876e-05,
+      "loss": 0.7889,
+      "step": 13739
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2518549462060103e-05,
+      "loss": 0.8531,
+      "step": 13740
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2503938498982304e-05,
+      "loss": 0.7797,
+      "step": 13741
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2489335792295898e-05,
+      "loss": 0.8432,
+      "step": 13742
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2474741342643948e-05,
+      "loss": 0.8528,
+      "step": 13743
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.246015515066905e-05,
+      "loss": 0.8697,
+      "step": 13744
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.244557721701356e-05,
+      "loss": 0.8489,
+      "step": 13745
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2431007542319406e-05,
+      "loss": 0.8209,
+      "step": 13746
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2416446127228187e-05,
+      "loss": 0.9084,
+      "step": 13747
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2401892972381057e-05,
+      "loss": 0.8438,
+      "step": 13748
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2387348078418904e-05,
+      "loss": 0.8383,
+      "step": 13749
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2372811445982235e-05,
+      "loss": 0.8849,
+      "step": 13750
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2358283075711074e-05,
+      "loss": 0.8616,
+      "step": 13751
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2343762968245332e-05,
+      "loss": 0.8305,
+      "step": 13752
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2329251124224273e-05,
+      "loss": 0.8307,
+      "step": 13753
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2314747544286964e-05,
+      "loss": 0.8961,
+      "step": 13754
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2300252229072095e-05,
+      "loss": 0.8694,
+      "step": 13755
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2285765179217934e-05,
+      "loss": 0.8392,
+      "step": 13756
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2271286395362436e-05,
+      "loss": 0.8656,
+      "step": 13757
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2256815878143158e-05,
+      "loss": 0.8654,
+      "step": 13758
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2242353628197367e-05,
+      "loss": 0.88,
+      "step": 13759
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2227899646161822e-05,
+      "loss": 0.8036,
+      "step": 13760
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2213453932673058e-05,
+      "loss": 0.8883,
+      "step": 13761
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.219901648836721e-05,
+      "loss": 0.8254,
+      "step": 13762
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.218458731387997e-05,
+      "loss": 0.9165,
+      "step": 13763
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2170166409846739e-05,
+      "loss": 0.869,
+      "step": 13764
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.215575377690259e-05,
+      "loss": 0.8781,
+      "step": 13765
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.2141349415682168e-05,
+      "loss": 0.8327,
+      "step": 13766
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.2126953326819723e-05,
+      "loss": 0.7982,
+      "step": 13767
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.2112565510949236e-05,
+      "loss": 0.8157,
+      "step": 13768
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.2098185968704268e-05,
+      "loss": 0.8411,
+      "step": 13769
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.2083814700717954e-05,
+      "loss": 0.8183,
+      "step": 13770
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.2069451707623258e-05,
+      "loss": 0.9215,
+      "step": 13771
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.2055096990052561e-05,
+      "loss": 0.8331,
+      "step": 13772
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.2040750548638002e-05,
+      "loss": 0.8174,
+      "step": 13773
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.2026412384011365e-05,
+      "loss": 0.9056,
+      "step": 13774
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.2012082496803945e-05,
+      "loss": 0.8373,
+      "step": 13775
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1997760887646815e-05,
+      "loss": 0.8124,
+      "step": 13776
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1983447557170602e-05,
+      "loss": 0.8631,
+      "step": 13777
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1969142506005648e-05,
+      "loss": 0.802,
+      "step": 13778
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1954845734781806e-05,
+      "loss": 0.8557,
+      "step": 13779
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1940557244128681e-05,
+      "loss": 0.7992,
+      "step": 13780
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1926277034675459e-05,
+      "loss": 0.8369,
+      "step": 13781
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1912005107050928e-05,
+      "loss": 0.8439,
+      "step": 13782
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1897741461883627e-05,
+      "loss": 0.8788,
+      "step": 13783
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1883486099801588e-05,
+      "loss": 0.8637,
+      "step": 13784
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1869239021432554e-05,
+      "loss": 0.8144,
+      "step": 13785
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1855000227403978e-05,
+      "loss": 0.8497,
+      "step": 13786
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.184076971834276e-05,
+      "loss": 0.8924,
+      "step": 13787
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1826547494875596e-05,
+      "loss": 0.7869,
+      "step": 13788
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1812333557628697e-05,
+      "loss": 0.8763,
+      "step": 13789
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1798127907228096e-05,
+      "loss": 0.8207,
+      "step": 13790
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1783930544299226e-05,
+      "loss": 0.8433,
+      "step": 13791
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1769741469467322e-05,
+      "loss": 0.8264,
+      "step": 13792
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1755560683357214e-05,
+      "loss": 0.8703,
+      "step": 13793
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1741388186593294e-05,
+      "loss": 0.8663,
+      "step": 13794
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1727223979799685e-05,
+      "loss": 0.943,
+      "step": 13795
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1713068063600108e-05,
+      "loss": 0.8452,
+      "step": 13796
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1698920438617933e-05,
+      "loss": 0.8589,
+      "step": 13797
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1684781105476105e-05,
+      "loss": 0.8432,
+      "step": 13798
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1670650064797305e-05,
+      "loss": 0.8854,
+      "step": 13799
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1656527317203769e-05,
+      "loss": 0.8432,
+      "step": 13800
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1642412863317332e-05,
+      "loss": 0.8915,
+      "step": 13801
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.162830670375965e-05,
+      "loss": 0.8444,
+      "step": 13802
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1614208839151785e-05,
+      "loss": 0.7625,
+      "step": 13803
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1600119270114573e-05,
+      "loss": 0.8644,
+      "step": 13804
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1586037997268472e-05,
+      "loss": 0.8184,
+      "step": 13805
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.15719650212335e-05,
+      "loss": 0.9165,
+      "step": 13806
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1557900342629403e-05,
+      "loss": 0.888,
+      "step": 13807
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1543843962075485e-05,
+      "loss": 0.8912,
+      "step": 13808
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1529795880190764e-05,
+      "loss": 0.8782,
+      "step": 13809
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1515756097593788e-05,
+      "loss": 0.8332,
+      "step": 13810
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1501724614902842e-05,
+      "loss": 0.8634,
+      "step": 13811
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1487701432735808e-05,
+      "loss": 0.8349,
+      "step": 13812
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.147368655171015e-05,
+      "loss": 0.8328,
+      "step": 13813
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1459679972443082e-05,
+      "loss": 0.8819,
+      "step": 13814
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1445681695551314e-05,
+      "loss": 0.8455,
+      "step": 13815
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.1431691721651327e-05,
+      "loss": 0.8511,
+      "step": 13816
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.1417710051359099e-05,
+      "loss": 0.8993,
+      "step": 13817
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.1403736685290357e-05,
+      "loss": 0.8534,
+      "step": 13818
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.1389771624060453e-05,
+      "loss": 0.9381,
+      "step": 13819
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.1375814868284228e-05,
+      "loss": 0.8841,
+      "step": 13820
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.1361866418576395e-05,
+      "loss": 0.8818,
+      "step": 13821
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.134792627555108e-05,
+      "loss": 0.8226,
+      "step": 13822
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.133399443982217e-05,
+      "loss": 0.8783,
+      "step": 13823
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.1320070912003178e-05,
+      "loss": 0.8372,
+      "step": 13824
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.1306155692707187e-05,
+      "loss": 0.8268,
+      "step": 13825
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.1292248782546977e-05,
+      "loss": 0.8075,
+      "step": 13826
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.1278350182134923e-05,
+      "loss": 0.8172,
+      "step": 13827
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.1264459892083113e-05,
+      "loss": 0.8708,
+      "step": 13828
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.1250577913003101e-05,
+      "loss": 0.8072,
+      "step": 13829
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.1236704245506247e-05,
+      "loss": 0.8893,
+      "step": 13830
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.122283889020348e-05,
+      "loss": 0.7702,
+      "step": 13831
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.1208981847705292e-05,
+      "loss": 0.8273,
+      "step": 13832
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.1195133118621991e-05,
+      "loss": 0.8781,
+      "step": 13833
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.1181292703563317e-05,
+      "loss": 0.9013,
+      "step": 13834
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.1167460603138758e-05,
+      "loss": 0.8892,
+      "step": 13835
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.1153636817957447e-05,
+      "loss": 0.8348,
+      "step": 13836
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.1139821348628055e-05,
+      "loss": 0.8321,
+      "step": 13837
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.112601419575896e-05,
+      "loss": 0.8773,
+      "step": 13838
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.1112215359958189e-05,
+      "loss": 0.8216,
+      "step": 13839
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.1098424841833387e-05,
+      "loss": 0.9395,
+      "step": 13840
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.108464264199176e-05,
+      "loss": 0.8573,
+      "step": 13841
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.1070868761040243e-05,
+      "loss": 0.8574,
+      "step": 13842
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.1057103199585395e-05,
+      "loss": 0.8565,
+      "step": 13843
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.1043345958233309e-05,
+      "loss": 0.8408,
+      "step": 13844
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.1029597037589833e-05,
+      "loss": 0.8745,
+      "step": 13845
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.1015856438260397e-05,
+      "loss": 0.8684,
+      "step": 13846
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.1002124160850113e-05,
+      "loss": 0.9118,
+      "step": 13847
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.098840020596359e-05,
+      "loss": 0.8347,
+      "step": 13848
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.0974684574205207e-05,
+      "loss": 0.8781,
+      "step": 13849
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.0960977266178973e-05,
+      "loss": 0.8209,
+      "step": 13850
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.0947278282488404e-05,
+      "loss": 0.8542,
+      "step": 13851
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.0933587623736818e-05,
+      "loss": 0.9037,
+      "step": 13852
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.0919905290527043e-05,
+      "loss": 0.8529,
+      "step": 13853
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.0906231283461555e-05,
+      "loss": 0.8355,
+      "step": 13854
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.0892565603142579e-05,
+      "loss": 0.8385,
+      "step": 13855
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.0878908250171771e-05,
+      "loss": 0.8806,
+      "step": 13856
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.0865259225150582e-05,
+      "loss": 0.889,
+      "step": 13857
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.0851618528680063e-05,
+      "loss": 0.823,
+      "step": 13858
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.0837986161360913e-05,
+      "loss": 0.8493,
+      "step": 13859
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.0824362123793342e-05,
+      "loss": 0.8284,
+      "step": 13860
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.0810746416577333e-05,
+      "loss": 0.8487,
+      "step": 13861
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.0797139040312477e-05,
+      "loss": 0.8827,
+      "step": 13862
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.078353999559789e-05,
+      "loss": 0.8537,
+      "step": 13863
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.0769949283032543e-05,
+      "loss": 0.8422,
+      "step": 13864
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.0756366903214799e-05,
+      "loss": 0.9233,
+      "step": 13865
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.0742792856742779e-05,
+      "loss": 0.8205,
+      "step": 13866
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.0729227144214248e-05,
+      "loss": 0.9353,
+      "step": 13867
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0715669766226533e-05,
+      "loss": 0.9093,
+      "step": 13868
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0702120723376662e-05,
+      "loss": 0.8768,
+      "step": 13869
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0688580016261206e-05,
+      "loss": 0.879,
+      "step": 13870
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0675047645476532e-05,
+      "loss": 0.8758,
+      "step": 13871
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0661523611618451e-05,
+      "loss": 0.8953,
+      "step": 13872
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0648007915282554e-05,
+      "loss": 0.8767,
+      "step": 13873
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0634500557063987e-05,
+      "loss": 0.8975,
+      "step": 13874
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0621001537557496e-05,
+      "loss": 0.8292,
+      "step": 13875
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0607510857357561e-05,
+      "loss": 0.9302,
+      "step": 13876
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.059402851705824e-05,
+      "loss": 0.8744,
+      "step": 13877
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.058055451725326e-05,
+      "loss": 0.8826,
+      "step": 13878
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0567088858535878e-05,
+      "loss": 0.8769,
+      "step": 13879
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0553631541499088e-05,
+      "loss": 0.7994,
+      "step": 13880
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0540182566735524e-05,
+      "loss": 0.9165,
+      "step": 13881
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0526741934837291e-05,
+      "loss": 0.9266,
+      "step": 13882
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0513309646396408e-05,
+      "loss": 0.8682,
+      "step": 13883
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0499885702004242e-05,
+      "loss": 0.8805,
+      "step": 13884
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0486470102251966e-05,
+      "loss": 0.8421,
+      "step": 13885
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0473062847730353e-05,
+      "loss": 0.8515,
+      "step": 13886
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0459663939029751e-05,
+      "loss": 0.8967,
+      "step": 13887
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.04462733767402e-05,
+      "loss": 0.8895,
+      "step": 13888
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.043289116145134e-05,
+      "loss": 0.8611,
+      "step": 13889
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0419517293752523e-05,
+      "loss": 0.9413,
+      "step": 13890
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0406151774232564e-05,
+      "loss": 0.909,
+      "step": 13891
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0392794603480082e-05,
+      "loss": 0.9344,
+      "step": 13892
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0379445782083253e-05,
+      "loss": 0.84,
+      "step": 13893
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0366105310629825e-05,
+      "loss": 0.8651,
+      "step": 13894
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0352773189707377e-05,
+      "loss": 0.8979,
+      "step": 13895
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.033944941990288e-05,
+      "loss": 0.811,
+      "step": 13896
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0326134001803112e-05,
+      "loss": 0.8732,
+      "step": 13897
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0312826935994336e-05,
+      "loss": 0.8619,
+      "step": 13898
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0299528223062615e-05,
+      "loss": 0.8694,
+      "step": 13899
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0286237863593528e-05,
+      "loss": 0.9066,
+      "step": 13900
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.027295585817225e-05,
+      "loss": 0.8574,
+      "step": 13901
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0259682207383803e-05,
+      "loss": 0.8174,
+      "step": 13902
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0246416911812539e-05,
+      "loss": 0.921,
+      "step": 13903
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0233159972042683e-05,
+      "loss": 0.8745,
+      "step": 13904
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.021991138865801e-05,
+      "loss": 0.8896,
+      "step": 13905
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0206671162241854e-05,
+      "loss": 0.8036,
+      "step": 13906
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0193439293377305e-05,
+      "loss": 0.8539,
+      "step": 13907
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0180215782647007e-05,
+      "loss": 0.8466,
+      "step": 13908
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0167000630633273e-05,
+      "loss": 0.8428,
+      "step": 13909
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0153793837918012e-05,
+      "loss": 0.8478,
+      "step": 13910
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0140595405082787e-05,
+      "loss": 0.818,
+      "step": 13911
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0127405332708839e-05,
+      "loss": 0.8957,
+      "step": 13912
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0114223621376884e-05,
+      "loss": 0.8366,
+      "step": 13913
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0101050271667522e-05,
+      "loss": 0.8435,
+      "step": 13914
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0087885284160736e-05,
+      "loss": 0.8363,
+      "step": 13915
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0074728659436261e-05,
+      "loss": 0.9169,
+      "step": 13916
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.00615803980735e-05,
+      "loss": 0.8948,
+      "step": 13917
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.004844050065139e-05,
+      "loss": 0.8915,
+      "step": 13918
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.0035308967748536e-05,
+      "loss": 0.8918,
+      "step": 13919
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.0022185799943229e-05,
+      "loss": 0.8835,
+      "step": 13920
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.0009070997813342e-05,
+      "loss": 0.8293,
+      "step": 13921
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.995964561936365e-06,
+      "loss": 0.8628,
+      "step": 13922
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.98286649288942e-06,
+      "loss": 0.8292,
+      "step": 13923
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.969776791249353e-06,
+      "loss": 0.8301,
+      "step": 13924
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.956695457592503e-06,
+      "loss": 0.8879,
+      "step": 13925
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.943622492494897e-06,
+      "loss": 0.8411,
+      "step": 13926
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.930557896532255e-06,
+      "loss": 0.8514,
+      "step": 13927
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.917501670279872e-06,
+      "loss": 0.8523,
+      "step": 13928
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.904453814312642e-06,
+      "loss": 0.8397,
+      "step": 13929
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.89141432920515e-06,
+      "loss": 0.848,
+      "step": 13930
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.878383215531606e-06,
+      "loss": 0.8767,
+      "step": 13931
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.86536047386577e-06,
+      "loss": 0.9167,
+      "step": 13932
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.852346104781186e-06,
+      "loss": 0.7733,
+      "step": 13933
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.839340108850881e-06,
+      "loss": 0.8664,
+      "step": 13934
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.8263424866476e-06,
+      "loss": 0.8144,
+      "step": 13935
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.813353238743705e-06,
+      "loss": 0.7851,
+      "step": 13936
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.80037236571114e-06,
+      "loss": 0.8437,
+      "step": 13937
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.787399868121538e-06,
+      "loss": 0.8686,
+      "step": 13938
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.77443574654615e-06,
+      "loss": 0.9158,
+      "step": 13939
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.761480001555855e-06,
+      "loss": 0.7936,
+      "step": 13940
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.748532633721153e-06,
+      "loss": 0.8397,
+      "step": 13941
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.73559364361214e-06,
+      "loss": 0.8464,
+      "step": 13942
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.722663031798674e-06,
+      "loss": 0.8535,
+      "step": 13943
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.709740798850031e-06,
+      "loss": 0.8792,
+      "step": 13944
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.69682694533538e-06,
+      "loss": 0.8588,
+      "step": 13945
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.683921471823287e-06,
+      "loss": 0.8203,
+      "step": 13946
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.671024378882076e-06,
+      "loss": 0.7691,
+      "step": 13947
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.658135667079694e-06,
+      "loss": 0.8267,
+      "step": 13948
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.645255336983638e-06,
+      "loss": 0.848,
+      "step": 13949
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.632383389161149e-06,
+      "loss": 0.8179,
+      "step": 13950
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.619519824178969e-06,
+      "loss": 0.8983,
+      "step": 13951
