diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,29275 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.9999230828397816,
+  "global_step": 4875,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.360544217687075e-07,
+      "loss": 2.2054,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.72108843537415e-07,
+      "loss": 1.9317,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0816326530612243e-07,
+      "loss": 1.8381,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.4421768707483e-07,
+      "loss": 1.8925,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.802721088435376e-07,
+      "loss": 2.0668,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.163265306122449e-07,
+      "loss": 1.9225,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.523809523809525e-07,
+      "loss": 1.9997,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.08843537414966e-06,
+      "loss": 1.9344,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2244897959183673e-06,
+      "loss": 1.8514,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3605442176870751e-06,
+      "loss": 1.9085,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4965986394557825e-06,
+      "loss": 1.8013,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6326530612244897e-06,
+      "loss": 1.8675,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7687074829931975e-06,
+      "loss": 1.8931,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.904761904761905e-06,
+      "loss": 1.6981,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0408163265306125e-06,
+      "loss": 1.8269,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.17687074829932e-06,
+      "loss": 1.9008,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3129251700680273e-06,
+      "loss": 1.8196,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4489795918367347e-06,
+      "loss": 1.7567,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5850340136054425e-06,
+      "loss": 1.8177,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7210884353741503e-06,
+      "loss": 1.842,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 1.9637,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.993197278911565e-06,
+      "loss": 1.763,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1292517006802725e-06,
+      "loss": 1.7435,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2653061224489794e-06,
+      "loss": 1.69,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4013605442176872e-06,
+      "loss": 1.6571,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.537414965986395e-06,
+      "loss": 1.8343,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6734693877551024e-06,
+      "loss": 1.7249,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.80952380952381e-06,
+      "loss": 1.9437,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.945578231292517e-06,
+      "loss": 1.6511,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.081632653061225e-06,
+      "loss": 1.696,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.217687074829933e-06,
+      "loss": 1.8953,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.35374149659864e-06,
+      "loss": 1.7559,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.489795918367348e-06,
+      "loss": 1.6156,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6258503401360546e-06,
+      "loss": 1.7349,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.761904761904762e-06,
+      "loss": 1.6926,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.897959183673469e-06,
+      "loss": 1.7203,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.034013605442177e-06,
+      "loss": 1.5929,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.170068027210885e-06,
+      "loss": 1.7476,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.306122448979593e-06,
+      "loss": 1.7339,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.442176870748301e-06,
+      "loss": 1.7675,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.578231292517007e-06,
+      "loss": 1.8059,
+      "step": 41
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.7142857142857145e-06,
+      "loss": 1.5962,
+      "step": 42
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.850340136054422e-06,
+      "loss": 1.7082,
+      "step": 43
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.98639455782313e-06,
+      "loss": 1.753,
+      "step": 44
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.122448979591837e-06,
+      "loss": 1.5706,
+      "step": 45
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.258503401360545e-06,
+      "loss": 1.6913,
+      "step": 46
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.394557823129253e-06,
+      "loss": 1.7386,
+      "step": 47
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.530612244897959e-06,
+      "loss": 1.6239,
+      "step": 48
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.5737,
+      "step": 49
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.8027210884353745e-06,
+      "loss": 1.7684,
+      "step": 50
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.938775510204082e-06,
+      "loss": 1.7642,
+      "step": 51
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.07482993197279e-06,
+      "loss": 1.3886,
+      "step": 52
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.210884353741497e-06,
+      "loss": 1.6409,
+      "step": 53
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.346938775510205e-06,
+      "loss": 1.6673,
+      "step": 54
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.482993197278913e-06,
+      "loss": 1.7243,
+      "step": 55
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.61904761904762e-06,
+      "loss": 1.567,
+      "step": 56
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.755102040816327e-06,
+      "loss": 1.6037,
+      "step": 57
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.891156462585034e-06,
+      "loss": 1.7143,
+      "step": 58
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.027210884353741e-06,
+      "loss": 1.7218,
+      "step": 59
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.16326530612245e-06,
+      "loss": 1.7142,
+      "step": 60
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.299319727891157e-06,
+      "loss": 1.6352,
+      "step": 61
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.435374149659866e-06,
+      "loss": 1.7653,
+      "step": 62
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.571428571428571e-06,
+      "loss": 1.7618,
+      "step": 63
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.70748299319728e-06,
+      "loss": 1.6692,
+      "step": 64
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.843537414965987e-06,
+      "loss": 1.5482,
+      "step": 65
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.979591836734695e-06,
+      "loss": 1.8431,
+      "step": 66
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.115646258503402e-06,
+      "loss": 1.7586,
+      "step": 67
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.251700680272109e-06,
+      "loss": 1.7047,
+      "step": 68
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.387755102040818e-06,
+      "loss": 1.7356,
+      "step": 69
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.523809523809525e-06,
+      "loss": 1.703,
+      "step": 70
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.659863945578232e-06,
+      "loss": 1.5398,
+      "step": 71
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.795918367346939e-06,
+      "loss": 1.7267,
+      "step": 72
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.931972789115647e-06,
+      "loss": 1.5982,
+      "step": 73
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0068027210884354e-05,
+      "loss": 1.7102,
+      "step": 74
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0204081632653063e-05,
+      "loss": 1.5244,
+      "step": 75
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.034013605442177e-05,
+      "loss": 1.6446,
+      "step": 76
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0476190476190477e-05,
+      "loss": 1.9369,
+      "step": 77
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0612244897959186e-05,
+      "loss": 1.6802,
+      "step": 78
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0748299319727893e-05,
+      "loss": 1.6758,
+      "step": 79
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0884353741496601e-05,
+      "loss": 1.7152,
+      "step": 80
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1020408163265306e-05,
+      "loss": 1.706,
+      "step": 81
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1156462585034013e-05,
+      "loss": 1.6785,
+      "step": 82
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1292517006802722e-05,
+      "loss": 1.5432,
+      "step": 83
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1428571428571429e-05,
+      "loss": 1.5693,
+      "step": 84
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1564625850340136e-05,
+      "loss": 1.8198,
+      "step": 85
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1700680272108845e-05,
+      "loss": 1.6864,
+      "step": 86
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1836734693877552e-05,
+      "loss": 1.6344,
+      "step": 87