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.606664642603647e-06,
+      "loss": 0.7812,
+      "step": 13952
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.593817845001196e-06,
+      "loss": 0.879,
+      "step": 13953
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.58097943193732e-06,
+      "loss": 0.8815,
+      "step": 13954
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.568149403977432e-06,
+      "loss": 0.818,
+      "step": 13955
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.555327761686394e-06,
+      "loss": 0.887,
+      "step": 13956
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.542514505628885e-06,
+      "loss": 0.9027,
+      "step": 13957
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.5297096363691e-06,
+      "loss": 0.8972,
+      "step": 13958
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.516913154470941e-06,
+      "loss": 0.8513,
+      "step": 13959
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.504125060497849e-06,
+      "loss": 0.9066,
+      "step": 13960
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.49134535501297e-06,
+      "loss": 0.9242,
+      "step": 13961
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.478574038579101e-06,
+      "loss": 0.8643,
+      "step": 13962
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.465811111758527e-06,
+      "loss": 0.8318,
+      "step": 13963
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.453056575113373e-06,
+      "loss": 0.9078,
+      "step": 13964
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.440310429205234e-06,
+      "loss": 0.8516,
+      "step": 13965
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.427572674595375e-06,
+      "loss": 0.8098,
+      "step": 13966
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.414843311844745e-06,
+      "loss": 0.8941,
+      "step": 13967
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.402122341513831e-06,
+      "loss": 0.8093,
+      "step": 13968
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.389409764162804e-06,
+      "loss": 0.8603,
+      "step": 13969
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.376705580351508e-06,
+      "loss": 0.8444,
+      "step": 13970
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.36400979063936e-06,
+      "loss": 0.8459,
+      "step": 13971
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.351322395585405e-06,
+      "loss": 0.8386,
+      "step": 13972
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.338643395748303e-06,
+      "loss": 0.8523,
+      "step": 13973
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.325972791686454e-06,
+      "loss": 0.8726,
+      "step": 13974
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.313310583957702e-06,
+      "loss": 0.8288,
+      "step": 13975
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.300656773119731e-06,
+      "loss": 0.8487,
+      "step": 13976
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.288011359729677e-06,
+      "loss": 0.8615,
+      "step": 13977
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.275374344344445e-06,
+      "loss": 0.8677,
+      "step": 13978
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.262745727520439e-06,
+      "loss": 0.8544,
+      "step": 13979
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.25012550981379e-06,
+      "loss": 0.8436,
+      "step": 13980
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.237513691780253e-06,
+      "loss": 0.892,
+      "step": 13981
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.224910273975119e-06,
+      "loss": 0.8406,
+      "step": 13982
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.212315256953497e-06,
+      "loss": 0.9011,
+      "step": 13983
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.1997286412699e-06,
+      "loss": 0.9001,
+      "step": 13984
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.187150427478618e-06,
+      "loss": 0.9267,
+      "step": 13985
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.174580616133565e-06,
+      "loss": 0.8826,
+      "step": 13986
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.162019207788187e-06,
+      "loss": 0.8021,
+      "step": 13987
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.149466202995682e-06,
+      "loss": 0.85,
+      "step": 13988
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.13692160230879e-06,
+      "loss": 0.846,
+      "step": 13989
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.124385406279955e-06,
+      "loss": 0.8198,
+      "step": 13990
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.111857615461138e-06,
+      "loss": 0.879,
+      "step": 13991
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.099338230404053e-06,
+      "loss": 0.8534,
+      "step": 13992
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.08682725165999e-06,
+      "loss": 0.8482,
+      "step": 13993
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.074324679779822e-06,
+      "loss": 0.8568,
+      "step": 13994
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.061830515314173e-06,
+      "loss": 0.8004,
+      "step": 13995
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.049344758813184e-06,
+      "loss": 0.8738,
+      "step": 13996
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.036867410826632e-06,
+      "loss": 0.8678,
+      "step": 13997
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.024398471904062e-06,
+      "loss": 0.8377,
+      "step": 13998
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.011937942594429e-06,
+      "loss": 0.7853,
+      "step": 13999
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 8.999485823446474e-06,
+      "loss": 0.8268,
+      "step": 14000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 8.987042115008536e-06,
+      "loss": 0.7962,
+      "step": 14001
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 8.97460681782858e-06,
+      "loss": 0.8374,
+      "step": 14002
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 8.962179932454185e-06,
+      "loss": 0.8462,
+      "step": 14003
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 8.949761459432538e-06,
+      "loss": 0.8561,
+      "step": 14004
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 8.937351399310557e-06,
+      "loss": 0.8898,
+      "step": 14005
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 8.92494975263467e-06,
+      "loss": 0.8812,
+      "step": 14006
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 8.912556519950954e-06,
+      "loss": 0.9128,
+      "step": 14007
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 8.900171701805215e-06,
+      "loss": 0.7957,
+      "step": 14008
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 8.887795298742796e-06,
+      "loss": 0.8583,
+      "step": 14009
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 8.875427311308659e-06,
+      "loss": 0.8581,
+      "step": 14010
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 8.86306774004746e-06,
+      "loss": 0.8272,
+      "step": 14011
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 8.850716585503471e-06,
+      "loss": 0.8877,
+      "step": 14012
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 8.838373848220482e-06,
+      "loss": 0.8899,
+      "step": 14013
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 8.826039528742147e-06,
+      "loss": 0.8685,
+      "step": 14014
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 8.813713627611497e-06,
+      "loss": 0.8991,
+      "step": 14015
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 8.801396145371343e-06,
+      "loss": 0.843,
+      "step": 14016
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 8.789087082564096e-06,
+      "loss": 0.8554,
+      "step": 14017
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 8.776786439731766e-06,
+      "loss": 0.8435,
+      "step": 14018
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 8.764494217416008e-06,
+      "loss": 0.8664,
+      "step": 14019
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 8.752210416158102e-06,
+      "loss": 0.8251,
+      "step": 14020
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 8.739935036499036e-06,
+      "loss": 0.8544,
+      "step": 14021
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.727668078979268e-06,
+      "loss": 0.8538,
+      "step": 14022
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.715409544139008e-06,
+      "loss": 0.8828,
+      "step": 14023
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.703159432518094e-06,
+      "loss": 0.8185,
+      "step": 14024
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.69091774465587e-06,
+      "loss": 0.8575,
+      "step": 14025
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.678684481091526e-06,
+      "loss": 0.7966,
+      "step": 14026
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.666459642363633e-06,
+      "loss": 0.786,
+      "step": 14027
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.654243229010584e-06,
+      "loss": 0.7973,
+      "step": 14028
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.642035241570322e-06,
+      "loss": 0.8131,
+      "step": 14029
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.629835680580401e-06,
+      "loss": 0.8627,
+      "step": 14030
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.617644546578052e-06,
+      "loss": 0.8888,
+      "step": 14031
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.605461840100072e-06,
+      "loss": 0.8365,
+      "step": 14032
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.593287561682983e-06,
+      "loss": 0.8366,
+      "step": 14033
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.581121711862827e-06,
+      "loss": 0.8431,
+      "step": 14034
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.568964291175352e-06,
+      "loss": 0.8191,
+      "step": 14035
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.55681530015595e-06,
+      "loss": 0.8536,
+      "step": 14036
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.54467473933953e-06,
+      "loss": 0.8812,
+      "step": 14037
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.53254260926073e-06,
+      "loss": 0.8826,
+      "step": 14038
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.520418910453787e-06,
+      "loss": 0.9023,
+      "step": 14039
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.508303643452608e-06,
+      "loss": 0.8561,
+      "step": 14040
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.496196808790635e-06,
+      "loss": 0.8887,
+      "step": 14041
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.484098407000995e-06,
+      "loss": 0.8775,
+      "step": 14042
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.472008438616508e-06,
+      "loss": 0.8073,
+      "step": 14043
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.459926904169436e-06,
+      "loss": 0.8241,
+      "step": 14044
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.447853804191907e-06,
+      "loss": 0.8348,
+      "step": 14045
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.435789139215499e-06,
+      "loss": 0.8675,
+      "step": 14046
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.423732909771498e-06,
+      "loss": 0.8225,
+      "step": 14047
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.411685116390811e-06,
+      "loss": 0.8685,
+      "step": 14048
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.399645759603924e-06,
+      "loss": 0.8659,
+      "step": 14049
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.387614839941016e-06,
+      "loss": 0.7788,
+      "step": 14050
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.37559235793186e-06,
+      "loss": 0.8332,
+      "step": 14051
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.363578314105903e-06,
+      "loss": 0.8927,
+      "step": 14052
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.351572708992138e-06,
+      "loss": 0.8387,
+      "step": 14053
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.339575543119215e-06,
+      "loss": 0.8401,
+      "step": 14054
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.327586817015509e-06,
+      "loss": 0.7902,
+      "step": 14055
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.315606531208887e-06,
+      "loss": 0.8276,
+      "step": 14056
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.30363468622688e-06,
+      "loss": 0.8555,
+      "step": 14057
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.291671282596713e-06,
+      "loss": 0.8923,
+      "step": 14058
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.27971632084521e-06,
+      "loss": 0.8551,
+      "step": 14059
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.267769801498747e-06,
+      "loss": 0.8479,
+      "step": 14060
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.255831725083418e-06,
+      "loss": 0.8837,
+      "step": 14061
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.243902092124955e-06,
+      "loss": 0.848,
+      "step": 14062
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.231980903148584e-06,
+      "loss": 0.8448,
+      "step": 14063
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.220068158679351e-06,
+      "loss": 0.8265,
+      "step": 14064
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.208163859241791e-06,
+      "loss": 0.8753,
+      "step": 14065
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.196268005360108e-06,
+      "loss": 0.825,
+      "step": 14066
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.184380597558172e-06,
+      "loss": 0.8156,
+      "step": 14067
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.172501636359386e-06,
+      "loss": 0.8737,
+      "step": 14068
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.160631122286866e-06,
+      "loss": 0.872,
+      "step": 14069
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.148769055863348e-06,
+      "loss": 0.7898,
+      "step": 14070
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.136915437611193e-06,
+      "loss": 0.94,
+      "step": 14071
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.125070268052315e-06,
+      "loss": 0.8396,
+      "step": 14072
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.113233547708344e-06,
+      "loss": 0.8927,
+      "step": 14073
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.10140527710055e-06,
+      "loss": 0.867,
+      "step": 14074
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.089585456749693e-06,
+      "loss": 0.922,
+      "step": 14075
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.077774087176382e-06,
+      "loss": 0.922,
+      "step": 14076
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.065971168900644e-06,
+      "loss": 0.8775,
+      "step": 14077
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.05417670244224e-06,
+      "loss": 0.8512,
+      "step": 14078
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.042390688320578e-06,
+      "loss": 0.8178,
+      "step": 14079
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.03061312705462e-06,
+      "loss": 0.8452,
+      "step": 14080
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.018844019162974e-06,
+      "loss": 0.749,
+      "step": 14081
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.007083365163915e-06,
+      "loss": 0.8462,
+      "step": 14082
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.995331165575337e-06,
+      "loss": 0.8018,
+      "step": 14083
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.983587420914717e-06,
+      "loss": 0.8611,
+      "step": 14084
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.97185213169922e-06,
+      "loss": 0.8336,
+      "step": 14085
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.960125298445608e-06,
+      "loss": 0.9462,
+      "step": 14086
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.948406921670248e-06,
+      "loss": 0.8489,
+      "step": 14087
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.936697001889171e-06,
+      "loss": 0.8483,
+      "step": 14088
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.92499553961803e-06,
+      "loss": 0.8911,
+      "step": 14089
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.913302535372102e-06,
+      "loss": 0.8331,
+      "step": 14090
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.901617989666266e-06,
+      "loss": 0.8611,
+      "step": 14091
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.889941903015063e-06,
+      "loss": 0.8129,
+      "step": 14092
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.878274275932662e-06,
+      "loss": 0.8281,
+      "step": 14093
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.866615108932807e-06,
+      "loss": 0.9319,
+      "step": 14094
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.854964402528996e-06,
+      "loss": 0.7879,
+      "step": 14095
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.843322157234179e-06,
+      "loss": 0.8996,
+      "step": 14096
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.831688373561053e-06,
+      "loss": 0.8273,
+      "step": 14097
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.820063052021942e-06,
+      "loss": 0.8617,
+      "step": 14098
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.808446193128705e-06,
+      "loss": 0.8711,
+      "step": 14099
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.796837797392931e-06,
+      "loss": 0.8943,
+      "step": 14100
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.78523786532579e-06,
+      "loss": 0.8705,
+      "step": 14101
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.773646397438117e-06,
+      "loss": 0.8982,
+      "step": 14102
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.762063394240283e-06,
+      "loss": 0.9265,
+      "step": 14103
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.750488856242366e-06,
+      "loss": 0.7902,
+      "step": 14104
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.738922783954095e-06,
+      "loss": 0.8043,
+      "step": 14105
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.727365177884682e-06,
+      "loss": 0.8223,
+      "step": 14106
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.71581603854319e-06,
+      "loss": 0.8391,
+      "step": 14107
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.704275366438096e-06,
+      "loss": 0.8333,
+      "step": 14108
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.692743162077621e-06,
+      "loss": 0.844,
+      "step": 14109
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.681219425969598e-06,
+      "loss": 0.8597,
+      "step": 14110
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.669704158621449e-06,
+      "loss": 0.8088,
+      "step": 14111
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.658197360540298e-06,
+      "loss": 0.8569,
+      "step": 14112
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.646699032232763e-06,
+      "loss": 0.8324,
+      "step": 14113
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.63520917420526e-06,
+      "loss": 0.8284,
+      "step": 14114
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.623727786963697e-06,
+      "loss": 0.8505,
+      "step": 14115
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.612254871013669e-06,
+      "loss": 0.9106,
+      "step": 14116
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.600790426860394e-06,
+      "loss": 0.8188,
+      "step": 14117
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.58933445500869e-06,
+      "loss": 0.8448,
+      "step": 14118
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.577886955963043e-06,
+      "loss": 0.885,
+      "step": 14119
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.566447930227516e-06,
+      "loss": 0.8561,
+      "step": 14120
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.555017378305862e-06,
+      "loss": 0.8714,
+      "step": 14121
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.543595300701389e-06,
+      "loss": 0.8225,
+      "step": 14122
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.532181697917074e-06,
+      "loss": 0.7965,
+      "step": 14123
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.520776570455557e-06,
+      "loss": 0.8154,
+      "step": 14124
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.509379918818993e-06,
+      "loss": 0.813,
+      "step": 14125
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.4979917435093136e-06,
+      "loss": 0.8158,
+      "step": 14126
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.486612045027941e-06,
+      "loss": 0.8476,
+      "step": 14127
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.475240823875984e-06,
+      "loss": 0.9048,
+      "step": 14128
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.463878080554221e-06,
+      "loss": 0.9197,
+      "step": 14129
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.45252381556294e-06,
+      "loss": 0.7637,
+      "step": 14130
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.441178029402163e-06,
+      "loss": 0.8215,
+      "step": 14131
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.429840722571513e-06,
+      "loss": 0.779,
+      "step": 14132
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.418511895570235e-06,
+      "loss": 0.9212,
+      "step": 14133
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.407191548897152e-06,
+      "loss": 0.8654,
+      "step": 14134
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.395879683050777e-06,
+      "loss": 0.8318,
+      "step": 14135
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.384576298529245e-06,
+      "loss": 0.8334,
+      "step": 14136
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.373281395830289e-06,
+      "loss": 0.902,
+      "step": 14137
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.3619949754512476e-06,
+      "loss": 0.9299,
+      "step": 14138
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.350717037889166e-06,
+      "loss": 0.8891,
+      "step": 14139
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.3394475836406685e-06,
+      "loss": 0.8397,
+      "step": 14140
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.32818661320196e-06,
+      "loss": 0.901,
+      "step": 14141
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.316934127068953e-06,
+      "loss": 0.8749,
+      "step": 14142
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.305690125737163e-06,
+      "loss": 0.8438,
+      "step": 14143
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.294454609701662e-06,
+      "loss": 0.8364,
+      "step": 14144
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.283227579457274e-06,
+      "loss": 0.8949,
+      "step": 14145
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.27200903549834e-06,
+      "loss": 0.8599,
+      "step": 14146
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.260798978318884e-06,
+      "loss": 0.8967,
+      "step": 14147
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.249597408412534e-06,
+      "loss": 0.8443,
+      "step": 14148
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.238404326272563e-06,
+      "loss": 0.8307,
+      "step": 14149
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.227219732391821e-06,
+      "loss": 0.8039,
+      "step": 14150
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.21604362726287e-06,
+      "loss": 0.879,
+      "step": 14151
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.204876011377848e-06,
+      "loss": 0.8369,
+      "step": 14152
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.193716885228474e-06,
+      "loss": 0.8654,
+      "step": 14153
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.182566249306177e-06,
+      "loss": 0.9213,
+      "step": 14154
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.171424104101987e-06,
+      "loss": 0.8473,
+      "step": 14155
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.160290450106488e-06,
+      "loss": 0.8159,
+      "step": 14156
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.149165287810045e-06,
+      "loss": 0.8647,
+      "step": 14157
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.1380486177024645e-06,
+      "loss": 0.8288,
+      "step": 14158
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.126940440273311e-06,
+      "loss": 0.853,
+      "step": 14159
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.115840756011749e-06,
+      "loss": 0.9262,
+      "step": 14160
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.104749565406521e-06,
+      "loss": 0.8943,
+      "step": 14161
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.093666868946036e-06,
+      "loss": 0.8333,
+      "step": 14162
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.082592667118327e-06,
+      "loss": 0.8729,
+      "step": 14163
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.071526960411046e-06,
+      "loss": 0.8537,
+      "step": 14164
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.060469749311472e-06,
+      "loss": 0.8732,
+      "step": 14165
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.049421034306481e-06,
+      "loss": 0.8013,
+      "step": 14166
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.0383808158826835e-06,
+      "loss": 0.8241,
+      "step": 14167
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.027349094526136e-06,
+      "loss": 0.8568,
+      "step": 14168
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.016325870722673e-06,
+      "loss": 0.8336,
+      "step": 14169
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.005311144957705e-06,
+      "loss": 0.8378,
+      "step": 14170
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.9943049177162655e-06,
+      "loss": 0.8041,
+      "step": 14171
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.983307189482968e-06,
+      "loss": 0.8247,
+      "step": 14172
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.9723179607421585e-06,
+      "loss": 0.8362,
+      "step": 14173
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.961337231977738e-06,
+      "loss": 0.8716,
+      "step": 14174
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.950365003673187e-06,
+      "loss": 0.8861,
+      "step": 14175
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.939401276311741e-06,
+      "loss": 0.882,
+      "step": 14176
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.928446050376147e-06,
+      "loss": 0.8137,
+      "step": 14177
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.91749932634882e-06,
+      "loss": 0.8411,
+      "step": 14178
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.906561104711817e-06,
+      "loss": 0.8427,
+      "step": 14179
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.895631385946799e-06,
+      "loss": 0.8558,
+      "step": 14180
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.884710170535025e-06,
+      "loss": 0.8179,
+      "step": 14181
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.873797458957443e-06,
+      "loss": 0.778,
+      "step": 14182
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.862893251694602e-06,
+      "loss": 0.8445,
+      "step": 14183
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.851997549226652e-06,
+      "loss": 0.8232,
+      "step": 14184
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.8411103520333645e-06,
+      "loss": 0.8401,
+      "step": 14185
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.830231660594222e-06,
+      "loss": 0.8889,
+      "step": 14186
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.819361475388153e-06,
+      "loss": 0.8588,
+      "step": 14187
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.8084997968939745e-06,
+      "loss": 0.903,
+      "step": 14188
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.797646625589859e-06,
+      "loss": 0.8868,
+      "step": 14189
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.786801961953804e-06,
+      "loss": 0.9278,
+      "step": 14190
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.775965806463291e-06,
+      "loss": 0.9257,
+      "step": 14191
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.765138159595541e-06,
+      "loss": 0.8981,
+      "step": 14192
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.754319021827349e-06,
+      "loss": 0.7945,
+      "step": 14193
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.743508393635045e-06,
+      "loss": 0.8772,
+      "step": 14194
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.732706275494826e-06,
+      "loss": 0.8872,
+      "step": 14195
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.721912667882246e-06,
+      "loss": 0.8478,
+      "step": 14196
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.7111275712726355e-06,
+      "loss": 0.8668,
+      "step": 14197
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.700350986140946e-06,
+      "loss": 0.8678,
+      "step": 14198
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.689582912961689e-06,
+      "loss": 0.8047,
+      "step": 14199
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.6788233522090404e-06,
+      "loss": 0.8629,
+      "step": 14200
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.668072304356798e-06,
+      "loss": 0.8055,
+      "step": 14201
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.657329769878429e-06,
+      "loss": 0.8714,
+      "step": 14202
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.646595749246909e-06,
+      "loss": 0.8537,
+      "step": 14203
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.63587024293495e-06,
+      "loss": 0.8818,
+      "step": 14204
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.625153251414862e-06,
+      "loss": 0.9302,
+      "step": 14205
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.614444775158513e-06,
+      "loss": 0.8573,
+      "step": 14206
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.6037448146375246e-06,
+      "loss": 0.81,
+      "step": 14207
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.593053370323011e-06,
+      "loss": 0.8098,
+      "step": 14208
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.582370442685793e-06,
+      "loss": 0.8985,
+      "step": 14209
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.571696032196317e-06,
+      "loss": 0.8199,
+      "step": 14210
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.561030139324587e-06,
+      "loss": 0.8034,
+      "step": 14211
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.550372764540291e-06,
+      "loss": 0.8575,
+      "step": 14212
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.539723908312723e-06,
+      "loss": 0.7713,
+      "step": 14213
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.529083571110839e-06,
+      "loss": 0.8767,
+      "step": 14214
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.518451753403154e-06,
+      "loss": 0.838,
+      "step": 14215
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.507828455657827e-06,
+      "loss": 0.81,
+      "step": 14216
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.497213678342728e-06,
+      "loss": 0.8018,
+      "step": 14217
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.486607421925173e-06,
+      "loss": 0.8101,
+      "step": 14218
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.4760096868722755e-06,
+      "loss": 0.8579,
+      "step": 14219
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.465420473650685e-06,
+      "loss": 0.8542,
+      "step": 14220
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.45483978272674e-06,
+      "loss": 0.8943,
+      "step": 14221
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.4442676145662906e-06,
+      "loss": 0.8089,
+      "step": 14222
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.433703969634941e-06,
+      "loss": 0.881,
+      "step": 14223
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.423148848397831e-06,
+      "loss": 0.8609,
+      "step": 14224
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.412602251319744e-06,
+      "loss": 0.9178,
+      "step": 14225
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.402064178865152e-06,
+      "loss": 0.8567,
+      "step": 14226
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.3915346314980415e-06,
+      "loss": 0.7986,
+      "step": 14227
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.381013609682107e-06,
+      "loss": 0.8295,
+      "step": 14228
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.370501113880667e-06,
+      "loss": 0.8311,
+      "step": 14229
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.359997144556595e-06,
+      "loss": 0.8198,
+      "step": 14230
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.3495017021724555e-06,
+      "loss": 0.8567,
+      "step": 14231
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.3390147871904115e-06,
+      "loss": 0.8382,
+      "step": 14232
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.328536400072271e-06,
+      "loss": 0.8536,
+      "step": 14233
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.318066541279399e-06,
+      "loss": 0.9253,
+      "step": 14234
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.307605211272894e-06,
+      "loss": 0.8563,
+      "step": 14235
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.297152410513407e-06,
+      "loss": 0.8603,
+      "step": 14236
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.2867081394611725e-06,
+      "loss": 0.8213,
+      "step": 14237
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.276272398576178e-06,
+      "loss": 0.8922,
+      "step": 14238
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.2658451883179425e-06,
+      "loss": 0.8612,
+      "step": 14239
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.25542650914559e-06,
+      "loss": 0.9261,
+      "step": 14240
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.245016361517975e-06,
+      "loss": 0.9393,
+      "step": 14241
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.234614745893441e-06,
+      "loss": 0.8823,
+      "step": 14242
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.2242216627300455e-06,
+      "loss": 0.8416,
+      "step": 14243
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.2138371124854656e-06,
+      "loss": 0.8316,
+      "step": 14244
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.2034610956170025e-06,
+      "loss": 0.9207,
+      "step": 14245
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.193093612581491e-06,
+      "loss": 0.7961,
+      "step": 14246
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.182734663835521e-06,
+      "loss": 0.8904,
+      "step": 14247
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.172384249835261e-06,
+      "loss": 0.8314,
+      "step": 14248
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.162042371036436e-06,
+      "loss": 0.831,
+      "step": 14249
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.151709027894481e-06,
+      "loss": 0.8487,
+      "step": 14250
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.14138422086441e-06,
+      "loss": 0.8062,
+      "step": 14251
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.131067950400926e-06,
+      "loss": 0.8918,
+      "step": 14252
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.120760216958266e-06,
+      "loss": 0.8715,
+      "step": 14253
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.11046102099031e-06,
+      "loss": 0.8597,
+      "step": 14254
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.1001703629506525e-06,
+      "loss": 0.7865,
+      "step": 14255
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.08988824329233e-06,
+      "loss": 0.8671,
+      "step": 14256
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.079614662468247e-06,
+      "loss": 0.8487,
+      "step": 14257
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.0693496209307084e-06,
+      "loss": 0.8563,
+      "step": 14258
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.059093119131776e-06,
+      "loss": 0.8499,
+      "step": 14259
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.048845157523109e-06,
+      "loss": 0.8045,
+      "step": 14260
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.038605736555924e-06,
+      "loss": 0.9005,
+      "step": 14261
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.02837485668113e-06,
+      "loss": 0.8475,
+      "step": 14262
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.018152518349273e-06,
+      "loss": 0.9104,
+      "step": 14263
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.0079387220104864e-06,
+      "loss": 0.8583,
+      "step": 14264
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.997733468114519e-06,
+      "loss": 0.8665,
+      "step": 14265
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.987536757110745e-06,
+      "loss": 0.8078,
+      "step": 14266
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.977348589448251e-06,
+      "loss": 0.8735,
+      "step": 14267
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.967168965575542e-06,
+      "loss": 0.8259,
+      "step": 14268
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.956997885941018e-06,
+      "loss": 0.8861,
+      "step": 14269
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.9468353509924745e-06,
+      "loss": 0.9054,
+      "step": 14270
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.936681361177465e-06,
+      "loss": 0.8491,
+      "step": 14271
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.9265359169430765e-06,
+      "loss": 0.8076,
+      "step": 14272
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.9163990187361075e-06,
+      "loss": 0.8768,
+      "step": 14273
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.906270667002911e-06,
+      "loss": 0.8511,
+      "step": 14274
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.8961508621894646e-06,
+      "loss": 0.8822,
+      "step": 14275
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.8860396047414555e-06,
+      "loss": 0.84,
+      "step": 14276
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.875936895104084e-06,
+      "loss": 0.8333,
+      "step": 14277
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.86584273372226e-06,
+      "loss": 0.8585,
+      "step": 14278
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.855757121040451e-06,
+      "loss": 0.8057,
+      "step": 14279
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.845680057502789e-06,
+      "loss": 0.8716,
+      "step": 14280
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.83561154355301e-06,
+      "loss": 0.8378,
+      "step": 14281
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.82555157963447e-06,
+      "loss": 0.8571,
+      "step": 14282
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.815500166190213e-06,
+      "loss": 0.9357,
+      "step": 14283
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.805457303662798e-06,
+      "loss": 0.8578,
+      "step": 14284
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.79542299249447e-06,
+      "loss": 0.8647,
+      "step": 14285
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.785397233127121e-06,
+      "loss": 0.8056,
+      "step": 14286
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.775380026002175e-06,
+      "loss": 0.8742,
+      "step": 14287
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.765371371560835e-06,
+      "loss": 0.849,
+      "step": 14288
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.755371270243748e-06,
+      "loss": 0.8048,
+      "step": 14289
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.745379722491295e-06,
+      "loss": 0.8635,
+      "step": 14290
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.735396728743481e-06,
+      "loss": 0.8634,
+      "step": 14291
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.7254222894398635e-06,
+      "loss": 0.8655,
+      "step": 14292
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.7154564050196705e-06,
+      "loss": 0.8398,
+      "step": 14293
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.705499075921772e-06,
+      "loss": 0.9022,
+      "step": 14294
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.6955503025846625e-06,
+      "loss": 0.8676,
+      "step": 14295
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.685610085446369e-06,
+      "loss": 0.9267,
+      "step": 14296
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.675678424944675e-06,
+      "loss": 0.8565,
+      "step": 14297
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.665755321516897e-06,
+      "loss": 0.8478,
+      "step": 14298
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.655840775599974e-06,
+      "loss": 0.8674,
+      "step": 14299
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.645934787630513e-06,
+      "loss": 0.8014,
+      "step": 14300
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.636037358044744e-06,
+      "loss": 0.9476,
+      "step": 14301
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.626148487278471e-06,
+      "loss": 0.8293,
+      "step": 14302
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.61626817576717e-06,
+      "loss": 0.8807,
+      "step": 14303
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.606396423945892e-06,
+      "loss": 0.8388,
+      "step": 14304
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.596533232249379e-06,
+      "loss": 0.8473,
+      "step": 14305
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.586678601111906e-06,
+      "loss": 0.857,
+      "step": 14306
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.576832530967502e-06,
+      "loss": 0.8449,
+      "step": 14307
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.566995022249666e-06,
+      "loss": 0.8679,
+      "step": 14308
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.557166075391607e-06,
+      "loss": 0.7767,
+      "step": 14309
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.547345690826178e-06,
+      "loss": 0.8822,
+      "step": 14310
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.537533868985789e-06,
+      "loss": 0.8526,
+      "step": 14311
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.527730610302495e-06,
+      "loss": 0.8208,
+      "step": 14312
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.517935915207995e-06,
+      "loss": 0.8472,
+      "step": 14313
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.508149784133609e-06,
+      "loss": 0.8535,
+      "step": 14314
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.498372217510239e-06,
+      "loss": 0.8688,
+      "step": 14315
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.488603215768495e-06,
+      "loss": 0.8565,
+      "step": 14316
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.478842779338544e-06,
+      "loss": 0.9041,
+      "step": 14317
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.469090908650087e-06,
+      "loss": 0.7843,
+      "step": 14318
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.459347604132692e-06,
+      "loss": 0.793,
+      "step": 14319
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.449612866215325e-06,
+      "loss": 0.9114,
+      "step": 14320
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.4398866953266455e-06,
+      "loss": 0.8327,
+      "step": 14321
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.430169091894999e-06,
+      "loss": 0.8686,
+      "step": 14322
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.420460056348264e-06,
+      "loss": 0.8008,
+      "step": 14323
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.410759589113989e-06,
+      "loss": 0.7485,
+      "step": 14324
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.401067690619299e-06,
+      "loss": 0.9082,
+      "step": 14325
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.391384361291052e-06,
+      "loss": 0.7937,
+      "step": 14326
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.381709601555573e-06,
+      "loss": 0.8473,
+      "step": 14327
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.372043411838945e-06,
+      "loss": 0.794,
+      "step": 14328
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.362385792566804e-06,
+      "loss": 0.8294,
+      "step": 14329
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.352736744164411e-06,
+      "loss": 0.8545,
+      "step": 14330
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.343096267056669e-06,
+      "loss": 0.8201,
+      "step": 14331
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.333464361668105e-06,
+      "loss": 0.8729,
+      "step": 14332
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.323841028422871e-06,
+      "loss": 0.8282,
+      "step": 14333
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.314226267744671e-06,
+      "loss": 0.7896,
+      "step": 14334
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.3046200800569655e-06,
+      "loss": 0.849,
+      "step": 14335
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.2950224657827285e-06,
+      "loss": 0.8868,
+      "step": 14336
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.285433425344577e-06,
+      "loss": 0.8374,
+      "step": 14337
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.275852959164795e-06,
+      "loss": 0.8507,
+      "step": 14338
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.266281067665246e-06,
+      "loss": 0.8846,
+      "step": 14339
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.256717751267415e-06,
+      "loss": 0.8773,
+      "step": 14340
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.247163010392453e-06,
+      "loss": 0.8251,
+      "step": 14341
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.2376168454610685e-06,
+      "loss": 0.8286,
+      "step": 14342
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.228079256893637e-06,
+      "loss": 0.875,
+      "step": 14343
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.218550245110154e-06,
+      "loss": 0.819,
+      "step": 14344
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.209029810530264e-06,
+      "loss": 0.8406,
+      "step": 14345
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.199517953573119e-06,
+      "loss": 0.8384,
+      "step": 14346
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.190014674657651e-06,
+      "loss": 0.8383,
+      "step": 14347
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.180519974202302e-06,
+      "loss": 0.8315,
+      "step": 14348
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.17103385262514e-06,
+      "loss": 0.783,
+      "step": 14349
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.16155631034394e-06,
+      "loss": 0.9747,
+      "step": 14350
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.152087347776036e-06,
+      "loss": 0.8341,
+      "step": 14351
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.142626965338382e-06,
+      "loss": 0.8619,
+      "step": 14352
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.133175163447534e-06,
+      "loss": 0.8478,
+      "step": 14353
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.123731942519782e-06,
+      "loss": 0.8897,
+      "step": 14354
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.114297302970905e-06,
+      "loss": 0.8603,
+      "step": 14355
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.104871245216325e-06,
+      "loss": 0.8644,
+      "step": 14356
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.095453769671199e-06,
+      "loss": 0.7844,
+      "step": 14357
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.086044876750174e-06,
+      "loss": 0.8857,
+      "step": 14358
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.0766445668675836e-06,
+      "loss": 0.7821,
+      "step": 14359
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.067252840437364e-06,
+      "loss": 0.7982,
+      "step": 14360
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.057869697873097e-06,
+      "loss": 0.8428,
+      "step": 14361
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.0484951395879386e-06,
+      "loss": 0.8956,
+      "step": 14362
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.039129165994716e-06,
+      "loss": 0.8881,
+      "step": 14363
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.029771777505876e-06,
+      "loss": 0.8068,
+      "step": 14364
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.020422974533423e-06,
+      "loss": 0.8636,
+      "step": 14365
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.011082757489094e-06,
+      "loss": 0.8901,
+      "step": 14366
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.00175112678416e-06,
+      "loss": 0.8087,
+      "step": 14367
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.9924280828295145e-06,
+      "loss": 0.8304,
+      "step": 14368
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.98311362603574e-06,
+      "loss": 0.8274,
+      "step": 14369
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.973807756812954e-06,
+      "loss": 0.9371,
+      "step": 14370
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.964510475570983e-06,
+      "loss": 0.781,
+      "step": 14371
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.955221782719189e-06,
+      "loss": 0.8525,
+      "step": 14372
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.945941678666666e-06,
+      "loss": 0.7888,
+      "step": 14373
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.936670163821977e-06,
+      "loss": 0.7812,
+      "step": 14374
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.9274072385934624e-06,
+      "loss": 0.7957,
+      "step": 14375
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.918152903388995e-06,
+      "loss": 0.8591,
+      "step": 14376
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.908907158616049e-06,