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.197278911564626e-05,
+      "loss": 1.5266,
+      "step": 88
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2108843537414967e-05,
+      "loss": 1.8137,
+      "step": 89
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2244897959183674e-05,
+      "loss": 1.7093,
+      "step": 90
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2380952380952383e-05,
+      "loss": 1.6609,
+      "step": 91
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.251700680272109e-05,
+      "loss": 1.6753,
+      "step": 92
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2653061224489798e-05,
+      "loss": 1.6357,
+      "step": 93
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2789115646258505e-05,
+      "loss": 1.6906,
+      "step": 94
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2925170068027212e-05,
+      "loss": 1.6901,
+      "step": 95
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3061224489795918e-05,
+      "loss": 1.4528,
+      "step": 96
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3197278911564626e-05,
+      "loss": 1.6027,
+      "step": 97
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 1.846,
+      "step": 98
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3469387755102042e-05,
+      "loss": 1.694,
+      "step": 99
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3605442176870749e-05,
+      "loss": 1.4134,
+      "step": 100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3741496598639456e-05,
+      "loss": 1.6667,
+      "step": 101
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3877551020408165e-05,
+      "loss": 1.6349,
+      "step": 102
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4013605442176872e-05,
+      "loss": 1.7919,
+      "step": 103
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.414965986394558e-05,
+      "loss": 1.5753,
+      "step": 104
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4285714285714287e-05,
+      "loss": 1.7043,
+      "step": 105
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4421768707482994e-05,
+      "loss": 1.5703,
+      "step": 106
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4557823129251703e-05,
+      "loss": 1.7664,
+      "step": 107
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.469387755102041e-05,
+      "loss": 1.6487,
+      "step": 108
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4829931972789118e-05,
+      "loss": 1.6095,
+      "step": 109
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4965986394557825e-05,
+      "loss": 1.8305,
+      "step": 110
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.510204081632653e-05,
+      "loss": 1.6448,
+      "step": 111
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.523809523809524e-05,
+      "loss": 1.6334,
+      "step": 112
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5374149659863945e-05,
+      "loss": 1.7572,
+      "step": 113
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5510204081632655e-05,
+      "loss": 1.5345,
+      "step": 114
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5646258503401362e-05,
+      "loss": 1.5893,
+      "step": 115
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.578231292517007e-05,
+      "loss": 1.7515,
+      "step": 116
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5918367346938776e-05,
+      "loss": 1.5527,
+      "step": 117
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6054421768707483e-05,
+      "loss": 1.6173,
+      "step": 118
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6190476190476193e-05,
+      "loss": 1.6438,
+      "step": 119
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.63265306122449e-05,
+      "loss": 1.5747,
+      "step": 120
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6462585034013607e-05,
+      "loss": 1.4945,
+      "step": 121
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6598639455782314e-05,
+      "loss": 1.6815,
+      "step": 122
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.673469387755102e-05,
+      "loss": 1.6926,
+      "step": 123
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.687074829931973e-05,
+      "loss": 1.5884,
+      "step": 124
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7006802721088435e-05,
+      "loss": 1.6569,
+      "step": 125
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7142857142857142e-05,
+      "loss": 1.5692,
+      "step": 126
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7278911564625852e-05,
+      "loss": 1.738,
+      "step": 127
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.741496598639456e-05,
+      "loss": 1.7065,
+      "step": 128
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7551020408163266e-05,
+      "loss": 1.4979,
+      "step": 129
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7687074829931973e-05,
+      "loss": 1.6857,
+      "step": 130
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.782312925170068e-05,
+      "loss": 1.8462,
+      "step": 131
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.795918367346939e-05,
+      "loss": 1.6003,
+      "step": 132
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8095238095238097e-05,
+      "loss": 1.531,
+      "step": 133
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8231292517006804e-05,
+      "loss": 1.4869,
+      "step": 134
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.836734693877551e-05,
+      "loss": 1.7549,
+      "step": 135
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8503401360544218e-05,
+      "loss": 1.5493,
+      "step": 136
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.863945578231293e-05,
+      "loss": 1.5068,
+      "step": 137
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8775510204081636e-05,
+      "loss": 1.5271,
+      "step": 138
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8911564625850343e-05,
+      "loss": 1.6093,
+      "step": 139
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.904761904761905e-05,
+      "loss": 1.6797,
+      "step": 140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9183673469387756e-05,
+      "loss": 1.759,
+      "step": 141
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9319727891156463e-05,
+      "loss": 1.4956,
+      "step": 142
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.945578231292517e-05,
+      "loss": 1.5532,
+      "step": 143
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9591836734693877e-05,
+      "loss": 1.816,
+      "step": 144
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9727891156462588e-05,
+      "loss": 1.5857,
+      "step": 145
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9863945578231295e-05,
+      "loss": 1.7439,
+      "step": 146
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 1.7172,
+      "step": 147
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999997792428403e-05,
+      "loss": 1.4999,
+      "step": 148
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999991169714585e-05,
+      "loss": 1.7715,
+      "step": 149
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999998013186146e-05,
+      "loss": 1.5687,
+      "step": 150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999996467887392e-05,
+      "loss": 1.6936,
+      "step": 151
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999944810758777e-05,
+      "loss": 1.5857,
+      "step": 152
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.99999205275248e-05,
+      "loss": 1.6094,
+      "step": 153
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999891829182717e-05,
+      "loss": 1.6897,
+      "step": 154
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999858715745195e-05,
+      "loss": 1.7484,
+      "step": 155
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999821187226853e-05,
+      "loss": 1.7313,
+      "step": 156
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999779243644264e-05,
+      "loss": 1.6879,
+      "step": 157
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999732885015946e-05,
+      "loss": 1.6293,
+      "step": 158
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999682111362368e-05,
+      "loss": 1.7648,
+      "step": 159
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999626922705944e-05,
+      "loss": 1.5696,
+      "step": 160
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999956731907104e-05,
+      "loss": 1.5775,
+      "step": 161
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999503300483978e-05,
+      "loss": 1.6914,
+      "step": 162
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999434866973018e-05,
+      "loss": 1.6716,
+      "step": 163
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999362018568375e-05,
+      "loss": 1.5525,
+      "step": 164
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999284755302213e-05,
+      "loss": 1.5314,
+      "step": 165
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999203077208644e-05,
+      "loss": 1.5227,
+      "step": 166
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999911698432373e-05,
+      "loss": 1.7998,
+      "step": 167
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999026476685484e-05,
+      "loss": 1.5992,
+      "step": 168
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999893155433387e-05,