+      "loss": 0.8255,
+      "step": 14377
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.8996700046818335e-06,
+      "loss": 0.8818,
+      "step": 14378
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.890441441993087e-06,
+      "loss": 0.8044,
+      "step": 14379
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.881221470956132e-06,
+      "loss": 0.8389,
+      "step": 14380
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.87201009197702e-06,
+      "loss": 0.8925,
+      "step": 14381
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.862807305461381e-06,
+      "loss": 0.8186,
+      "step": 14382
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.853613111814448e-06,
+      "loss": 0.8811,
+      "step": 14383
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.84442751144103e-06,
+      "loss": 0.8272,
+      "step": 14384
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.835250504745736e-06,
+      "loss": 0.8285,
+      "step": 14385
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.8260820921326e-06,
+      "loss": 0.9711,
+      "step": 14386
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.81692227400532e-06,
+      "loss": 0.8225,
+      "step": 14387
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.8077710507673066e-06,
+      "loss": 0.8324,
+      "step": 14388
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.798628422821527e-06,
+      "loss": 0.9024,
+      "step": 14389
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.789494390570548e-06,
+      "loss": 0.8441,
+      "step": 14390
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.780368954416625e-06,
+      "loss": 0.9088,
+      "step": 14391
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.771252114761593e-06,
+      "loss": 0.7831,
+      "step": 14392
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.762143872006863e-06,
+      "loss": 0.8363,
+      "step": 14393
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.75304422655356e-06,
+      "loss": 0.9104,
+      "step": 14394
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.743953178802407e-06,
+      "loss": 0.8608,
+      "step": 14395
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.73487072915364e-06,
+      "loss": 0.8561,
+      "step": 14396
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.7257968780073154e-06,
+      "loss": 0.8699,
+      "step": 14397
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.716731625762938e-06,
+      "loss": 0.8263,
+      "step": 14398
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.707674972819676e-06,
+      "loss": 0.8407,
+      "step": 14399
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.69862691957641e-06,
+      "loss": 0.8494,
+      "step": 14400
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.689587466431511e-06,
+      "loss": 0.8418,
+      "step": 14401
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.6805566137830605e-06,
+      "loss": 0.8984,
+      "step": 14402
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.671534362028696e-06,
+      "loss": 0.8131,
+      "step": 14403
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.662520711565788e-06,
+      "loss": 0.8268,
+      "step": 14404
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.653515662791197e-06,
+      "loss": 0.844,
+      "step": 14405
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.644519216101406e-06,
+      "loss": 0.866,
+      "step": 14406
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.6355313718926985e-06,
+      "loss": 0.8903,
+      "step": 14407
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.6265521305607795e-06,
+      "loss": 0.8265,
+      "step": 14408
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.617581492501044e-06,
+      "loss": 0.8073,
+      "step": 14409
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.6086194581085315e-06,
+      "loss": 0.793,
+      "step": 14410
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.599666027777927e-06,
+      "loss": 0.8845,
+      "step": 14411
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.590721201903403e-06,
+      "loss": 0.8746,
+      "step": 14412
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.581784980878911e-06,
+      "loss": 0.8112,
+      "step": 14413
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.572857365097982e-06,
+      "loss": 0.8557,
+      "step": 14414
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.563938354953634e-06,
+      "loss": 0.8189,
+      "step": 14415
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.555027950838753e-06,
+      "loss": 0.8375,
+      "step": 14416
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.546126153145602e-06,
+      "loss": 0.854,
+      "step": 14417
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.537232962266202e-06,
+      "loss": 0.8954,
+      "step": 14418
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.528348378592218e-06,
+      "loss": 0.9087,
+      "step": 14419
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.519472402514801e-06,
+      "loss": 0.8473,
+      "step": 14420
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.510605034424864e-06,
+      "loss": 0.8228,
+      "step": 14421
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.501746274712848e-06,
+      "loss": 0.8626,
+      "step": 14422
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.492896123768864e-06,
+      "loss": 0.8213,
+      "step": 14423
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.484054581982622e-06,
+      "loss": 0.8358,
+      "step": 14424
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.475221649743477e-06,
+      "loss": 0.8086,
+      "step": 14425
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.466397327440363e-06,
+      "loss": 0.8988,
+      "step": 14426
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.4575816154618585e-06,
+      "loss": 0.8069,
+      "step": 14427
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.448774514196186e-06,
+      "loss": 0.8402,
+      "step": 14428
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.439976024031145e-06,
+      "loss": 0.8804,
+      "step": 14429
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.431186145354205e-06,
+      "loss": 0.8696,
+      "step": 14430
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.422404878552411e-06,
+      "loss": 0.8878,
+      "step": 14431
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.4136322240124315e-06,
+      "loss": 0.8307,
+      "step": 14432
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.404868182120603e-06,
+      "loss": 0.827,
+      "step": 14433
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.396112753262793e-06,
+      "loss": 0.8431,
+      "step": 14434
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.387365937824605e-06,
+      "loss": 0.8086,
+      "step": 14435
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.378627736191199e-06,
+      "loss": 0.8947,
+      "step": 14436
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.369898148747331e-06,
+      "loss": 0.8391,
+      "step": 14437
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.361177175877451e-06,
+      "loss": 0.8618,
+      "step": 14438
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.3524648179655405e-06,
+      "loss": 0.8416,
+      "step": 14439
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.34376107539527e-06,
+      "loss": 0.8684,
+      "step": 14440
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.335065948549888e-06,
+      "loss": 0.8587,
+      "step": 14441
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.326379437812356e-06,
+      "loss": 0.8845,
+      "step": 14442
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.3177015435651e-06,
+      "loss": 0.838,
+      "step": 14443
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.309032266190282e-06,
+      "loss": 0.8377,
+      "step": 14444
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.300371606069686e-06,
+      "loss": 0.8319,
+      "step": 14445
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.291719563584607e-06,
+      "loss": 0.856,
+      "step": 14446
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.283076139116138e-06,
+      "loss": 0.7626,
+      "step": 14447
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.274441333044799e-06,
+      "loss": 0.8193,
+      "step": 14448
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.265815145750885e-06,
+      "loss": 0.8864,
+      "step": 14449
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.257197577614225e-06,
+      "loss": 0.8997,
+      "step": 14450
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.248588629014294e-06,
+      "loss": 0.8361,
+      "step": 14451
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.239988300330211e-06,
+      "loss": 0.8924,
+      "step": 14452
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.23139659194065e-06,
+      "loss": 0.8048,
+      "step": 14453
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.222813504223999e-06,
+      "loss": 0.8551,
+      "step": 14454
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.214239037558154e-06,
+      "loss": 0.8342,
+      "step": 14455
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.205673192320725e-06,
+      "loss": 0.8467,
+      "step": 14456
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.197115968888921e-06,
+      "loss": 0.8043,
+      "step": 14457
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.188567367639551e-06,
+      "loss": 0.9077,
+      "step": 14458
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.180027388949026e-06,
+      "loss": 0.8487,
+      "step": 14459
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.171496033193423e-06,
+      "loss": 0.8761,
+      "step": 14460
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.162973300748463e-06,
+      "loss": 0.9411,
+      "step": 14461
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.154459191989357e-06,
+      "loss": 0.8366,
+      "step": 14462
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.145953707291094e-06,
+      "loss": 0.898,
+      "step": 14463
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.137456847028198e-06,
+      "loss": 0.9154,
+      "step": 14464
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.12896861157479e-06,
+      "loss": 0.8307,
+      "step": 14465
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.120489001304706e-06,
+      "loss": 0.8728,
+      "step": 14466
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.112018016591313e-06,
+      "loss": 0.954,
+      "step": 14467
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.103555657807623e-06,
+      "loss": 0.8098,
+      "step": 14468
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.095101925326317e-06,
+      "loss": 0.8867,
+      "step": 14469
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.086656819519608e-06,
+      "loss": 0.7727,
+      "step": 14470
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.078220340759398e-06,
+      "loss": 0.8371,
+      "step": 14471
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.0697924894171906e-06,
+      "loss": 0.9277,
+      "step": 14472
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.0613732658641105e-06,
+      "loss": 0.8484,
+      "step": 14473
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.052962670470884e-06,
+      "loss": 0.823,
+      "step": 14474
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.0445607036078805e-06,
+      "loss": 0.9557,
+      "step": 14475
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.036167365645094e-06,
+      "loss": 0.8843,
+      "step": 14476
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.027782656952073e-06,
+      "loss": 0.8124,
+      "step": 14477
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.019406577898099e-06,
+      "loss": 0.8473,
+      "step": 14478
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.011039128851968e-06,
+      "loss": 0.8648,
+      "step": 14479
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.002680310182183e-06,
+      "loss": 0.8289,
+      "step": 14480
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.994330122256807e-06,
+      "loss": 0.8257,
+      "step": 14481
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.985988565443521e-06,
+      "loss": 0.8273,
+      "step": 14482
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.977655640109656e-06,
+      "loss": 0.7974,
+      "step": 14483
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.969331346622163e-06,
+      "loss": 0.8221,
+      "step": 14484
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.961015685347613e-06,
+      "loss": 0.899,
+      "step": 14485
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.95270865665216e-06,
+      "loss": 0.8778,
+      "step": 14486
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.944410260901599e-06,
+      "loss": 0.8976,
+      "step": 14487
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.936120498461393e-06,
+      "loss": 0.8527,
+      "step": 14488
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.92783936969654e-06,
+      "loss": 0.8623,
+      "step": 14489
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.919566874971703e-06,
+      "loss": 0.8596,
+      "step": 14490
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.911303014651191e-06,
+      "loss": 0.898,
+      "step": 14491
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.9030477890988904e-06,
+      "loss": 0.8216,
+      "step": 14492
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.894801198678289e-06,
+      "loss": 0.8464,
+      "step": 14493
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.886563243752562e-06,
+      "loss": 0.8213,
+      "step": 14494
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.878333924684463e-06,
+      "loss": 0.8594,
+      "step": 14495
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.870113241836326e-06,
+      "loss": 0.7765,
+      "step": 14496
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.861901195570239e-06,
+      "loss": 0.8501,
+      "step": 14497
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.8536977862477345e-06,
+      "loss": 0.7696,
+      "step": 14498
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.845503014230079e-06,
+      "loss": 0.8595,
+      "step": 14499
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.837316879878161e-06,
+      "loss": 0.8472,
+      "step": 14500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.829139383552405e-06,
+      "loss": 0.8741,
+      "step": 14501
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.820970525612944e-06,
+      "loss": 0.901,
+      "step": 14502
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.8128103064194677e-06,
+      "loss": 0.8475,
+      "step": 14503
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.8046587263313336e-06,
+      "loss": 0.8502,
+      "step": 14504
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.7965157857074775e-06,
+      "loss": 0.7894,
+      "step": 14505
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.7883814849065004e-06,
+      "loss": 0.8947,
+      "step": 14506
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.7802558242865825e-06,
+      "loss": 0.8266,
+      "step": 14507
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.772138804205483e-06,
+      "loss": 0.8208,
+      "step": 14508
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.764030425020737e-06,
+      "loss": 0.8279,
+      "step": 14509
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.7559306870893263e-06,
+      "loss": 0.8556,
+      "step": 14510
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.747839590767965e-06,
+      "loss": 0.8663,
+      "step": 14511
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.739757136412925e-06,
+      "loss": 0.767,
+      "step": 14512
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.7316833243800754e-06,
+      "loss": 0.8332,
+      "step": 14513
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.7236181550250215e-06,
+      "loss": 0.809,
+      "step": 14514
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.715561628702857e-06,
+      "loss": 0.8001,
+      "step": 14515
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.7075137457683875e-06,
+      "loss": 0.8227,
+      "step": 14516
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.6994745065759726e-06,
+      "loss": 0.9002,
+      "step": 14517
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.691443911479642e-06,
+      "loss": 0.8555,
+      "step": 14518
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.6834219608330224e-06,
+      "loss": 0.8206,
+      "step": 14519
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.675408654989343e-06,
+      "loss": 0.9104,
+      "step": 14520
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.667403994301477e-06,
+      "loss": 0.8625,
+      "step": 14521
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.6594079791219205e-06,
+      "loss": 0.8571,
+      "step": 14522
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.651420609802769e-06,
+      "loss": 0.8453,
+      "step": 14523
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.643441886695742e-06,
+      "loss": 0.8661,
+      "step": 14524
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.6354718101522023e-06,
+      "loss": 0.8994,
+      "step": 14525
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.6275103805230915e-06,
+      "loss": 0.8517,
+      "step": 14526
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.6195575981589734e-06,
+      "loss": 0.82,
+      "step": 14527
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.6116134634101017e-06,
+      "loss": 0.8684,
+      "step": 14528
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.6036779766262628e-06,
+      "loss": 0.8591,
+      "step": 14529
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5957511381568887e-06,
+      "loss": 0.8164,
+      "step": 14530
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5878329483510776e-06,
+      "loss": 0.8828,
+      "step": 14531
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5799234075574617e-06,
+      "loss": 0.8641,
+      "step": 14532
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5720225161243624e-06,
+      "loss": 0.8453,
+      "step": 14533
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5641302743996795e-06,
+      "loss": 0.9032,
+      "step": 14534
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5562466827309793e-06,
+      "loss": 0.9056,
+      "step": 14535
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5483717414653617e-06,
+      "loss": 0.8206,
+      "step": 14536
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5405054509496606e-06,
+      "loss": 0.8459,
+      "step": 14537
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5326478115302208e-06,
+      "loss": 0.8491,
+      "step": 14538
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.524798823553077e-06,
+      "loss": 0.8053,
+      "step": 14539
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.516958487363864e-06,
+      "loss": 0.8114,
+      "step": 14540
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5091268033078164e-06,
+      "loss": 0.8426,
+      "step": 14541
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.501303771729814e-06,
+      "loss": 0.9399,
+      "step": 14542
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.4934893929743362e-06,
+      "loss": 0.8955,
+      "step": 14543
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.4856836673854865e-06,
+      "loss": 0.9001,
+      "step": 14544
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.4778865953070117e-06,
+      "loss": 0.8743,
+      "step": 14545
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.4700981770821928e-06,
+      "loss": 0.8436,
+      "step": 14546
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.4623184130540886e-06,
+      "loss": 0.9407,
+      "step": 14547
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.4545473035652254e-06,
+      "loss": 0.8144,
+      "step": 14548
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.446784848957796e-06,
+      "loss": 0.8367,
+      "step": 14549
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.4390310495736377e-06,
+      "loss": 0.8749,
+      "step": 14550
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.431285905754189e-06,
+      "loss": 0.7915,
+      "step": 14551
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.4235494178405103e-06,
+      "loss": 0.8577,
+      "step": 14552
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.4158215861732624e-06,
+      "loss": 0.8012,
+      "step": 14553
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.4081024110927505e-06,
+      "loss": 0.8245,
+      "step": 14554
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.4003918929389033e-06,
+      "loss": 0.8378,
+      "step": 14555
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.392690032051227e-06,
+      "loss": 0.8508,
+      "step": 14556
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.384996828768894e-06,
+      "loss": 0.7752,
+      "step": 14557
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.3773122834306337e-06,
+      "loss": 0.8461,
+      "step": 14558
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.369636396374909e-06,
+      "loss": 0.8501,
+      "step": 14559
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.3619691679396715e-06,
+      "loss": 0.8667,
+      "step": 14560
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.35431059846254e-06,
+      "loss": 0.8368,
+      "step": 14561
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.3466606882808225e-06,
+      "loss": 0.8557,
+      "step": 14562
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.3390194377313166e-06,
+      "loss": 0.8549,
+      "step": 14563
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.3313868471505307e-06,
+      "loss": 0.8942,
+      "step": 14564
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.323762916874551e-06,
+      "loss": 0.9508,
+      "step": 14565
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.316147647239154e-06,
+      "loss": 0.9135,
+      "step": 14566
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.3085410385796045e-06,
+      "loss": 0.8293,
+      "step": 14567
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.3009430912309013e-06,
+      "loss": 0.8927,
+      "step": 14568
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.2933538055276216e-06,
+      "loss": 0.8527,
+      "step": 14569
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.2857731818039194e-06,
+      "loss": 0.8533,
+      "step": 14570
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.278201220393662e-06,
+      "loss": 0.823,
+      "step": 14571
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.2706379216302263e-06,
+      "loss": 0.825,
+      "step": 14572
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.263083285846724e-06,
+      "loss": 0.8834,
+      "step": 14573
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.2555373133757783e-06,
+      "loss": 0.8963,
+      "step": 14574
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.2480000045496783e-06,
+      "loss": 0.8528,
+      "step": 14575
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.2404713597003587e-06,
+      "loss": 0.8282,
+      "step": 14576
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.232951379159288e-06,
+      "loss": 0.8776,
+      "step": 14577
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.2254400632576897e-06,
+      "loss": 0.8206,
+      "step": 14578
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.217937412326255e-06,
+      "loss": 0.8392,
+      "step": 14579
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.2104434266953865e-06,
+      "loss": 0.8772,
+      "step": 14580
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.2029581066951086e-06,
+      "loss": 0.8592,
+      "step": 14581
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.1954814526549803e-06,
+      "loss": 0.8477,
+      "step": 14582
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.1880134649042936e-06,
+      "loss": 0.8212,
+      "step": 14583
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.1805541437718523e-06,
+      "loss": 0.8203,
+      "step": 14584
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.1731034895861934e-06,
+      "loss": 0.8845,
+      "step": 14585
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.165661502675321e-06,
+      "loss": 0.7717,
+      "step": 14586
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.1582281833669956e-06,
+      "loss": 0.8467,
+      "step": 14587
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.1508035319885554e-06,
+      "loss": 0.8765,
+      "step": 14588
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.143387548866894e-06,
+      "loss": 0.9051,
+      "step": 14589
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.1359802343286393e-06,
+      "loss": 0.8578,
+      "step": 14590
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.1285815886999306e-06,
+      "loss": 0.8483,
+      "step": 14591
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.1211916123065733e-06,
+      "loss": 0.8918,
+      "step": 14592
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.1138103054739744e-06,
+      "loss": 0.8755,
+      "step": 14593
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.1064376685271847e-06,
+      "loss": 0.7944,
+      "step": 14594
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.0990737017908555e-06,
+      "loss": 0.888,
+      "step": 14595
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.091718405589261e-06,
+      "loss": 0.8592,
+      "step": 14596
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.084371780246298e-06,
+      "loss": 0.8502,
+      "step": 14597
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.0770338260854627e-06,
+      "loss": 0.8752,
+      "step": 14598
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.0697045434298744e-06,
+      "loss": 0.8244,
+      "step": 14599
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.0623839326022974e-06,
+      "loss": 0.9073,
+      "step": 14600
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.055071993925096e-06,
+      "loss": 0.8964,
+      "step": 14601
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.047768727720235e-06,
+      "loss": 0.8605,
+      "step": 14602
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.0404741343093236e-06,
+      "loss": 0.826,
+      "step": 14603
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.0331882140135934e-06,
+      "loss": 0.8561,
+      "step": 14604
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.0259109671538332e-06,
+      "loss": 0.8616,
+      "step": 14605
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.0186423940505416e-06,
+      "loss": 0.8712,
+      "step": 14606
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.0113824950237955e-06,
+      "loss": 0.9083,
+      "step": 14607
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.004131270393229e-06,
+      "loss": 0.8478,
+      "step": 14608
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.9968887204782304e-06,
+      "loss": 0.9446,
+      "step": 14609
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.989654845597656e-06,
+      "loss": 0.9064,
+      "step": 14610
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.982429646070073e-06,
+      "loss": 0.8888,
+      "step": 14611
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.9752131222136715e-06,
+      "loss": 0.8105,
+      "step": 14612
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.9680052743461752e-06,
+      "loss": 0.8685,
+      "step": 14613
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.9608061027850187e-06,
+      "loss": 0.8358,
+      "step": 14614
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.9536156078472154e-06,
+      "loss": 0.8545,
+      "step": 14615
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.946433789849401e-06,
+      "loss": 0.8473,
+      "step": 14616
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.939260649107811e-06,
+      "loss": 0.8421,
+      "step": 14617
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.932096185938327e-06,
+      "loss": 0.8186,
+      "step": 14618
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.9249404006564287e-06,
+      "loss": 0.8921,
+      "step": 14619
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.9177932935772423e-06,
+      "loss": 0.8829,
+      "step": 14620
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.91065486501545e-06,
+      "loss": 0.9284,
+      "step": 14621
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.9035251152854213e-06,
+      "loss": 0.889,
+      "step": 14622
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.8964040447011287e-06,
+      "loss": 0.835,
+      "step": 14623
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.889291653576098e-06,
+      "loss": 0.8517,
+      "step": 14624
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.8821879422235686e-06,
+      "loss": 0.8698,
+      "step": 14625
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.8750929109563562e-06,
+      "loss": 0.8169,
+      "step": 14626
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.8680065600868556e-06,
+      "loss": 0.7712,
+      "step": 14627
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.86092888992715e-06,
+      "loss": 0.8867,
+      "step": 14628
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.85385990078888e-06,
+      "loss": 0.8535,
+      "step": 14629
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.8467995929833292e-06,
+      "loss": 0.8733,
+      "step": 14630
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.8397479668214045e-06,
+      "loss": 0.8155,
+      "step": 14631
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.8327050226136354e-06,
+      "loss": 0.7939,
+      "step": 14632
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.8256707606701294e-06,
+      "loss": 0.8253,
+      "step": 14633
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.818645181300661e-06,
+      "loss": 0.8082,
+      "step": 14634
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.8116282848146267e-06,
+      "loss": 0.8404,
+      "step": 14635
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.804620071520958e-06,
+      "loss": 0.8416,
+      "step": 14636
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.7976205417282963e-06,
+      "loss": 0.8982,
+      "step": 14637
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.790629695744884e-06,
+      "loss": 0.8482,
+      "step": 14638
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.783647533878497e-06,
+      "loss": 0.8605,
+      "step": 14639
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.77667405643669e-06,
+      "loss": 0.864,
+      "step": 14640
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.769709263726461e-06,
+      "loss": 0.8021,
+      "step": 14641
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.7627531560545205e-06,
+      "loss": 0.8135,
+      "step": 14642
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.7558057337272234e-06,
+      "loss": 0.8497,
+      "step": 14643
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.7488669970504367e-06,
+      "loss": 0.8737,
+      "step": 14644
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.7419369463297816e-06,
+      "loss": 0.8368,
+      "step": 14645
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.735015581870326e-06,
+      "loss": 0.8166,
+      "step": 14646
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.7281029039769367e-06,
+      "loss": 0.8296,
+      "step": 14647
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.7211989129539926e-06,
+      "loss": 0.9065,
+      "step": 14648
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.7143036091054953e-06,
+      "loss": 0.873,
+      "step": 14649
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.7074169927350903e-06,
+      "loss": 0.8262,
+      "step": 14650
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.7005390641460014e-06,
+      "loss": 0.8958,
+      "step": 14651
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.693669823641143e-06,
+      "loss": 0.9472,
+      "step": 14652
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.6868092715229832e-06,
+      "loss": 0.8674,
+      "step": 14653
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.6799574080936364e-06,
+      "loss": 0.8497,
+      "step": 14654
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.6731142336547943e-06,
+      "loss": 0.9095,
+      "step": 14655
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.666279748507816e-06,
+      "loss": 0.8876,
+      "step": 14656
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.6594539529536832e-06,
+      "loss": 0.8221,
+      "step": 14657
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.652636847292911e-06,
+      "loss": 0.8797,
+      "step": 14658
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.6458284318257475e-06,
+      "loss": 0.8581,
+      "step": 14659
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.639028706851976e-06,
+      "loss": 0.7608,
+      "step": 14660
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.6322376726710007e-06,
+      "loss": 0.7849,
+      "step": 14661
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.625455329581916e-06,
+      "loss": 0.8362,
+      "step": 14662
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.6186816778833503e-06,
+      "loss": 0.8798,
+      "step": 14663
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.6119167178735746e-06,
+      "loss": 0.8704,
+      "step": 14664
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.605160449850508e-06,
+      "loss": 0.8344,
+      "step": 14665
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5984128741116443e-06,
+      "loss": 0.9068,
+      "step": 14666
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.591673990954124e-06,
+      "loss": 0.8944,
+      "step": 14667
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5849438006746885e-06,
+      "loss": 0.8785,
+      "step": 14668
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5782223035696994e-06,
+      "loss": 0.8809,
+      "step": 14669
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5715094999351207e-06,
+      "loss": 0.8693,
+      "step": 14670
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5648053900666046e-06,
+      "loss": 0.8569,
+      "step": 14671
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5581099742593374e-06,
+      "loss": 0.804,
+      "step": 14672
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5514232528081274e-06,
+      "loss": 0.8313,
+      "step": 14673
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.54474522600745e-06,
+      "loss": 0.8081,
+      "step": 14674
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.538075894151359e-06,
+      "loss": 0.9301,
+      "step": 14675
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.531415257533576e-06,
+      "loss": 0.8293,
+      "step": 14676
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5247633164473315e-06,
+      "loss": 0.8502,
+      "step": 14677
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5181200711856147e-06,
+      "loss": 0.8373,
+      "step": 14678
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5114855220409017e-06,
+      "loss": 0.8651,
+      "step": 14679
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5048596693054037e-06,
+      "loss": 0.8048,
+      "step": 14680
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4982425132708655e-06,
+      "loss": 0.8951,
+      "step": 14681
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4916340542286532e-06,
+      "loss": 0.8464,
+      "step": 14682
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.485034292469779e-06,
+      "loss": 0.8496,
+      "step": 14683
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.478443228284877e-06,
+      "loss": 0.7773,
+      "step": 14684
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4718608619641816e-06,
+      "loss": 0.8451,
+      "step": 14685
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4652871937975497e-06,
+      "loss": 0.9,
+      "step": 14686
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.458722224074439e-06,
+      "loss": 0.9013,
+      "step": 14687
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.452165953083951e-06,
+      "loss": 0.8265,
+      "step": 14688
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.445618381114767e-06,
+      "loss": 0.8692,
+      "step": 14689
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4390795084552555e-06,
+      "loss": 0.7828,
+      "step": 14690
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.4325493353933193e-06,
+      "loss": 0.8203,
+      "step": 14691
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.426027862216529e-06,
+      "loss": 0.8683,
+      "step": 14692
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.4195150892120546e-06,
+      "loss": 0.8501,
+      "step": 14693
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.413011016666689e-06,
+      "loss": 0.8334,
+      "step": 14694
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.406515644866825e-06,
+      "loss": 0.885,
+      "step": 14695
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.4000289740985004e-06,
+      "loss": 0.8618,
+      "step": 14696
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3935510046473542e-06,
+      "loss": 0.8731,
+      "step": 14697
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3870817367986464e-06,
+      "loss": 0.8667,
+      "step": 14698
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3806211708372383e-06,
+      "loss": 0.8464,
+      "step": 14699
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3741693070476357e-06,
+      "loss": 0.8857,
+      "step": 14700
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3677261457139222e-06,
+      "loss": 0.87,
+      "step": 14701
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3612916871198486e-06,
+      "loss": 0.8475,
+      "step": 14702
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3548659315487663e-06,
+      "loss": 0.9765,
+      "step": 14703
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3484488792836046e-06,
+      "loss": 0.8921,
+      "step": 14704
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3420405306069373e-06,
+      "loss": 0.9398,
+      "step": 14705
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3356408858009824e-06,
+      "loss": 0.8859,
+      "step": 14706
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.329249945147538e-06,
+      "loss": 0.8558,
+      "step": 14707
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3228677089280004e-06,
+      "loss": 0.8885,
+      "step": 14708
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.316494177423456e-06,
+      "loss": 0.8924,
+      "step": 14709
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.310129350914547e-06,
+      "loss": 0.7925,
+      "step": 14710
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3037732296815384e-06,
+      "loss": 0.849,
+      "step": 14711
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.2974258140043394e-06,
+      "loss": 0.8357,
+      "step": 14712
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.291087104162437e-06,
+      "loss": 0.7945,
+      "step": 14713
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.2847571004349645e-06,
+      "loss": 0.8549,
+      "step": 14714
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.278435803100676e-06,
+      "loss": 0.8649,
+      "step": 14715
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.2721232124379268e-06,
+      "loss": 0.8599,
+      "step": 14716
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.265819328724672e-06,
+      "loss": 0.8632,
+      "step": 14717
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.2595241522385123e-06,
+      "loss": 0.8082,
+      "step": 14718
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.2532376832566703e-06,
+      "loss": 0.8215,
+      "step": 14719
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.246959922055947e-06,
+      "loss": 0.8645,
+      "step": 14720
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.240690868912809e-06,
+      "loss": 0.8135,
+      "step": 14721
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.2344305241033036e-06,
+      "loss": 0.8237,
+      "step": 14722
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.2281788879030983e-06,
+      "loss": 0.8506,
+      "step": 14723
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.2219359605875066e-06,
+      "loss": 0.8489,
+      "step": 14724
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.2157017424314197e-06,
+      "loss": 0.7937,
+      "step": 14725
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.209476233709351e-06,
+      "loss": 0.8494,
+      "step": 14726
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.203259434695437e-06,
+      "loss": 0.8012,
+      "step": 14727
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.197051345663459e-06,
+      "loss": 0.8273,
+      "step": 14728
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.1908519668867754e-06,
+      "loss": 0.9205,
+      "step": 14729
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.1846612986383905e-06,
+      "loss": 0.8853,
+      "step": 14730
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.178479341190909e-06,
+      "loss": 0.8507,
+      "step": 14731
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.172306094816534e-06,
+      "loss": 0.7925,
+      "step": 14732
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.166141559787094e-06,
+      "loss": 0.7973,
+      "step": 14733
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.1599857363740815e-06,
+      "loss": 0.8868,
+      "step": 14734
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.15383862484857e-06,
+      "loss": 0.8117,
+      "step": 14735
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.1477002254812084e-06,
+      "loss": 0.846,
+      "step": 14736
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.141570538542337e-06,
+      "loss": 0.9187,
+      "step": 14737
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.13544956430185e-06,
+      "loss": 0.8753,
+      "step": 14738
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.1293373030292883e-06,
+      "loss": 0.8423,
+      "step": 14739
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.1232337549938366e-06,
+      "loss": 0.8589,
+      "step": 14740
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.1171389204642344e-06,
+      "loss": 0.8395,
+      "step": 14741
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.111052799708868e-06,
+      "loss": 0.8584,
+      "step": 14742
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.104975392995767e-06,
+      "loss": 0.8731,
+      "step": 14743
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.098906700592518e-06,
+      "loss": 0.8778,
+      "step": 14744
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0928467227663505e-06,
+      "loss": 0.8783,
+      "step": 14745
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0867954597841634e-06,
+      "loss": 0.8469,
+      "step": 14746
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.080752911912387e-06,
+      "loss": 0.8335,
+      "step": 14747
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0747190794170977e-06,
+      "loss": 0.9319,
+      "step": 14748
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.068693962564017e-06,
+      "loss": 0.833,
+      "step": 14749
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0626775616184645e-06,
+      "loss": 0.8912,
+      "step": 14750
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0566698768453407e-06,
+      "loss": 0.8525,
+      "step": 14751
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0506709085092333e-06,
+      "loss": 0.8325,
+      "step": 14752
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0446806568742873e-06,
+      "loss": 0.835,
+      "step": 14753
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0386991222042904e-06,
+      "loss": 0.8506,
+      "step": 14754
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.032726304762611e-06,
+      "loss": 0.8743,
+      "step": 14755
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0267622048123045e-06,
+      "loss": 0.8567,
+      "step": 14756
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0208068226159837e-06,
+      "loss": 0.8182,
+      "step": 14757
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0148601584358608e-06,
+      "loss": 0.8807,
+      "step": 14758
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.008922212533859e-06,
+      "loss": 0.844,
+      "step": 14759
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0029929851714145e-06,
+      "loss": 0.8757,
+      "step": 14760
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9970724766096295e-06,
+      "loss": 0.846,
+      "step": 14761
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.991160687109228e-06,
+      "loss": 0.8803,
+      "step": 14762
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.985257616930514e-06,
+      "loss": 0.8474,
+      "step": 14763
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9793632663334337e-06,
+      "loss": 0.9491,
+      "step": 14764
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9734776355775584e-06,
+      "loss": 0.9201,
+      "step": 14765
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9676007249220586e-06,
+      "loss": 0.9602,
+      "step": 14766
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9617325346257044e-06,
+      "loss": 0.8542,
+      "step": 14767
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9558730649469116e-06,
+      "loss": 0.8691,
+      "step": 14768
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9500223161437182e-06,
+      "loss": 0.8865,
+      "step": 14769
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9441802884737405e-06,
+      "loss": 0.8451,
+      "step": 14770
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.938346982194261e-06,
+      "loss": 0.8899,
+      "step": 14771
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.932522397562098e-06,
+      "loss": 0.8046,
+      "step": 14772
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9267065348337776e-06,
+      "loss": 0.874,
+      "step": 14773
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9208993942654294e-06,
+      "loss": 0.8637,
+      "step": 14774
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9151009761126937e-06,
+      "loss": 0.8232,
+      "step": 14775
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9093112806309654e-06,
+      "loss": 0.8663,
+      "step": 14776
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.903530308075152e-06,
+      "loss": 0.8597,
+      "step": 14777
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.8977580586998722e-06,
+      "loss": 0.8293,
+      "step": 14778
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.891994532759256e-06,
+      "loss": 0.8,
+      "step": 14779
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.8862397305071e-06,
+      "loss": 0.8495,
+      "step": 14780
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.8804936521968685e-06,
+      "loss": 0.957,
+      "step": 14781
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.874756298081537e-06,
+      "loss": 0.8156,
+      "step": 14782
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.869027668413792e-06,
+      "loss": 0.8663,
+      "step": 14783
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.863307763445854e-06,
+      "loss": 0.8633,
+      "step": 14784
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.8575965834296327e-06,
+      "loss": 0.8491,
+      "step": 14785
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.8518941286165937e-06,
+      "loss": 0.8509,
+      "step": 14786
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.8462003992578469e-06,
+      "loss": 0.9199,
+      "step": 14787
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.8405153956041476e-06,
+      "loss": 0.8521,
+      "step": 14788
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.8348391179057845e-06,
+      "loss": 0.8419,
+      "step": 14789
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.8291715664127572e-06,
+      "loss": 0.879,
+      "step": 14790
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.8235127413746e-06,
+      "loss": 0.8558,
+      "step": 14791
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.8178626430405355e-06,
+      "loss": 0.8462,
+      "step": 14792
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.8122212716593645e-06,
+      "loss": 0.8075,
+      "step": 14793
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.8065886274794663e-06,
+      "loss": 0.7941,
+      "step": 14794
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.8009647107489091e-06,
+      "loss": 0.8613,
+      "step": 14795
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.7953495217153172e-06,
+      "loss": 0.9023,
+      "step": 14796
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.7897430606259812e-06,
+      "loss": 0.7845,
+      "step": 14797
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.784145327727771e-06,
+      "loss": 0.887,
+      "step": 14798
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.7785563232671553e-06,
+      "loss": 0.8835,
+      "step": 14799
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.7729760474903157e-06,
+      "loss": 0.858,
+      "step": 14800
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.7674045006428997e-06,
+      "loss": 0.8177,
+      "step": 14801