+      "loss": 1.6339,
+      "step": 169
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999883221731079e-05,
+      "loss": 1.5137,
+      "step": 170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998728465660105e-05,
+      "loss": 1.4989,
+      "step": 171
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998620299427624e-05,
+      "loss": 1.6107,
+      "step": 172
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998507718661104e-05,
+      "loss": 1.629,
+      "step": 173
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999839072341025e-05,
+      "loss": 1.6199,
+      "step": 174
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998269313726722e-05,
+      "loss": 1.6348,
+      "step": 175
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998143489664114e-05,
+      "loss": 1.4972,
+      "step": 176
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998013251277988e-05,
+      "loss": 1.6121,
+      "step": 177
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997878598625845e-05,
+      "loss": 1.7005,
+      "step": 178
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997739531767132e-05,
+      "loss": 1.6006,
+      "step": 179
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997596050763253e-05,
+      "loss": 1.5954,
+      "step": 180
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999744815567755e-05,
+      "loss": 1.7201,
+      "step": 181
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999729584657533e-05,
+      "loss": 1.5947,
+      "step": 182
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999713912352384e-05,
+      "loss": 1.6259,
+      "step": 183
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996977986592263e-05,
+      "loss": 1.7148,
+      "step": 184
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996812435851754e-05,
+      "loss": 1.438,
+      "step": 185
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996642471375406e-05,
+      "loss": 1.6635,
+      "step": 186
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996468093238256e-05,
+      "loss": 1.5015,
+      "step": 187
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.99962893015173e-05,
+      "loss": 1.5034,
+      "step": 188
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999610609629147e-05,
+      "loss": 1.4632,
+      "step": 189
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995918477641657e-05,
+      "loss": 1.6049,
+      "step": 190
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995726445650698e-05,
+      "loss": 1.8895,
+      "step": 191
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995530000403378e-05,
+      "loss": 1.5521,
+      "step": 192
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999532914198643e-05,
+      "loss": 1.7053,
+      "step": 193
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995123870488537e-05,
+      "loss": 1.7173,
+      "step": 194
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999491418600033e-05,
+      "loss": 1.6423,
+      "step": 195
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999470008861438e-05,
+      "loss": 1.7326,
+      "step": 196
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994481578425225e-05,
+      "loss": 1.9087,
+      "step": 197
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994258655529336e-05,
+      "loss": 1.5748,
+      "step": 198
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999403132002513e-05,
+      "loss": 1.6909,
+      "step": 199
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999379957201299e-05,
+      "loss": 1.5826,
+      "step": 200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999356341159523e-05,
+      "loss": 1.667,
+      "step": 201
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993322838876116e-05,
+      "loss": 1.6582,
+      "step": 202
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993077853961874e-05,
+      "loss": 1.5196,
+      "step": 203
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992828456960654e-05,
+      "loss": 1.6064,
+      "step": 204
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999257464798258e-05,
+      "loss": 1.5091,
+      "step": 205
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999231642713971e-05,
+      "loss": 1.5139,
+      "step": 206
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999205379454605e-05,
+      "loss": 1.7734,
+      "step": 207
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991786750317555e-05,
+      "loss": 1.5538,
+      "step": 208
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999151529457213e-05,
+      "loss": 1.4208,
+      "step": 209
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999123942742963e-05,
+      "loss": 1.7543,
+      "step": 210
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999095914901185e-05,
+      "loss": 1.6356,
+      "step": 211
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990674459442536e-05,
+      "loss": 1.5622,
+      "step": 212
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990385358847385e-05,
+      "loss": 1.6637,
+      "step": 213
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990091847354043e-05,
+      "loss": 1.5312,
+      "step": 214
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989793925092092e-05,
+      "loss": 1.4405,
+      "step": 215
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989491592193074e-05,
+      "loss": 1.5693,
+      "step": 216
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998918484879047e-05,
+      "loss": 1.6806,
+      "step": 217
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988873695019716e-05,
+      "loss": 1.5026,
+      "step": 218
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988558131018188e-05,
+      "loss": 1.664,
+      "step": 219
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998823815692521e-05,
+      "loss": 1.5622,
+      "step": 220
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987913772882058e-05,
+      "loss": 1.5389,
+      "step": 221
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987584979031954e-05,
+      "loss": 1.7978,
+      "step": 222
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987251775520056e-05,
+      "loss": 1.6628,
+      "step": 223
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998691416249349e-05,
+      "loss": 1.6411,
+      "step": 224
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986572140101312e-05,
+      "loss": 1.6526,
+      "step": 225
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998622570849453e-05,
+      "loss": 1.6851,
+      "step": 226
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985874867826095e-05,
+      "loss": 1.4972,
+      "step": 227
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985519618250912e-05,
+      "loss": 1.3894,
+      "step": 228
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998515995992583e-05,
+      "loss": 1.7018,
+      "step": 229
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984795893009643e-05,
+      "loss": 1.5166,
+      "step": 230
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984427417663085e-05,
+      "loss": 1.605,
+      "step": 231
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984054534048854e-05,
+      "loss": 1.6448,
+      "step": 232
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983677242331573e-05,
+      "loss": 1.6911,
+      "step": 233
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998329554267783e-05,
+      "loss": 1.6296,
+      "step": 234
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9982909435256143e-05,
+      "loss": 1.7005,
+      "step": 235
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998251892023699e-05,
+      "loss": 1.7285,
+      "step": 236
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998212399779279e-05,
+      "loss": 1.4675,
+      "step": 237
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.99817246680979e-05,
+      "loss": 1.5663,
+      "step": 238
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998132093132864e-05,
+      "loss": 1.4669,
+      "step": 239
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980912787663253e-05,
+      "loss": 1.663,
+      "step": 240
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980500237281948e-05,
+      "loss": 1.4481,
+      "step": 241
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980083280366874e-05,
+      "loss": 1.7369,
+      "step": 242
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979661917102116e-05,
+      "loss": 1.5867,
+      "step": 243
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979236147673718e-05,
+      "loss": 1.6083,
+      "step": 244
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997880597226966e-05,
+      "loss": 1.6638,
+      "step": 245
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978371391079874e-05,
+      "loss": 1.5065,
+      "step": 246
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997793240429623e-05,
+      "loss": 1.4402,
+      "step": 247
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997748901211255e-05,
+      "loss": 1.7069,
+      "step": 248
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977041214724594e-05,
+      "loss": 1.5265,
+      "step": 249
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997658901233008e-05,
+      "loss": 1.6213,