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.7618416829702888e-06,
+      "loss": 0.8342,
+      "step": 14802
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.756287594717465e-06,
+      "loss": 0.8861,
+      "step": 14803
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.7507422361289437e-06,
+      "loss": 0.8622,
+      "step": 14804
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.7452056074489742e-06,
+      "loss": 0.8146,
+      "step": 14805
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.739677708921339e-06,
+      "loss": 0.8574,
+      "step": 14806
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.7341585407894434e-06,
+      "loss": 0.8082,
+      "step": 14807
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.7286481032963376e-06,
+      "loss": 0.8992,
+      "step": 14808
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.7231463966846718e-06,
+      "loss": 0.8702,
+      "step": 14809
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.7176534211967187e-06,
+      "loss": 0.9434,
+      "step": 14810
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.712169177074352e-06,
+      "loss": 0.881,
+      "step": 14811
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.7066936645590892e-06,
+      "loss": 0.8441,
+      "step": 14812
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.7012268838920264e-06,
+      "loss": 0.8273,
+      "step": 14813
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.6957688353138823e-06,
+      "loss": 0.8681,
+      "step": 14814
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.6903195190650423e-06,
+      "loss": 0.8614,
+      "step": 14815
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.6848789353854255e-06,
+      "loss": 0.8345,
+      "step": 14816
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.6794470845146404e-06,
+      "loss": 0.7922,
+      "step": 14817
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.674023966691851e-06,
+      "loss": 0.8071,
+      "step": 14818
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.6686095821558888e-06,
+      "loss": 0.8139,
+      "step": 14819
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.6632039311451408e-06,
+      "loss": 0.8148,
+      "step": 14820
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.6578070138977054e-06,
+      "loss": 0.8417,
+      "step": 14821
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.65241883065117e-06,
+      "loss": 0.8136,
+      "step": 14822
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.6470393816428564e-06,
+      "loss": 0.8741,
+      "step": 14823
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.6416686671096192e-06,
+      "loss": 0.8788,
+      "step": 14824
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.6363066872879585e-06,
+      "loss": 0.9401,
+      "step": 14825
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.6309534424139738e-06,
+      "loss": 0.846,
+      "step": 14826
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.6256089327234326e-06,
+      "loss": 0.8372,
+      "step": 14827
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.6202731584516795e-06,
+      "loss": 0.8177,
+      "step": 14828
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.614946119833638e-06,
+      "loss": 0.9484,
+      "step": 14829
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.6096278171039204e-06,
+      "loss": 0.8924,
+      "step": 14830
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.6043182504967168e-06,
+      "loss": 0.8773,
+      "step": 14831
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5990174202457742e-06,
+      "loss": 0.8268,
+      "step": 14832
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5937253265846165e-06,
+      "loss": 0.8725,
+      "step": 14833
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5884419697461905e-06,
+      "loss": 0.8605,
+      "step": 14834
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5831673499631995e-06,
+      "loss": 0.7915,
+      "step": 14835
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5779014674679015e-06,
+      "loss": 0.8852,
+      "step": 14836
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.572644322492156e-06,
+      "loss": 0.8633,
+      "step": 14837
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.567395915267489e-06,
+      "loss": 0.877,
+      "step": 14838
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.562156246025004e-06,
+      "loss": 0.9015,
+      "step": 14839
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5569253149954276e-06,
+      "loss": 0.8367,
+      "step": 14840
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5517031224091095e-06,
+      "loss": 0.8948,
+      "step": 14841
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5464896684960205e-06,
+      "loss": 0.8372,
+      "step": 14842
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.541284953485711e-06,
+      "loss": 0.8128,
+      "step": 14843
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.536088977607375e-06,
+      "loss": 0.8459,
+      "step": 14844
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5309017410898296e-06,
+      "loss": 0.8493,
+      "step": 14845
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5257232441614922e-06,
+      "loss": 0.8971,
+      "step": 14846
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5205534870504023e-06,
+      "loss": 0.8683,
+      "step": 14847
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5153924699842004e-06,
+      "loss": 0.9209,
+      "step": 14848
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5102401931901267e-06,
+      "loss": 0.8657,
+      "step": 14849
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.505096656895133e-06,
+      "loss": 0.9324,
+      "step": 14850
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4999618613256383e-06,
+      "loss": 0.8469,
+      "step": 14851
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.494835806707795e-06,
+      "loss": 0.8175,
+      "step": 14852
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4897184932673115e-06,
+      "loss": 0.8796,
+      "step": 14853
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4846099212295627e-06,
+      "loss": 0.8696,
+      "step": 14854
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4795100908194582e-06,
+      "loss": 0.8592,
+      "step": 14855
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4744190022615957e-06,
+      "loss": 0.8212,
+      "step": 14856
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4693366557801514e-06,
+      "loss": 0.8309,
+      "step": 14857
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4642630515989463e-06,
+      "loss": 0.8774,
+      "step": 14858
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4591981899413798e-06,
+      "loss": 0.8165,
+      "step": 14859
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.454142071030473e-06,
+      "loss": 0.8445,
+      "step": 14860
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4490946950888929e-06,
+      "loss": 0.879,
+      "step": 14861
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4440560623388832e-06,
+      "loss": 0.9054,
+      "step": 14862
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4390261730023337e-06,
+      "loss": 0.8309,
+      "step": 14863
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4340050273007332e-06,
+      "loss": 0.8704,
+      "step": 14864
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4289926254551722e-06,
+      "loss": 0.7936,
+      "step": 14865
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.423988967686407e-06,
+      "loss": 0.8017,
+      "step": 14866
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4189940542147285e-06,
+      "loss": 0.8577,
+      "step": 14867
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4140078852601378e-06,
+      "loss": 0.7855,
+      "step": 14868
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.409030461042149e-06,
+      "loss": 0.8025,
+      "step": 14869
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4040617817799863e-06,
+      "loss": 0.9796,
+      "step": 14870
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.3991018476924078e-06,
+      "loss": 0.8336,
+      "step": 14871
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.3941506589978614e-06,
+      "loss": 0.854,
+      "step": 14872
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.3892082159143505e-06,
+      "loss": 0.8317,
+      "step": 14873
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.3842745186595229e-06,
+      "loss": 0.8839,
+      "step": 14874
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.3793495674506274e-06,
+      "loss": 0.8771,
+      "step": 14875
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.3744333625045348e-06,
+      "loss": 0.8416,
+      "step": 14876
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.3695259040377384e-06,
+      "loss": 0.8767,
+      "step": 14877
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.3646271922663546e-06,
+      "loss": 0.8755,
+      "step": 14878
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.359737227406055e-06,
+      "loss": 0.8702,
+      "step": 14879
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.3548560096721786e-06,
+      "loss": 0.8847,
+      "step": 14880
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.3499835392797088e-06,
+      "loss": 0.8668,
+      "step": 14881
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.3451198164431633e-06,
+      "loss": 0.8635,
+      "step": 14882
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.3402648413767483e-06,
+      "loss": 0.8232,
+      "step": 14883
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.335418614294226e-06,
+      "loss": 0.8188,
+      "step": 14884
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.3305811354090037e-06,
+      "loss": 0.8617,
+      "step": 14885
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.3257524049341107e-06,
+      "loss": 0.8245,
+      "step": 14886
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.3209324230821773e-06,
+      "loss": 0.9406,
+      "step": 14887
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.316121190065478e-06,
+      "loss": 0.826,
+      "step": 14888
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.3113187060957987e-06,
+      "loss": 0.8806,
+      "step": 14889
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.306524971384704e-06,
+      "loss": 0.9301,
+      "step": 14890
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.3017399861432245e-06,
+      "loss": 0.831,
+      "step": 14891
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.296963750582103e-06,
+      "loss": 0.8136,
+      "step": 14892
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.2921962649116382e-06,
+      "loss": 0.8048,
+      "step": 14893
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.287437529341795e-06,
+      "loss": 0.8235,
+      "step": 14894
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.282687544082073e-06,
+      "loss": 0.852,
+      "step": 14895
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.2779463093417044e-06,
+      "loss": 0.9034,
+      "step": 14896
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.2732138253294334e-06,
+      "loss": 0.829,
+      "step": 14897
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.268490092253627e-06,
+      "loss": 0.871,
+      "step": 14898
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.2637751103223627e-06,
+      "loss": 0.8376,
+      "step": 14899
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.2590688797432081e-06,
+      "loss": 0.8295,
+      "step": 14900
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.2543714007234197e-06,
+      "loss": 0.8639,
+      "step": 14901
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.249682673469854e-06,
+      "loss": 0.8256,
+      "step": 14902
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.2450026981889906e-06,
+      "loss": 0.845,
+      "step": 14903
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.2403314750868867e-06,
+      "loss": 0.8902,
+      "step": 14904
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.2356690043692664e-06,
+      "loss": 0.8966,
+      "step": 14905
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.2310152862414104e-06,
+      "loss": 0.8887,
+      "step": 14906
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.2263703209082878e-06,
+      "loss": 0.8462,
+      "step": 14907
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.2217341085743793e-06,
+      "loss": 0.893,
+      "step": 14908
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.217106649443922e-06,
+      "loss": 0.884,
+      "step": 14909
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.2124879437205972e-06,
+      "loss": 0.8797,
+      "step": 14910
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.207877991607864e-06,
+      "loss": 0.8566,
+      "step": 14911
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.203276793308672e-06,
+      "loss": 0.8535,
+      "step": 14912
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1986843490256583e-06,
+      "loss": 0.8565,
+      "step": 14913
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1941006589610393e-06,
+      "loss": 0.853,
+      "step": 14914
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1895257233166756e-06,
+      "loss": 0.8187,
+      "step": 14915
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1849595422940285e-06,
+      "loss": 0.8714,
+      "step": 14916
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1804021160941147e-06,
+      "loss": 0.8764,
+      "step": 14917
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.175853444917685e-06,
+      "loss": 0.8786,
+      "step": 14918
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1713135289650234e-06,
+      "loss": 0.8379,
+      "step": 14919
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1667823684360146e-06,
+      "loss": 0.8415,
+      "step": 14920
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1622599635302323e-06,
+      "loss": 0.8925,
+      "step": 14921
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1577463144468059e-06,
+      "loss": 0.875,
+      "step": 14922
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1532414213844655e-06,
+      "loss": 0.9078,
+      "step": 14923
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1487452845416302e-06,
+      "loss": 0.8087,
+      "step": 14924
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1442579041162527e-06,
+      "loss": 0.8482,
+      "step": 14925
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1397792803059304e-06,
+      "loss": 0.872,
+      "step": 14926
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1353094133079057e-06,
+      "loss": 0.8253,
+      "step": 14927
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.130848303319021e-06,
+      "loss": 0.9626,
+      "step": 14928
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1263959505356747e-06,
+      "loss": 0.818,
+      "step": 14929
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1219523551539547e-06,
+      "loss": 0.8745,
+      "step": 14930
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1175175173695484e-06,
+      "loss": 0.9009,
+      "step": 14931
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1130914373777002e-06,
+      "loss": 0.8966,
+      "step": 14932
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1086741153733426e-06,
+      "loss": 0.8422,
+      "step": 14933
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1042655515510092e-06,
+      "loss": 0.8598,
+      "step": 14934
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.099865746104789e-06,
+      "loss": 0.9145,
+      "step": 14935
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0954746992284604e-06,
+      "loss": 0.811,
+      "step": 14936
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0910924111153575e-06,
+      "loss": 0.8255,
+      "step": 14937
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0867188819584816e-06,
+      "loss": 0.8367,
+      "step": 14938
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0823541119504121e-06,
+      "loss": 0.871,
+      "step": 14939
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0779981012833506e-06,
+      "loss": 0.8209,
+      "step": 14940
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0736508501490994e-06,
+      "loss": 0.8227,
+      "step": 14941
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.069312358739105e-06,
+      "loss": 0.8983,
+      "step": 14942
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0649826272444153e-06,
+      "loss": 0.846,
+      "step": 14943
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.060661655855677e-06,
+      "loss": 0.9263,
+      "step": 14944
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0563494447631605e-06,
+      "loss": 0.8635,
+      "step": 14945
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0520459941567805e-06,
+      "loss": 0.8396,
+      "step": 14946
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.04775130422603e-06,
+      "loss": 0.9294,
+      "step": 14947
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.043465375160002e-06,
+      "loss": 0.8543,
+      "step": 14948
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0391882071474568e-06,
+      "loss": 0.8412,
+      "step": 14949
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0349198003767324e-06,
+      "loss": 0.789,
+      "step": 14950
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0306601550357676e-06,
+      "loss": 0.8058,
+      "step": 14951
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0264092713121675e-06,
+      "loss": 0.8106,
+      "step": 14952
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0221671493930939e-06,
+      "loss": 0.9151,
+      "step": 14953
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.017933789465375e-06,
+      "loss": 0.8152,
+      "step": 14954
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.013709191715395e-06,
+      "loss": 0.861,
+      "step": 14955
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0094933563292054e-06,
+      "loss": 0.8453,
+      "step": 14956
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0052862834924347e-06,
+      "loss": 0.8426,
+      "step": 14957
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0010879733903578e-06,
+      "loss": 0.8721,
+      "step": 14958
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.968984262078707e-07,
+      "loss": 0.8444,
+      "step": 14959
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.927176421294039e-07,
+      "loss": 0.8501,
+      "step": 14960
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.885456213390765e-07,
+      "loss": 0.946,
+      "step": 14961
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.843823640206306e-07,
+      "loss": 0.8859,
+      "step": 14962
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.80227870357364e-07,
+      "loss": 0.8455,
+      "step": 14963
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.760821405322417e-07,
+      "loss": 0.8422,
+      "step": 14964
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.719451747278064e-07,
+      "loss": 0.7571,
+      "step": 14965
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.678169731262233e-07,
+      "loss": 0.8606,
+      "step": 14966
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.636975359093248e-07,
+      "loss": 0.8978,
+      "step": 14967
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.595868632584547e-07,
+      "loss": 0.7916,
+      "step": 14968
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.554849553546463e-07,
+      "loss": 0.8463,
+      "step": 14969
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.513918123785104e-07,
+      "loss": 0.7789,
+      "step": 14970
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.473074345103472e-07,
+      "loss": 0.8932,
+      "step": 14971
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.432318219299463e-07,
+      "loss": 0.8629,
+      "step": 14972
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.391649748167863e-07,
+      "loss": 0.9057,
+      "step": 14973
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.351068933500129e-07,
+      "loss": 0.871,
+      "step": 14974
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.310575777082387e-07,
+      "loss": 0.8297,
+      "step": 14975
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.270170280698321e-07,
+      "loss": 0.8251,
+      "step": 14976
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.229852446126952e-07,
+      "loss": 0.8812,
+      "step": 14977
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.18962227514375e-07,
+      "loss": 0.8546,
+      "step": 14978
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.149479769520186e-07,
+      "loss": 0.814,
+      "step": 14979
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.109424931023957e-07,
+      "loss": 0.8533,
+      "step": 14980
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.069457761418986e-07,
+      "loss": 0.8316,
+      "step": 14981
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.029578262465199e-07,
+      "loss": 0.8787,
+      "step": 14982
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.989786435918523e-07,
+      "loss": 0.8743,
+      "step": 14983
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.950082283531335e-07,
+      "loss": 0.918,
+      "step": 14984
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.910465807052015e-07,
+      "loss": 0.8127,
+      "step": 14985
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.870937008225166e-07,
+      "loss": 0.8811,
+      "step": 14986
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.831495888791174e-07,
+      "loss": 0.807,
+      "step": 14987
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.792142450486873e-07,
+      "loss": 0.8414,
+      "step": 14988
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.752876695045542e-07,
+      "loss": 0.8277,
+      "step": 14989
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.713698624195799e-07,
+      "loss": 0.8413,
+      "step": 14990
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.674608239663151e-07,
+      "loss": 0.8876,
+      "step": 14991
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.635605543168889e-07,
+      "loss": 0.8411,
+      "step": 14992
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.596690536430308e-07,
+      "loss": 0.8753,
+      "step": 14993
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.557863221161144e-07,
+      "loss": 0.726,
+      "step": 14994
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.519123599071366e-07,
+      "loss": 0.8417,
+      "step": 14995
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.480471671866497e-07,
+      "loss": 0.8837,
+      "step": 14996
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.441907441248953e-07,
+      "loss": 0.8764,
+      "step": 14997
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.40343090891671e-07,
+      "loss": 0.9003,
+      "step": 14998
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.365042076563967e-07,
+      "loss": 0.8611,
+      "step": 14999
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.326740945881373e-07,
+      "loss": 0.8757,
+      "step": 15000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.288527518555356e-07,
+      "loss": 0.8443,
+      "step": 15001
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.250401796269014e-07,
+      "loss": 0.8494,
+      "step": 15002
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.212363780700782e-07,
+      "loss": 0.8232,
+      "step": 15003
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.174413473525766e-07,
+      "loss": 0.8741,
+      "step": 15004
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.136550876415294e-07,
+      "loss": 0.8646,
+      "step": 15005
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.098775991036478e-07,
+      "loss": 0.8692,
+      "step": 15006
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.061088819052876e-07,
+      "loss": 0.8242,
+      "step": 15007
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.023489362123826e-07,
+      "loss": 0.8198,
+      "step": 15008
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.985977621905117e-07,
+      "loss": 0.8273,
+      "step": 15009
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.948553600048758e-07,
+      "loss": 0.8623,
+      "step": 15010
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.911217298202323e-07,
+      "loss": 0.8459,
+      "step": 15011
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.873968718010494e-07,
+      "loss": 0.8903,
+      "step": 15012
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.83680786111285e-07,
+      "loss": 0.91,
+      "step": 15013
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.799734729146302e-07,