+      "step": 250
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976132405128646e-05,
+      "loss": 1.6345,
+      "step": 251
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997567139332191e-05,
+      "loss": 1.6033,
+      "step": 252
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9975205977113398e-05,
+      "loss": 1.6856,
+      "step": 253
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997473615670861e-05,
+      "loss": 1.8683,
+      "step": 254
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997426193231497e-05,
+      "loss": 1.824,
+      "step": 255
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997378330414186e-05,
+      "loss": 1.641,
+      "step": 256
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.99733002724006e-05,
+      "loss": 1.7038,
+      "step": 257
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972812837304454e-05,
+      "loss": 1.4866,
+      "step": 258
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972320999068636e-05,
+      "loss": 1.7403,
+      "step": 259
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971824757910295e-05,
+      "loss": 1.5893,
+      "step": 260
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971324114048533e-05,
+      "loss": 1.7671,
+      "step": 261
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997081906770438e-05,
+      "loss": 1.7711,
+      "step": 262
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997030961910084e-05,
+      "loss": 1.6466,
+      "step": 263
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969795768462824e-05,
+      "loss": 1.4688,
+      "step": 264
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969277516017214e-05,
+      "loss": 1.662,
+      "step": 265
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996875486199282e-05,
+      "loss": 1.651,
+      "step": 266
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968227806620413e-05,
+      "loss": 1.5727,
+      "step": 267
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9967696350132684e-05,
+      "loss": 1.7939,
+      "step": 268
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996716049276428e-05,
+      "loss": 1.6602,
+      "step": 269
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966620234751794e-05,
+      "loss": 1.6687,
+      "step": 270
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966075576333757e-05,
+      "loss": 1.6927,
+      "step": 271
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996552651775064e-05,
+      "loss": 1.6387,
+      "step": 272
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996497305924486e-05,
+      "loss": 1.6943,
+      "step": 273
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964415201060782e-05,
+      "loss": 1.3389,
+      "step": 274
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963852943444703e-05,
+      "loss": 1.5606,
+      "step": 275
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963286286644875e-05,
+      "loss": 1.6758,
+      "step": 276
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962715230911477e-05,
+      "loss": 1.6869,
+      "step": 277
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996213977649664e-05,
+      "loss": 1.6984,
+      "step": 278
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996155992365444e-05,
+      "loss": 1.4167,
+      "step": 279
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9960975672640887e-05,
+      "loss": 1.7717,
+      "step": 280
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9960387023713935e-05,
+      "loss": 1.6857,
+      "step": 281
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959793977133483e-05,
+      "loss": 1.5171,
+      "step": 282
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.995919653316137e-05,
+      "loss": 1.4732,
+      "step": 283
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9958594692061374e-05,
+      "loss": 1.5648,
+      "step": 284
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995798845409922e-05,
+      "loss": 1.5543,
+      "step": 285
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957377819542568e-05,
+      "loss": 1.776,
+      "step": 286
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956762788661018e-05,
+      "loss": 1.7748,
+      "step": 287
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956143361726123e-05,
+      "loss": 1.7336,
+      "step": 288
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955519539011362e-05,
+      "loss": 1.6824,
+      "step": 289
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954891320792166e-05,
+      "loss": 1.7506,
+      "step": 290
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954258707345903e-05,
+      "loss": 1.5711,
+      "step": 291
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953621698951875e-05,
+      "loss": 1.6099,
+      "step": 292
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9952980295891335e-05,
+      "loss": 1.7158,
+      "step": 293
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9952334498447473e-05,
+      "loss": 1.4926,
+      "step": 294
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951684306905418e-05,
+      "loss": 1.747,
+      "step": 295
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995102972155223e-05,
+      "loss": 1.6313,
+      "step": 296
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995037074267693e-05,
+      "loss": 1.7343,
+      "step": 297
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949707370570456e-05,
+      "loss": 1.6337,
+      "step": 298
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949039605525705e-05,
+      "loss": 1.7952,
+      "step": 299
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9948367447837503e-05,
+      "loss": 1.5182,
+      "step": 300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994769089780261e-05,
+      "loss": 1.6427,
+      "step": 301
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994700995571974e-05,
+      "loss": 1.7528,
+      "step": 302
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946324621889538e-05,
+      "loss": 1.7347,
+      "step": 303
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9945634896614584e-05,
+      "loss": 1.776,
+      "step": 304
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944940780199407e-05,
+      "loss": 1.7954,
+      "step": 305
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944242272950468e-05,
+      "loss": 1.7076,
+      "step": 306
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943539375176164e-05,
+      "loss": 1.5313,
+      "step": 307
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942832087186842e-05,
+      "loss": 1.7238,
+      "step": 308
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942120409294768e-05,
+      "loss": 1.5554,
+      "step": 309
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9941404341814175e-05,
+      "loss": 1.5804,
+      "step": 310
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99406838850612e-05,
+      "loss": 1.6121,
+      "step": 311
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939959039353947e-05,
+      "loss": 1.6604,
+      "step": 312
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939229805012435e-05,
+      "loss": 1.5501,
+      "step": 313
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938496182358643e-05,
+      "loss": 1.6636,
+      "step": 314
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937758171716468e-05,
+      "loss": 1.6708,
+      "step": 315
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937015773411757e-05,
+      "loss": 1.6708,
+      "step": 316
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9936268987772286e-05,
+      "loss": 1.8348,
+      "step": 317
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993551781512777e-05,
+      "loss": 1.5023,
+      "step": 318
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934762255809866e-05,
+      "loss": 1.8314,
+      "step": 319
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934002310152167e-05,
+      "loss": 1.6247,
+      "step": 320
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933237978490193e-05,
+      "loss": 1.5488,
+      "step": 321
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932469261161414e-05,
+      "loss": 1.7784,
+      "step": 322
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993169615850522e-05,
+      "loss": 1.7966,
+      "step": 323
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993091867086296e-05,
+      "loss": 1.6944,
+      "step": 324
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930136798577898e-05,
+      "loss": 1.49,
+      "step": 325
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9929350541995237e-05,
+      "loss": 1.588,
+      "step": 326
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992855990146213e-05,
+      "loss": 1.6606,
+      "step": 327
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9927764877327654e-05,
+      "loss": 1.5651,
+      "step": 328
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9926965469942826e-05,
+      "loss": 1.5676,
+      "step": 329
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9926161679660588e-05,
+      "loss": 1.6222,
+      "step": 330
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9925353506835827e-05,
+      "loss": 1.6568,
+      "step": 331
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9924540951825368e-05,
+      "loss": 1.7568,
+      "step": 332
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9923724014987963e-05,
+      "loss": 1.5687,
+      "step": 333
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9922902696684295e-05,
+      "loss": 1.6,
+      "step": 334
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9922076997277e-05,
+      "loss": 1.6941,
+      "step": 335
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9921246917130627e-05,
+      "loss": 1.6185,
+      "step": 336
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992041245661167e-05,
+      "loss": 1.5062,
+      "step": 337
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9919573616088558e-05,
+      "loss": 1.5925,
+      "step": 338
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9918730395931648e-05,
+      "loss": 1.6455,
+      "step": 339
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9917882796513238e-05,
+      "loss": 1.6393,
+      "step": 340
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991703081820755e-05,
+      "loss": 1.7708,
+      "step": 341
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991617446139075e-05,
+      "loss": 1.6145,
+      "step": 342
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9915313726440922e-05,
+      "loss": 1.5781,
+      "step": 343
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9914448613738107e-05,
+      "loss": 1.6344,
+      "step": 344
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9913579123664253e-05,
+      "loss": 1.7069,
+      "step": 345
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9912705256603258e-05,
+      "loss": 1.6422,
+      "step": 346
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9911827012940944e-05,
+      "loss": 1.6068,
+      "step": 347
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9910944393065073e-05,
+      "loss": 1.6164,
+      "step": 348
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9910057397365328e-05,
+      "loss": 1.5468,
+      "step": 349
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990916602623333e-05,
+      "loss": 1.6608,
+      "step": 350
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9908270280062643e-05,
+      "loss": 1.5416,
+      "step": 351
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990737015924874e-05,
+      "loss": 1.6495,
+      "step": 352
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990646566418904e-05,
+      "loss": 1.6003,
+      "step": 353
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9905556795282894e-05,
+      "loss": 1.8159,
+      "step": 354
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9904643552931578e-05,
+      "loss": 1.7428,
+      "step": 355
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.99037259375383e-05,
+      "loss": 1.6802,
+      "step": 356
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9902803949508206e-05,
+      "loss": 1.5533,
+      "step": 357
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990187758924836e-05,
+      "loss": 1.6668,
+      "step": 358
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9900946857167768e-05,
+      "loss": 1.6223,
+      "step": 359
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9900011753677363e-05,
+      "loss": 1.7831,
+      "step": 360
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989907227919e-05,
+      "loss": 1.7085,
+      "step": 361
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9898128434120477e-05,
+      "loss": 1.7375,
+      "step": 362
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989718021888551e-05,
+      "loss": 1.4788,
+      "step": 363
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9896227633903752e-05,
+      "loss": 1.8388,
+      "step": 364
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989527067959578e-05,
+      "loss": 1.6895,
+      "step": 365
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989430935638411e-05,
+      "loss": 1.7023,
+      "step": 366
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9893343664693177e-05,
+      "loss": 1.5976,
+      "step": 367
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989237360494934e-05,
+      "loss": 1.6778,
+      "step": 368
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989139917758091e-05,
+      "loss": 1.554,
+      "step": 369
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9890420383018094e-05,
+      "loss": 1.5968,
+      "step": 370
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9889437221693053e-05,
+      "loss": 1.6417,
+      "step": 371
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9888449694039866e-05,
+      "loss": 1.5951,
+      "step": 372
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988745780049454e-05,
+      "loss": 1.5368,
+      "step": 373
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9886461541495007e-05,
+      "loss": 1.5071,
+      "step": 374
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9885460917481137e-05,
+      "loss": 1.6209,
+      "step": 375
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988445592889471e-05,
+      "loss": 1.6357,
+      "step": 376
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9883446576179454e-05,
+      "loss": 1.6914,
+      "step": 377
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9882432859781002e-05,
+      "loss": 1.8024,
+      "step": 378
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988141478014693e-05,
+      "loss": 1.6189,
+      "step": 379
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9880392337726736e-05,
+      "loss": 1.6457,
+      "step": 380
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.987936553297184e-05,
+      "loss": 1.6095,
+      "step": 381
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987833436633559e-05,
+      "loss": 1.4077,
+      "step": 382
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9877298838273263e-05,
+      "loss": 1.6502,
+      "step": 383
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987625894924206e-05,
+      "loss": 1.5356,
+      "step": 384
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9875214699701108e-05,
+      "loss": 1.7991,
+      "step": 385
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987416609011145e-05,
+      "loss": 1.6621,
+      "step": 386
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9873113120936074e-05,
+      "loss": 1.5659,
+      "step": 387
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9872055792639875e-05,
+      "loss": 1.4915,
+      "step": 388
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9870994105689675e-05,
+      "loss": 1.5945,
+      "step": 389
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9869928060554234e-05,
+      "loss": 1.7863,
+      "step": 390
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9868857657704215e-05,
+      "loss": 1.6172,
+      "step": 391
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986778289761222e-05,
+      "loss": 1.6226,
+      "step": 392
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9866703780752777e-05,
+      "loss": 1.5661,
+      "step": 393
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9865620307602325e-05,
+      "loss": 1.5877,
+      "step": 394
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9864532478639233e-05,
+      "loss": 1.7502,
+      "step": 395
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9863440294343793e-05,
+      "loss": 1.6019,
+      "step": 396
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9862343755198227e-05,
+      "loss": 1.765,
+      "step": 397
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9861242861686665e-05,
+      "loss": 1.6868,
+      "step": 398
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.986013761429517e-05,
+      "loss": 1.7615,
+      "step": 399
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9859028013511725e-05,
+      "loss": 1.6832,
+      "step": 400
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9857914059826232e-05,
+      "loss": 1.6171,
+      "step": 401
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985679575373052e-05,
+      "loss": 1.5194,
+      "step": 402
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9855673095718338e-05,
+      "loss": 1.6351,
+      "step": 403
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9854546086285353e-05,
+      "loss": 1.6114,
+      "step": 404
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9853414725929157e-05,
+      "loss": 1.5549,
+      "step": 405
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985227901514926e-05,
+      "loss": 1.5601,
+      "step": 406
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9851138954447097e-05,
+      "loss": 1.4674,
+      "step": 407
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984999454432602e-05,
+      "loss": 1.6081,
+      "step": 408
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9848845785291305e-05,
+      "loss": 1.5996,
+      "step": 409
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9847692677850142e-05,
+      "loss": 1.7471,
+      "step": 410
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9846535222511648e-05,
+      "loss": 1.6294,
+      "step": 411
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984537341978685e-05,