+      "loss": 0.8508,
+      "step": 15014
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.76274932374288e-07,
+      "loss": 0.8332,
+      "step": 15015
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.725851646531723e-07,
+      "loss": 0.8774,
+      "step": 15016
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.689041699137534e-07,
+      "loss": 0.8262,
+      "step": 15017
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.652319483181014e-07,
+      "loss": 0.8946,
+      "step": 15018
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.615685000279093e-07,
+      "loss": 0.839,
+      "step": 15019
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.579138252045592e-07,
+      "loss": 0.8536,
+      "step": 15020
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.542679240089224e-07,
+      "loss": 0.8481,
+      "step": 15021
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.506307966015813e-07,
+      "loss": 0.8177,
+      "step": 15022
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.470024431426747e-07,
+      "loss": 0.848,
+      "step": 15023
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.433828637920081e-07,
+      "loss": 0.994,
+      "step": 15024
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.397720587089208e-07,
+      "loss": 0.9259,
+      "step": 15025
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.36170028052463e-07,
+      "loss": 0.8339,
+      "step": 15026
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.325767719812193e-07,
+      "loss": 0.8393,
+      "step": 15027
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.289922906534407e-07,
+      "loss": 0.8902,
+      "step": 15028
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.254165842269566e-07,
+      "loss": 0.8816,
+      "step": 15029
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.218496528592189e-07,
+      "loss": 0.8706,
+      "step": 15030
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.18291496707324e-07,
+      "loss": 0.8404,
+      "step": 15031
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.147421159279244e-07,
+      "loss": 0.8261,
+      "step": 15032
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.112015106773173e-07,
+      "loss": 0.8652,
+      "step": 15033
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.076696811114447e-07,
+      "loss": 0.8392,
+      "step": 15034
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.041466273858044e-07,
+      "loss": 0.7925,
+      "step": 15035
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.00632349655539e-07,
+      "loss": 0.7871,
+      "step": 15036
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.971268480753912e-07,
+      "loss": 0.8154,
+      "step": 15037
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.936301227997488e-07,
+      "loss": 0.8704,
+      "step": 15038
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.901421739825775e-07,
+      "loss": 0.8662,
+      "step": 15039
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.866630017774656e-07,
+      "loss": 0.8841,
+      "step": 15040
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.831926063376237e-07,
+      "loss": 0.8533,
+      "step": 15041
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.797309878158631e-07,
+      "loss": 0.8727,
+      "step": 15042
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.762781463646394e-07,
+      "loss": 0.887,
+      "step": 15043
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.728340821359424e-07,
+      "loss": 0.8279,
+      "step": 15044
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.693987952815173e-07,
+      "loss": 0.8019,
+      "step": 15045
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.659722859525541e-07,
+      "loss": 0.7735,
+      "step": 15046
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.625545542999989e-07,
+      "loss": 0.8471,
+      "step": 15047
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.591456004743091e-07,
+      "loss": 0.8767,
+      "step": 15048
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.557454246256312e-07,
+      "loss": 0.8611,
+      "step": 15049
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.523540269036677e-07,
+      "loss": 0.9495,
+      "step": 15050
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.489714074577657e-07,
+      "loss": 0.8722,
+      "step": 15051
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.455975664368952e-07,
+      "loss": 0.8405,
+      "step": 15052
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.422325039896038e-07,
+      "loss": 0.8823,
+      "step": 15053
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.388762202640841e-07,
+      "loss": 0.7931,
+      "step": 15054
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.355287154081069e-07,
+      "loss": 0.848,
+      "step": 15055
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.321899895691096e-07,
+      "loss": 0.8289,
+      "step": 15056
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.288600428941083e-07,
+      "loss": 0.8964,
+      "step": 15057
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.255388755297187e-07,
+      "loss": 0.9016,
+      "step": 15058
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.222264876222017e-07,
+      "loss": 0.8515,
+      "step": 15059
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.189228793173963e-07,
+      "loss": 0.8503,
+      "step": 15060
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.15628050760808e-07,
+      "loss": 0.8344,
+      "step": 15061
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.123420020975434e-07,
+      "loss": 0.8347,
+      "step": 15062
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.090647334722199e-07,
+      "loss": 0.8678,
+      "step": 15063
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.057962450292554e-07,
+      "loss": 0.8844,
+      "step": 15064
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.025365369124902e-07,
+      "loss": 0.885,
+      "step": 15065
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.992856092655208e-07,
+      "loss": 0.8143,
+      "step": 15066
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.96043462231477e-07,
+      "loss": 0.8737,
+      "step": 15067
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.928100959531335e-07,
+      "loss": 0.8712,
+      "step": 15068
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.895855105728875e-07,
+      "loss": 0.8343,
+      "step": 15069
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.863697062326923e-07,
+      "loss": 0.8315,
+      "step": 15070
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.831626830742121e-07,
+      "loss": 0.8461,
+      "step": 15071
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.79964441238623e-07,
+      "loss": 0.8259,
+      "step": 15072
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.767749808667678e-07,
+      "loss": 0.8658,
+      "step": 15073
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.735943020991119e-07,
+      "loss": 0.8688,
+      "step": 15074
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.70422405075699e-07,
+      "loss": 0.8877,
+      "step": 15075
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.672592899362172e-07,
+      "loss": 0.7951,
+      "step": 15076
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.64104956819933e-07,
+      "loss": 0.8496,
+      "step": 15077
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.609594058657796e-07,
+      "loss": 0.8903,
+      "step": 15078
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.578226372122464e-07,
+      "loss": 0.8757,
+      "step": 15079
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.546946509974671e-07,
+      "loss": 0.8471,
+      "step": 15080
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.515754473591983e-07,
+      "loss": 0.8698,
+      "step": 15081
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.484650264347524e-07,
+      "loss": 0.9111,
+      "step": 15082
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.453633883611309e-07,
+      "loss": 0.8632,
+      "step": 15083
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.422705332749134e-07,
+      "loss": 0.8616,
+      "step": 15084
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.391864613123021e-07,
+      "loss": 0.8186,
+      "step": 15085
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.361111726090773e-07,
+      "loss": 0.8607,
+      "step": 15086
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.330446673006639e-07,
+      "loss": 0.8389,
+      "step": 15087
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.299869455221096e-07,
+      "loss": 0.8463,
+      "step": 15088
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.269380074080621e-07,
+      "loss": 0.9028,
+      "step": 15089
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.238978530927918e-07,
+      "loss": 0.8206,
+      "step": 15090
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.20866482710125e-07,
+      "loss": 0.8573,
+      "step": 15091
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.178438963935772e-07,
+      "loss": 0.8959,
+      "step": 15092
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.148300942762641e-07,
+      "loss": 0.8138,
+      "step": 15093
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.118250764908794e-07,
+      "loss": 0.8319,
+      "step": 15094
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.088288431697618e-07,
+      "loss": 0.8579,
+      "step": 15095
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.058413944448281e-07,
+      "loss": 0.8857,
+      "step": 15096
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.028627304476396e-07,
+      "loss": 0.8604,
+      "step": 15097
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.998928513093803e-07,
+      "loss": 0.8901,
+      "step": 15098
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.969317571608123e-07,
+      "loss": 0.7955,
+      "step": 15099
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.939794481323423e-07,
+      "loss": 0.9258,
+      "step": 15100
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.910359243539331e-07,
+      "loss": 0.8625,
+      "step": 15101
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.881011859552587e-07,
+      "loss": 0.8748,
+      "step": 15102
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.851752330655046e-07,
+      "loss": 0.8547,
+      "step": 15103
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.822580658135456e-07,
+      "loss": 0.834,
+      "step": 15104
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.793496843278345e-07,
+      "loss": 0.8627,
+      "step": 15105
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.7645008873642427e-07,
+      "loss": 0.8534,
+      "step": 15106
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.7355927916703515e-07,
+      "loss": 0.8439,
+      "step": 15107
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.706772557468986e-07,
+      "loss": 0.7842,
+      "step": 15108
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.678040186030019e-07,
+      "loss": 0.8709,
+      "step": 15109
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.649395678618218e-07,
+      "loss": 0.8826,
+      "step": 15110
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.6208390364950173e-07,
+      "loss": 0.8175,
+      "step": 15111
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.5923702609178554e-07,
+      "loss": 0.8218,
+      "step": 15112
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.563989353140397e-07,
+      "loss": 0.8618,
+      "step": 15113
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.5356963144125296e-07,
+      "loss": 0.8096,
+      "step": 15114
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.507491145979925e-07,
+      "loss": 0.8468,
+      "step": 15115
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.4793738490849224e-07,
+      "loss": 0.8199,
+      "step": 15116
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.45134442496542e-07,
+      "loss": 0.814,
+      "step": 15117
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.4234028748555424e-07,
+      "loss": 0.9232,
+      "step": 15118
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.395549199986082e-07,
+      "loss": 0.8735,
+      "step": 15119
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.3677834015831695e-07,
+      "loss": 0.8673,
+      "step": 15120
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.340105480869827e-07,
+      "loss": 0.8685,
+      "step": 15121
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.312515439064635e-07,
+      "loss": 0.7986,
+      "step": 15122
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.285013277382621e-07,
+      "loss": 0.8339,
+      "step": 15123
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.2575989970350393e-07,
+      "loss": 0.8762,
+      "step": 15124
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.2302725992284796e-07,
+      "loss": 0.8448,
+      "step": 15125
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.203034085166868e-07,
+      "loss": 0.8493,
+      "step": 15126
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.1758834560492457e-07,
+      "loss": 0.8858,
+      "step": 15127
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.148820713071544e-07,
+      "loss": 0.7796,
+      "step": 15128
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.121845857425255e-07,
+      "loss": 0.901,
+      "step": 15129
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.094958890298095e-07,
+      "loss": 0.8914,
+      "step": 15130
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.0681598128744504e-07,
+      "loss": 0.9149,
+      "step": 15131
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.0414486263338215e-07,
+      "loss": 0.8225,
+      "step": 15132
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.014825331853045e-07,
+      "loss": 0.9137,
+      "step": 15133
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.9882899306040725e-07,
+      "loss": 0.8956,
+      "step": 15134
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.961842423755746e-07,
+      "loss": 0.8008,
+      "step": 15135
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.9354828124722466e-07,
+      "loss": 0.8939,
+      "step": 15136
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.9092110979146447e-07,
+      "loss": 0.8473,
+      "step": 15137
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.883027281239571e-07,
+      "loss": 0.8645,
+      "step": 15138
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.8569313636001023e-07,
+      "loss": 0.8347,
+      "step": 15139
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.830923346145543e-07,
+      "loss": 0.84,
+      "step": 15140
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.805003230021198e-07,
+      "loss": 0.7983,
+      "step": 15141
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.7791710163681547e-07,
+      "loss": 0.8362,
+      "step": 15142
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.75342670632417e-07,
+      "loss": 0.8774,
+      "step": 15143
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.7277703010225594e-07,
+      "loss": 0.8472,
+      "step": 15144
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.70220180159353e-07,
+      "loss": 0.9016,
+      "step": 15145
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.676721209162848e-07,
+      "loss": 0.8484,
+      "step": 15146
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.651328524852504e-07,
+      "loss": 0.8711,
+      "step": 15147
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.626023749780716e-07,
+      "loss": 0.8034,
+      "step": 15148
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.6008068850617025e-07,
+      "loss": 0.7401,
+      "step": 15149
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5756779318059095e-07,
+      "loss": 0.8213,
+      "step": 15150
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.550636891120007e-07,
+      "loss": 0.8686,
+      "step": 15151
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5256837641064467e-07,
+      "loss": 0.8221,
+      "step": 15152
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.500818551864349e-07,
+      "loss": 0.8691,
+      "step": 15153
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.476041255488394e-07,
+      "loss": 0.8798,
+      "step": 15154
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.4513518760697083e-07,
+      "loss": 0.8027,
+      "step": 15155
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.426750414695645e-07,
+      "loss": 0.821,
+      "step": 15156
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.4022368724491163e-07,
+      "loss": 0.9209,
+      "step": 15157
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.3778112504101456e-07,
+      "loss": 0.8378,
+      "step": 15158
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.3534735496538736e-07,
+      "loss": 0.8516,
+      "step": 15159
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.329223771252332e-07,
+      "loss": 0.8983,
+      "step": 15160
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.3050619162731113e-07,
+      "loss": 0.7764,
+      "step": 15161
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.280987985780248e-07,
+      "loss": 0.8436,
+      "step": 15162
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.2570019808337847e-07,
+      "loss": 0.8858,
+      "step": 15163
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.233103902490209e-07,
+      "loss": 0.8147,
+      "step": 15164
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.2092937518015677e-07,
+      "loss": 0.8435,
+      "step": 15165
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.185571529816356e-07,
+      "loss": 0.8805,
+      "step": 15166
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.1619372375795153e-07,
+      "loss": 0.8599,
+      "step": 15167
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.138390876131325e-07,
+      "loss": 0.796,
+      "step": 15168
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.114932446508956e-07,
+      "loss": 0.8754,
+      "step": 15169
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0915619497453587e-07,
+      "loss": 0.7819,
+      "step": 15170
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.06827938686971e-07,
+      "loss": 0.8184,
+      "step": 15171
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0450847589069685e-07,
+      "loss": 0.8728,
+      "step": 15172
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.021978066878761e-07,
+      "loss": 0.8884,
+      "step": 15173
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.9989593118027184e-07,
+      "loss": 0.8839,
+      "step": 15174
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.976028494692029e-07,
+      "loss": 0.9079,
+      "step": 15175
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.953185616556775e-07,
+      "loss": 0.8772,
+      "step": 15176
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.9304306784028177e-07,
+      "loss": 0.8051,
+      "step": 15177
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.907763681232023e-07,
+      "loss": 0.8784,
+      "step": 15178
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.885184626042703e-07,
+      "loss": 0.8257,
+      "step": 15179
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.8626935138291735e-07,
+      "loss": 0.8069,
+      "step": 15180
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.8402903455815313e-07,
+      "loss": 0.8871,
+      "step": 15181
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.8179751222865424e-07,
+      "loss": 0.8572,
+      "step": 15182
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.795747844926977e-07,
+      "loss": 0.8392,
+      "step": 15183
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.773608514481385e-07,
+      "loss": 0.8171,
+      "step": 15184
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.7515571319245425e-07,
+      "loss": 0.8708,
+      "step": 15185
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.7295936982278945e-07,
+      "loss": 0.8168,
+      "step": 15186
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.7077182143582235e-07,
+      "loss": 0.9344,
+      "step": 15187
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.6859306812789807e-07,
+      "loss": 0.8993,
+      "step": 15188
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.664231099949399e-07,
+      "loss": 0.917,
+      "step": 15189
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.6426194713256027e-07,
+      "loss": 0.8618,
+      "step": 15190
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.621095796358608e-07,
+      "loss": 0.8457,
+      "step": 15191
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.599660075996324e-07,
+      "loss": 0.8829,
+      "step": 15192
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.5783123111828846e-07,
+      "loss": 0.9186,
+      "step": 15193
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.557052502858426e-07,
+      "loss": 0.8414,
+      "step": 15194
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.535880651958644e-07,
+      "loss": 0.8387,
+      "step": 15195
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.514796759416349e-07,
+      "loss": 0.9095,
+      "step": 15196
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.4938008261596867e-07,
+      "loss": 0.8947,
+      "step": 15197
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.4728928531134733e-07,
+      "loss": 0.8452,
+      "step": 15198
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.452072841198083e-07,
+      "loss": 0.8402,
+      "step": 15199
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.431340791330339e-07,
+      "loss": 0.8206,
+      "step": 15200
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.4106967044235093e-07,
+      "loss": 0.8496,
+      "step": 15201
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.3901405813864243e-07,
+      "loss": 0.8072,
+      "step": 15202
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.3696724231241363e-07,
+      "loss": 0.8857,
+      "step": 15203
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.3492922305381472e-07,
+      "loss": 0.8919,
+      "step": 15204
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.3290000045259607e-07,
+      "loss": 0.8909,
+      "step": 15205
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.308795745980863e-07,
+      "loss": 0.8694,
+      "step": 15206
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.2886794557928082e-07,
+      "loss": 0.8898,
+      "step": 15207
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.2686511348477546e-07,
+      "loss": 0.902,
+      "step": 15208
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.2487107840272191e-07,
+      "loss": 0.9351,
+      "step": 15209
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.2288584042096106e-07,
+      "loss": 0.8011,
+      "step": 15210
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.2090939962688961e-07,
+      "loss": 0.8777,
+      "step": 15211
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.189417561075713e-07,
+      "loss": 0.8238,
+      "step": 15212
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.1698290994960347e-07,
+      "loss": 0.8807,
+      "step": 15213
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.150328612392949e-07,
+      "loss": 0.8603,
+      "step": 15214
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.13091610062488e-07,
+      "loss": 0.8119,
+      "step": 15215
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.1115915650467e-07,
+      "loss": 0.8879,
+      "step": 15216
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.0923550065092835e-07,
+      "loss": 0.8398,
+      "step": 15217
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.0732064258599527e-07,
+      "loss": 0.8455,
+      "step": 15218
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.054145823941811e-07,
+      "loss": 0.8118,
+      "step": 15219
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.0351732015939652e-07,
+      "loss": 0.8239,
+      "step": 15220
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.0162885596521907e-07,
+      "loss": 0.8564,
+      "step": 15221
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.997491898947823e-07,
+      "loss": 0.7539,
+      "step": 15222
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9787832203088663e-07,
+      "loss": 0.8358,
+      "step": 15223
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.960162524558884e-07,
+      "loss": 0.9105,
+      "step": 15224
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9416298125178867e-07,
+      "loss": 0.8665,
+      "step": 15225
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9231850850021105e-07,
+      "loss": 0.8403,
+      "step": 15226