+      "loss": 1.6205,
+      "step": 412
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984420727018871e-05,
+      "loss": 1.5973,
+      "step": 413
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984303677423209e-05,
+      "loss": 1.7832,
+      "step": 414
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9841861932433784e-05,
+      "loss": 1.4733,
+      "step": 415
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.984068274531251e-05,
+      "loss": 1.6144,
+      "step": 416
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983949921338888e-05,
+      "loss": 1.5223,
+      "step": 417
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9838311337185452e-05,
+      "loss": 1.787,
+      "step": 418
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983711911722669e-05,
+      "loss": 1.5369,
+      "step": 419
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983592255403897e-05,
+      "loss": 1.622,
+      "step": 420
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9834721648150593e-05,
+      "loss": 1.6897,
+      "step": 421
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983351640009178e-05,
+      "loss": 1.5803,
+      "step": 422
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9832306810394665e-05,
+      "loss": 1.5418,
+      "step": 423
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.98310928795933e-05,
+      "loss": 1.6037,
+      "step": 424
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9829874608223646e-05,
+      "loss": 1.6188,
+      "step": 425
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9828651996823593e-05,
+      "loss": 1.7262,
+      "step": 426
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982742504593294e-05,
+      "loss": 1.4492,
+      "step": 427
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9826193756093407e-05,
+      "loss": 1.6529,
+      "step": 428
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9824958127848618e-05,
+      "loss": 1.7214,
+      "step": 429
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9823718161744127e-05,
+      "loss": 1.5739,
+      "step": 430
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9822473858327398e-05,
+      "loss": 1.7709,
+      "step": 431
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.98212252181478e-05,
+      "loss": 1.6085,
+      "step": 432
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9819972241756637e-05,
+      "loss": 1.625,
+      "step": 433
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9818714929707106e-05,
+      "loss": 1.6741,
+      "step": 434
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9817453282554334e-05,
+      "loss": 1.5998,
+      "step": 435
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981618730085536e-05,
+      "loss": 1.59,
+      "step": 436
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9814916985169123e-05,
+      "loss": 1.6829,
+      "step": 437
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9813642336056488e-05,
+      "loss": 1.6627,
+      "step": 438
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981236335408024e-05,
+      "loss": 1.6941,
+      "step": 439
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9811080039805052e-05,
+      "loss": 1.6808,
+      "step": 440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9809792393797543e-05,
+      "loss": 1.5172,
+      "step": 441
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9808500416626218e-05,
+      "loss": 1.6989,
+      "step": 442
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.98072041088615e-05,
+      "loss": 1.4928,
+      "step": 443
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9805903471075735e-05,
+      "loss": 1.5253,
+      "step": 444
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980459850384317e-05,
+      "loss": 1.4803,
+      "step": 445
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9803289207739967e-05,
+      "loss": 1.6226,
+      "step": 446
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.98019755833442e-05,
+      "loss": 1.6644,
+      "step": 447
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.980065763123585e-05,
+      "loss": 1.5961,
+      "step": 448
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9799335351996814e-05,
+      "loss": 1.7708,
+      "step": 449
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9798008746210894e-05,
+      "loss": 1.5671,
+      "step": 450
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9796677814463812e-05,
+      "loss": 1.6289,
+      "step": 451
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9795342557343188e-05,
+      "loss": 1.4305,
+      "step": 452
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979400297543856e-05,
+      "loss": 1.6084,
+      "step": 453
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9792659069341372e-05,
+      "loss": 1.6732,
+      "step": 454
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9791310839644976e-05,
+      "loss": 1.8997,
+      "step": 455
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9789958286944632e-05,
+      "loss": 1.6183,
+      "step": 456
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978860141183752e-05,
+      "loss": 1.5416,
+      "step": 457
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9787240214922713e-05,
+      "loss": 1.4839,
+      "step": 458
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.97858746968012e-05,
+      "loss": 1.7001,
+      "step": 459
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9784504858075883e-05,
+      "loss": 1.6428,
+      "step": 460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9783130699351556e-05,
+      "loss": 1.6093,
+      "step": 461
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9781752221234932e-05,
+      "loss": 1.4511,
+      "step": 462
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9780369424334633e-05,
+      "loss": 1.7809,
+      "step": 463
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977898230926118e-05,
+      "loss": 1.6578,
+      "step": 464
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9777590876627012e-05,
+      "loss": 1.5328,
+      "step": 465
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9776195127046453e-05,
+      "loss": 1.6307,
+      "step": 466
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9774795061135754e-05,
+      "loss": 1.3929,
+      "step": 467
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9773390679513068e-05,
+      "loss": 1.6582,
+      "step": 468
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977198198279844e-05,
+      "loss": 1.7103,
+      "step": 469
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9770568971613837e-05,
+      "loss": 1.7032,
+      "step": 470
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9769151646583122e-05,
+      "loss": 1.5408,
+      "step": 471
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9767730008332062e-05,
+      "loss": 1.6452,
+      "step": 472
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9766304057488335e-05,
+      "loss": 1.5822,
+      "step": 473
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9764873794681514e-05,
+      "loss": 1.6383,
+      "step": 474
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9763439220543084e-05,
+      "loss": 1.8773,
+      "step": 475
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976200033570643e-05,
+      "loss": 1.5865,
+      "step": 476
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9760557140806838e-05,
+      "loss": 1.5191,
+      "step": 477
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.97591096364815e-05,
+      "loss": 1.6886,
+      "step": 478
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9757657823369508e-05,
+      "loss": 1.562,
+      "step": 479
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9756201702111863e-05,
+      "loss": 1.6834,
+      "step": 480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975474127335146e-05,
+      "loss": 1.6029,
+      "step": 481
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.97532765377331e-05,
+      "loss": 1.7272,
+      "step": 482
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9751807495903484e-05,
+      "loss": 1.6437,
+      "step": 483
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9750334148511218e-05,
+      "loss": 1.4556,
+      "step": 484
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9748856496206804e-05,
+      "loss": 1.766,
+      "step": 485
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9747374539642643e-05,
+      "loss": 1.4968,
+      "step": 486
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9745888279473046e-05,
+      "loss": 1.5956,
+      "step": 487
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9744397716354214e-05,
+      "loss": 1.5747,
+      "step": 488
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9742902850944257e-05,
+      "loss": 1.79,
+      "step": 489
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9741403683903172e-05,
+      "loss": 1.4672,
+      "step": 490
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973990021589287e-05,
+      "loss": 1.4745,
+      "step": 491
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9738392447577148e-05,
+      "loss": 1.6633,
+      "step": 492
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973688037962171e-05,