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9048283428237946e-07,
+      "loss": 0.8825,
+      "step": 15227
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.886559586790959e-07,
+      "loss": 0.8708,
+      "step": 15228
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.868378817708294e-07,
+      "loss": 0.9051,
+      "step": 15229
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8502860363764918e-07,
+      "loss": 0.8477,
+      "step": 15230
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8322812435920267e-07,
+      "loss": 0.9091,
+      "step": 15231
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.81436444014782e-07,
+      "loss": 0.8259,
+      "step": 15232
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.796535626833018e-07,
+      "loss": 0.8951,
+      "step": 15233
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.7787948044323267e-07,
+      "loss": 0.8891,
+      "step": 15234
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.7611419737273427e-07,
+      "loss": 0.9256,
+      "step": 15235
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.7435771354954444e-07,
+      "loss": 0.8511,
+      "step": 15236
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.726100290509791e-07,
+      "loss": 0.8494,
+      "step": 15237
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.7087114395399895e-07,
+      "loss": 0.8569,
+      "step": 15238
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.6914105833518713e-07,
+      "loss": 0.8425,
+      "step": 15239
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.6741977227074934e-07,
+      "loss": 0.8127,
+      "step": 15240
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.6570728583644722e-07,
+      "loss": 0.8047,
+      "step": 15241
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.6400359910770934e-07,
+      "loss": 0.8274,
+      "step": 15242
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.6230871215956456e-07,
+      "loss": 0.8434,
+      "step": 15243
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.6062262506659763e-07,
+      "loss": 0.876,
+      "step": 15244
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.5894533790312694e-07,
+      "loss": 0.8239,
+      "step": 15245
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.5727685074298225e-07,
+      "loss": 0.9084,
+      "step": 15246
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.55617163659616e-07,
+      "loss": 0.8654,
+      "step": 15247
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.5396627672612517e-07,
+      "loss": 0.8369,
+      "step": 15248
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.5232419001522945e-07,
+      "loss": 0.8902,
+      "step": 15249
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.5069090359918214e-07,
+      "loss": 0.8569,
+      "step": 15250
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4906641754997007e-07,
+      "loss": 0.8617,
+      "step": 15251
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4745073193909166e-07,
+      "loss": 0.8277,
+      "step": 15252
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4584384683771213e-07,
+      "loss": 0.8357,
+      "step": 15253
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4424576231655275e-07,
+      "loss": 0.8744,
+      "step": 15254
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4265647844602382e-07,
+      "loss": 0.8889,
+      "step": 15255
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4107599529611382e-07,
+      "loss": 0.7971,
+      "step": 15256
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.3950431293638933e-07,
+      "loss": 0.8319,
+      "step": 15257
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.3794143143606165e-07,
+      "loss": 0.8419,
+      "step": 15258
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.363873508639646e-07,
+      "loss": 0.8872,
+      "step": 15259
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.3484207128855454e-07,
+      "loss": 0.8705,
+      "step": 15260
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.3330559277784371e-07,
+      "loss": 0.8949,
+      "step": 15261
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.3177791539948913e-07,
+      "loss": 0.8376,
+      "step": 15262
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.302590392207703e-07,
+      "loss": 0.8448,
+      "step": 15263
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.2874896430858928e-07,
+      "loss": 0.8339,
+      "step": 15264
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.2724769072942622e-07,
+      "loss": 0.8763,
+      "step": 15265
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.2575521854938377e-07,
+      "loss": 0.9288,
+      "step": 15266
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.2427154783420936e-07,
+      "loss": 0.8028,
+      "step": 15267
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.227966786492063e-07,
+      "loss": 0.8448,
+      "step": 15268
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.2133061105932265e-07,
+      "loss": 0.8391,
+      "step": 15269
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.1987334512915115e-07,
+      "loss": 0.9012,
+      "step": 15270
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.1842488092281834e-07,
+      "loss": 0.8903,
+      "step": 15271
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.1698521850411758e-07,
+      "loss": 0.8984,
+      "step": 15272
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.1555435793646486e-07,
+      "loss": 0.8639,
+      "step": 15273
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.1413229928285418e-07,
+      "loss": 0.9584,
+      "step": 15274
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.1271904260592437e-07,
+      "loss": 0.8265,
+      "step": 15275
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.1131458796787008e-07,
+      "loss": 0.8892,
+      "step": 15276
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.0991893543057519e-07,
+      "loss": 0.8647,
+      "step": 15277
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.085320850554572e-07,
+      "loss": 0.8529,
+      "step": 15278
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.071540369036228e-07,
+      "loss": 0.8512,
+      "step": 15279
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.0578479103575678e-07,
+      "loss": 0.8485,
+      "step": 15280
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.0442434751212205e-07,
+      "loss": 0.8963,
+      "step": 15281
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.0307270639264843e-07,
+      "loss": 0.8905,
+      "step": 15282
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.0172986773684389e-07,
+      "loss": 0.8507,
+      "step": 15283
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.0039583160383891e-07,
+      "loss": 0.83,
+      "step": 15284
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 9.90705980523865e-08,
+      "loss": 0.8314,
+      "step": 15285
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 9.775416714083995e-08,
+      "loss": 0.7803,
+      "step": 15286
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 9.644653892717515e-08,
+      "loss": 0.7994,
+      "step": 15287
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 9.514771346896823e-08,
+      "loss": 0.7933,
+      "step": 15288
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 9.385769082341789e-08,
+      "loss": 0.7959,
+      "step": 15289
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 9.257647104732314e-08,
+      "loss": 0.8858,
+      "step": 15290
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 9.130405419710553e-08,
+      "loss": 0.8485,
+      "step": 15291
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 9.004044032878689e-08,
+      "loss": 0.8691,
+      "step": 15292
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.878562949801161e-08,
+      "loss": 0.8389,
+      "step": 15293
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.75396217600466e-08,
+      "loss": 0.8458,
+      "step": 15294
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.630241716975907e-08,
+      "loss": 0.8402,
+      "step": 15295
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.507401578163876e-08,
+      "loss": 0.8882,
+      "step": 15296
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.385441764975354e-08,
+      "loss": 0.8791,
+      "step": 15297
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.264362282781602e-08,
+      "loss": 0.8212,
+      "step": 15298
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.144163136913907e-08,
+      "loss": 0.8601,
+      "step": 15299
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.024844332668036e-08,
+      "loss": 0.8887,
+      "step": 15300
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.906405875295342e-08,
+      "loss": 0.9142,
+      "step": 15301
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.788847770011654e-08,
+      "loss": 0.8962,
+      "step": 15302
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.672170021995052e-08,
+      "loss": 0.8277,
+      "step": 15303
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.556372636383646e-08,
+      "loss": 0.8684,
+      "step": 15304
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.441455618273363e-08,
+      "loss": 0.8043,
+      "step": 15305
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.327418972726819e-08,
+      "loss": 0.9403,
+      "step": 15306
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.214262704766662e-08,
+      "loss": 0.7795,
+      "step": 15307
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.101986819373352e-08,
+      "loss": 0.7768,
+      "step": 15308
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.990591321494044e-08,
+      "loss": 0.8527,
+      "step": 15309
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.880076216029263e-08,
+      "loss": 0.854,
+      "step": 15310
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.770441507850667e-08,
+      "loss": 0.8135,
+      "step": 15311
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.661687201783284e-08,
+      "loss": 0.8734,
+      "step": 15312
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.553813302616618e-08,
+      "loss": 0.8493,
+      "step": 15313
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.446819815100203e-08,
+      "loss": 0.8267,
+      "step": 15314
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.340706743948044e-08,
+      "loss": 0.8567,
+      "step": 15315
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.23547409382974e-08,
+      "loss": 0.8501,
+      "step": 15316
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.131121869379363e-08,
+      "loss": 0.9054,
+      "step": 15317
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.027650075195456e-08,
+      "loss": 0.853,
+      "step": 15318
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.9250587158299344e-08,
+      "loss": 0.9069,
+      "step": 15319
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.8233477958036244e-08,
+      "loss": 0.8665,
+      "step": 15320
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.7225173195951666e-08,
+      "loss": 0.8051,
+      "step": 15321
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.622567291641012e-08,
+      "loss": 0.8051,
+      "step": 15322
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.5234977163487464e-08,
+      "loss": 0.7988,
+      "step": 15323
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.425308598074885e-08,
+      "loss": 0.7688,
+      "step": 15324
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.327999941147077e-08,
+      "loss": 0.8972,
+      "step": 15325
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.231571749848563e-08,
+      "loss": 0.8691,
+      "step": 15326
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.136024028424835e-08,
+      "loss": 0.8611,
+      "step": 15327
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.04135678108586e-08,
+      "loss": 0.8636,
+      "step": 15328
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.9475700119971936e-08,
+      "loss": 0.8751,
+      "step": 15329
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.854663725291087e-08,
+      "loss": 0.8098,
+      "step": 15330
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.762637925059821e-08,
+      "loss": 0.9194,
+      "step": 15331
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.671492615351269e-08,
+      "loss": 0.8059,
+      "step": 15332
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.58122780018444e-08,
+      "loss": 0.8562,
+      "step": 15333
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.4918434835294895e-08,
+      "loss": 0.8819,
+      "step": 15334
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.4033396693254905e-08,
+      "loss": 0.8432,
+      "step": 15335
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.3157163614671034e-08,
+      "loss": 0.8603,
+      "step": 15336
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.228973563815686e-08,
+      "loss": 0.8478,
+      "step": 15337
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.1431112801881834e-08,
+      "loss": 0.8973,
+      "step": 15338
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.058129514368236e-08,
+      "loss": 0.8293,
+      "step": 15339
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.974028270095076e-08,
+      "loss": 0.7684,
+      "step": 15340
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.890807551074627e-08,
+      "loss": 0.8315,
+      "step": 15341
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.8084673609706245e-08,
+      "loss": 0.9186,
+      "step": 15342
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.727007703409058e-08,
+      "loss": 0.8605,
+      "step": 15343
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.646428581978168e-08,
+      "loss": 0.8855,
+      "step": 15344
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.566730000224006e-08,
+      "loss": 0.8988,
+      "step": 15345
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.487911961654877e-08,
+      "loss": 0.8267,
+      "step": 15346
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.40997446974578e-08,
+      "loss": 0.8255,
+      "step": 15347
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.332917527927304e-08,
+      "loss": 0.9011,
+      "step": 15348
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.256741139590069e-08,
+      "loss": 0.8301,
+      "step": 15349
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.1814453080913906e-08,
+      "loss": 0.8993,
+      "step": 15350
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.1070300367463945e-08,
+      "loss": 0.8695,
+      "step": 15351
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.033495328830238e-08,
+      "loss": 0.8515,
+      "step": 15352
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.960841187584773e-08,
+      "loss": 0.8773,
+      "step": 15353
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.8890676162030008e-08,
+      "loss": 0.8703,
+      "step": 15354
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.8181746178534973e-08,
+      "loss": 0.8567,
+      "step": 15355
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.7481621956515492e-08,
+      "loss": 0.8208,
+      "step": 15356
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.6790303526813555e-08,
+      "loss": 0.8703,
+      "step": 15357
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.6107790919893683e-08,
+      "loss": 0.8015,
+      "step": 15358
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.5434084165798512e-08,
+      "loss": 0.8091,
+      "step": 15359
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.4769183294193198e-08,
+      "loss": 0.8803,
+      "step": 15360
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.411308833436543e-08,
+      "loss": 0.8577,
+      "step": 15361
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.3465799315203206e-08,
+      "loss": 0.8377,
+      "step": 15362
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.2827316265194853e-08,
+      "loss": 0.8496,
+      "step": 15363
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.219763921247342e-08,
+      "loss": 0.8626,
+      "step": 15364
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.1576768184750073e-08,
+      "loss": 0.9083,
+      "step": 15365
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.096470320938071e-08,
+      "loss": 0.8153,
+      "step": 15366
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.0361444313321542e-08,
+      "loss": 0.8524,
+      "step": 15367
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9766991523129108e-08,
+      "loss": 0.8833,
+      "step": 15368
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.918134486496026e-08,
+      "loss": 0.8995,
+      "step": 15369
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8604504364638788e-08,
+      "loss": 0.8509,
+      "step": 15370
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8036470047544384e-08,
+      "loss": 0.8865,
+      "step": 15371
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.7477241938701482e-08,
+      "loss": 0.82,
+      "step": 15372
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.6926820062712624e-08,
+      "loss": 0.7833,
+      "step": 15373
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.638520444384728e-08,
+      "loss": 0.848,
+      "step": 15374
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5852395105953044e-08,
+      "loss": 0.8842,
+      "step": 15375
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.532839207245562e-08,
+      "loss": 0.8343,
+      "step": 15376
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4813195366469856e-08,
+      "loss": 0.9153,
+      "step": 15377
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4306805010666502e-08,
+      "loss": 0.8245,
+      "step": 15378
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3809221027338836e-08,
+      "loss": 0.826,
+      "step": 15379
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3320443438402663e-08,
+      "loss": 0.8093,
+      "step": 15380
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2840472265374103e-08,
+      "loss": 0.8838,
+      "step": 15381
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.236930752941401e-08,
+      "loss": 0.9673,
+      "step": 15382
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1906949251261346e-08,
+      "loss": 0.7917,
+      "step": 15383
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1453397451233194e-08,
+      "loss": 0.8888,
+      "step": 15384
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1008652149357979e-08,
+      "loss": 0.862,
+      "step": 15385
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0572713365197828e-08,
+      "loss": 0.8312,
+      "step": 15386
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0145581117959602e-08,
+      "loss": 0.891,
+      "step": 15387
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.727255426428272e-09,
+      "loss": 0.9174,
+      "step": 15388
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.317736309055746e-09,
+      "loss": 0.8605,
+      "step": 15389
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.917023783849844e-09,
+      "loss": 0.856,
+      "step": 15390
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.525117868463105e-09,
+      "loss": 0.9057,
+      "step": 15391
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.142018580148403e-09,
+      "loss": 0.7962,
+      "step": 15392
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.767725935781123e-09,
+      "loss": 0.8594,
+      "step": 15393
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.40223995185918e-09,
+      "loss": 0.8357,
+      "step": 15394
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.045560644436399e-09,
+      "loss": 0.8461,
+      "step": 15395
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.697688029255744e-09,
+      "loss": 0.8207,
+      "step": 15396
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.358622121638292e-09,
+      "loss": 0.8612,
+      "step": 15397
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.028362936483234e-09,
+      "loss": 0.8607,
+      "step": 15398
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.706910488356698e-09,
+      "loss": 0.8325,
+      "step": 15399
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.39426479142513e-09,
+      "loss": 0.9113,
+      "step": 15400
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.090425859410886e-09,
+      "loss": 0.7796,
+      "step": 15401
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.7953937057476635e-09,
+      "loss": 0.8919,
+      "step": 15402
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.509168343380665e-09,
+      "loss": 0.8544,
+      "step": 15403
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.2317497849442275e-09,
+      "loss": 0.8887,
+      "step": 15404
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.9631380426508045e-09,
+      "loss": 0.8289,
+      "step": 15405
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.703333128313169e-09,
+      "loss": 0.904,
+      "step": 15406
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.452335053366618e-09,
+      "loss": 0.8369,
+      "step": 15407
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.2101438289133814e-09,
+      "loss": 0.8206,
+      "step": 15408
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.976759465544987e-09,
+      "loss": 0.8131,
+      "step": 15409
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.752181973608714e-09,
+      "loss": 0.8532,
+      "step": 15410
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.536411362941138e-09,
+      "loss": 0.7975,
+      "step": 15411
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.3294476430679722e-09,
+      "loss": 0.8779,
+      "step": 15412
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.1312908231152506e-09,
+      "loss": 0.8636,
+      "step": 15413
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9419409117871214e-09,
+      "loss": 0.8678,
+      "step": 15414
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.761397917410257e-09,
+      "loss": 0.8539,
+      "step": 15415
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.5896618479782633e-09,
+      "loss": 0.8301,
+      "step": 15416
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4267327110184524e-09,
+      "loss": 0.8633,
+      "step": 15417
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.2726105137028653e-09,
+      "loss": 0.8454,
+      "step": 15418
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.127295262826067e-09,
+      "loss": 0.8545,
+      "step": 15419
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 9.907869648051459e-10,
+      "loss": 0.8877,
+      "step": 15420
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.63085625635307e-10,
+      "loss": 0.8724,
+      "step": 15421
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.441912509342786e-10,
+      "loss": 0.904,
+      "step": 15422
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.341038459645177e-10,
+      "loss": 0.8321,
+      "step": 15423
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.328234155221879e-10,
+      "loss": 0.8525,
+      "step": 15424
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.4034996411479456e-10,
+      "loss": 0.8294,
+      "step": 15425
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.566834958057541e-10,
+      "loss": 0.7848,
+      "step": 15426
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.818240142810069e-10,
+      "loss": 0.8639,
+      "step": 15427
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.1577152280460867e-10,
+      "loss": 0.8884,
+      "step": 15428
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.5852602430754816e-10,
+      "loss": 0.8169,
+      "step": 15429
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.1008752132113386e-10,
+      "loss": 0.8747,
+      "step": 15430
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.045601597699402e-11,
+      "loss": 0.8418,
+      "step": 15431
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.9631510007076545e-11,
+      "loss": 0.8474,
+      "step": 15432
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.761400476585351e-11,
+      "loss": 0.8337,
+      "step": 15433
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.403501230321183e-12,
+      "loss": 0.8442,
+      "step": 15434
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 0.9265,
+      "step": 15435
+    },
+    {
+      "epoch": 3.0,
+      "step": 15435,
+      "total_flos": 1.7492833725644276e+19,
+      "train_loss": 0.6371766319080275,
+      "train_runtime": 68889.0709,
+      "train_samples_per_second": 28.68,
+      "train_steps_per_second": 0.224
+    }
+  ],
+  "max_steps": 15435,
+  "num_train_epochs": 3,
+  "total_flos": 1.7492833725644276e+19,
+  "trial_name": null,
+  "trial_params": null
+}