+      "loss": 1.6329,
+      "step": 493
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9735364012694156e-05,
+      "loss": 1.6768,
+      "step": 494
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9733843347463982e-05,
+      "loss": 1.6309,
+      "step": 495
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9732318384602587e-05,
+      "loss": 1.6632,
+      "step": 496
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.973078912478326e-05,
+      "loss": 1.5889,
+      "step": 497
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972925556868119e-05,
+      "loss": 1.6008,
+      "step": 498
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972771771697347e-05,
+      "loss": 1.7235,
+      "step": 499
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972617557033908e-05,
+      "loss": 1.5646,
+      "step": 500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.97246291294589e-05,
+      "loss": 1.4414,
+      "step": 501
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9723078395015703e-05,
+      "loss": 1.6484,
+      "step": 502
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9721523367694167e-05,
+      "loss": 1.7167,
+      "step": 503
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9719964048180853e-05,
+      "loss": 1.5283,
+      "step": 504
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9718400437164225e-05,
+      "loss": 1.4766,
+      "step": 505
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9716832535334636e-05,
+      "loss": 1.5933,
+      "step": 506
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971526034338435e-05,
+      "loss": 1.6459,
+      "step": 507
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9713683862007494e-05,
+      "loss": 1.6161,
+      "step": 508
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9712103091900122e-05,
+      "loss": 1.6651,
+      "step": 509
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9710518033760158e-05,
+      "loss": 1.707,
+      "step": 510
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.970892868828743e-05,
+      "loss": 1.6008,
+      "step": 511
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9707335056183657e-05,
+      "loss": 1.7186,
+      "step": 512
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970573713815245e-05,
+      "loss": 1.4774,
+      "step": 513
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9704134934899317e-05,
+      "loss": 1.7674,
+      "step": 514
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9702528447131647e-05,
+      "loss": 1.4762,
+      "step": 515
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9700917675558733e-05,
+      "loss": 1.5765,
+      "step": 516
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969930262089175e-05,
+      "loss": 1.557,
+      "step": 517
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969768328384377e-05,
+      "loss": 1.475,
+      "step": 518
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969605966512975e-05,
+      "loss": 1.7818,
+      "step": 519
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9694431765466545e-05,
+      "loss": 1.6617,
+      "step": 520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9692799585572894e-05,
+      "loss": 1.6284,
+      "step": 521
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9691163126169428e-05,
+      "loss": 1.6712,
+      "step": 522
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9689522387978666e-05,
+      "loss": 1.648,
+      "step": 523
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9687877371725022e-05,
+      "loss": 1.4765,
+      "step": 524
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968622807813479e-05,
+      "loss": 1.3396,
+      "step": 525
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9684574507936155e-05,
+      "loss": 1.5757,
+      "step": 526
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9682916661859197e-05,
+      "loss": 1.596,
+      "step": 527
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9681254540635877e-05,
+      "loss": 1.5511,
+      "step": 528
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9679588145000044e-05,
+      "loss": 1.6206,
+      "step": 529
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9677917475687435e-05,
+      "loss": 1.6483,
+      "step": 530
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967624253343568e-05,
+      "loss": 1.499,
+      "step": 531
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9674563318984282e-05,
+      "loss": 1.6957,
+      "step": 532
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9672879833074642e-05,
+      "loss": 1.5084,
+      "step": 533
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9671192076450047e-05,
+      "loss": 1.76,
+      "step": 534
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9669500049855657e-05,
+      "loss": 1.5803,
+      "step": 535
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966780375403853e-05,
+      "loss": 1.4897,
+      "step": 536
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966610318974761e-05,
+      "loss": 1.438,
+      "step": 537
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9664398357733715e-05,
+      "loss": 1.5708,
+      "step": 538
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9662689258749555e-05,
+      "loss": 1.7605,
+      "step": 539
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9660975893549722e-05,
+      "loss": 1.6458,
+      "step": 540
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9659258262890683e-05,
+      "loss": 1.6337,
+      "step": 541
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965753636753081e-05,
+      "loss": 1.6316,
+      "step": 542
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9655810208230334e-05,
+      "loss": 1.6601,
+      "step": 543
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9654079785751385e-05,
+      "loss": 1.615,
+      "step": 544
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9652345100857966e-05,
+      "loss": 1.6297,
+      "step": 545
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9650606154315967e-05,
+      "loss": 1.6174,
+      "step": 546
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9648862946893158e-05,
+      "loss": 1.5553,
+      "step": 547
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964711547935919e-05,
+      "loss": 1.5651,
+      "step": 548
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9645363752485594e-05,
+      "loss": 1.6189,
+      "step": 549
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9643607767045778e-05,
+      "loss": 1.72,
+      "step": 550
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964184752381504e-05,
+      "loss": 1.5313,
+      "step": 551
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964008302357056e-05,
+      "loss": 1.5454,
+      "step": 552
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963831426709138e-05,
+      "loss": 1.4561,
+      "step": 553
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963654125515843e-05,
+      "loss": 1.4341,
+      "step": 554
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963476398855452e-05,
+      "loss": 1.6456,
+      "step": 555
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963298246806435e-05,
+      "loss": 1.4956,
+      "step": 556
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9631196694474475e-05,
+      "loss": 1.5821,
+      "step": 557
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9629406668573346e-05,
+      "loss": 1.6651,
+      "step": 558
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9627612391151278e-05,
+      "loss": 1.5882,
+      "step": 559
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9625813863000477e-05,
+      "loss": 1.6264,
+      "step": 560
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.962401108491502e-05,
+      "loss": 1.6283,
+      "step": 561
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9622204057690856e-05,
+      "loss": 1.772,
+      "step": 562
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.962039278212581e-05,
+      "loss": 1.548,
+      "step": 563
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9618577259019595e-05,
+      "loss": 1.4568,
+      "step": 564
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961675748917378e-05,
+      "loss": 1.6455,
+      "step": 565
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9614933473391825e-05,
+      "loss": 1.5601,
+      "step": 566
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961310521247906e-05,
+      "loss": 1.4095,
+      "step": 567
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961127270724269e-05,
+      "loss": 1.3433,
+      "step": 568
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9609435958491785e-05,
+      "loss": 1.4758,
+      "step": 569
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.96075949670373e-05,
+      "loss": 1.5793,
+      "step": 570
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9605749733692063e-05,
+      "loss": 1.7595,
+      "step": 571
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9603900259270765e-05,
+      "loss": 1.7898,
+      "step": 572
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960204654458998e-05,
+      "loss": 1.6643,
+      "step": 573
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9600188590468144e-05,
+      "loss": 1.6926,
+      "step": 574
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9598326397725577e-05,
+      "loss": 1.4618,
+      "step": 575
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9596459967184458e-05,
+      "loss": 1.7158,
+      "step": 576
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9594589299668846e-05,
+      "loss": 1.7114,
+      "step": 577
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9592714396004665e-05,
+      "loss": 1.6013,
+      "step": 578
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9590835257019715e-05,
+      "loss": 1.8002,
+      "step": 579
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958895188354366e-05,
+      "loss": 1.6863,
+      "step": 580
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9587064276408043e-05,
+      "loss": 1.7051,
+      "step": 581
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958517243644626e-05,
+      "loss": 1.4915,
+      "step": 582
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958327636449359e-05,
+      "loss": 1.7515,
+      "step": 583
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9581376061387174e-05,
+      "loss": 1.7508,
+      "step": 584
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9579471527966026e-05,
+      "loss": 1.6445,
+      "step": 585
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9577562765071024e-05,
+      "loss": 1.3606,
+      "step": 586
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9575649773544914e-05,
+      "loss": 1.7449,
+      "step": 587
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957373255423231e-05,
+      "loss": 1.5362,
+      "step": 588
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9571811107979687e-05,
+      "loss": 1.5664,
+      "step": 589
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956988543563539e-05,
+      "loss": 1.7692,
+      "step": 590
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9567955538049643e-05,
+      "loss": 1.7307,
+      "step": 591
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9566021416074513e-05,
+      "loss": 1.5732,
+      "step": 592
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9564083070563948e-05,
+      "loss": 1.3883,
+      "step": 593
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9562140502373753e-05,
+      "loss": 1.6809,
+      "step": 594
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9560193712361597e-05,
+      "loss": 1.4126,
+      "step": 595
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955824270138702e-05,
+      "loss": 1.4092,
+      "step": 596
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9556287470311418e-05,
+      "loss": 1.6105,
+      "step": 597
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9554328019998055e-05,
+      "loss": 1.594,
+      "step": 598
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9552364351312056e-05,
+      "loss": 1.5213,
+      "step": 599
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955039646512041e-05,
+      "loss": 1.722,
+      "step": 600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9548424362291963e-05,
+      "loss": 1.7328,
+      "step": 601
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954644804369743e-05,
+      "loss": 1.6705,
+      "step": 602
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954446751020939e-05,
+      "loss": 1.3883,
+      "step": 603
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9542482762702264e-05,
+      "loss": 1.6117,
+      "step": 604
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9540493802052355e-05,
+      "loss": 1.4389,
+      "step": 605
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9538500629137814e-05,
+      "loss": 1.3845,
+      "step": 606
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953650324483866e-05,
+      "loss": 1.5546,
+      "step": 607
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9534501650036764e-05,
+      "loss": 1.5457,
+      "step": 608
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9532495845615854e-05,
+      "loss": 1.6544,
+      "step": 609
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.953048583246153e-05,
+      "loss": 1.7206,
+      "step": 610
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9528471611461235e-05,
+      "loss": 1.5617,
+      "step": 611
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952645318350428e-05,
+      "loss": 1.4918,
+      "step": 612
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9524430549481827e-05,
+      "loss": 1.7825,
+      "step": 613
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9522403710286904e-05,
+      "loss": 1.6037,
+      "step": 614
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952037266681438e-05,
+      "loss": 1.6588,
+      "step": 615
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9518337419961e-05,
+      "loss": 1.678,
+      "step": 616
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9516297970625344e-05,
+      "loss": 1.632,
+      "step": 617
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9514254319707868e-05,
+      "loss": 1.4636,
+      "step": 618
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9512206468110863e-05,
+      "loss": 1.4955,
+      "step": 619
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9510154416738495e-05,
+      "loss": 1.7379,
+      "step": 620
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950809816649677e-05,
+      "loss": 1.4619,
+      "step": 621
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950603771829355e-05,
+      "loss": 1.5194,
+      "step": 622
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9503973073038554e-05,
+      "loss": 1.7175,
+      "step": 623
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9501904231643353e-05,
+      "loss": 1.5593,
+      "step": 624
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.949983119502137e-05,
+      "loss": 1.5182,
+      "step": 625
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949775396408788e-05,
+      "loss": 1.8002,
+      "step": 626
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949567253976001e-05,
+      "loss": 1.7214,
+      "step": 627
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9493586922956736e-05,
+      "loss": 1.6373,
+      "step": 628
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9491497114598892e-05,
+      "loss": 1.5392,
+      "step": 629
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9489403115609157e-05,
+      "loss": 1.6721,
+      "step": 630
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948730492691206e-05,
+      "loss": 1.6799,
+      "step": 631
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9485202549433986e-05,
+      "loss": 1.7387,
+      "step": 632
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948309598410316e-05,
+      "loss": 1.752,
+      "step": 633
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948098523184966e-05,
+      "loss": 1.5648,
+      "step": 634
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9478870293605416e-05,
+      "loss": 1.5138,
+      "step": 635
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9476751170304206e-05,
+      "loss": 1.5918,
+      "step": 636
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9474627862881644e-05,
+      "loss": 1.6359,
+      "step": 637
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9472500372275212e-05,
+      "loss": 1.4255,
+      "step": 638
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947036869942422e-05,
+      "loss": 1.5994,
+      "step": 639
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9468232845269837e-05,
+      "loss": 1.4497,
+      "step": 640
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946609281075507e-05,
+      "loss": 1.6148,
+      "step": 641
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9463948596824774e-05,
+      "loss": 1.6318,
+      "step": 642
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9461800204425653e-05,
+      "loss": 1.6085,
+      "step": 643
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9459647634506253e-05,
+      "loss": 1.5614,
+      "step": 644
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9457490888016962e-05,
+      "loss": 1.7067,
+      "step": 645
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9455329965910016e-05,
+      "loss": 1.6461,
+      "step": 646
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9453164869139488e-05,
+      "loss": 1.808,
+      "step": 647
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945099559866131e-05,
+      "loss": 1.5588,
+      "step": 648
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9448822155433235e-05,
+      "loss": 1.6787,
+      "step": 649
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9446644540414882e-05,
+      "loss": 1.6981,
+      "step": 650
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9444462754567682e-05,
+      "loss": 1.498,
+      "step": 651
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944227679885494e-05,
+      "loss": 1.6146,
+      "step": 652
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944008667424178e-05,
+      "loss": 1.6685,
+      "step": 653
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9437892381695174e-05,
+      "loss": 1.6024,
+      "step": 654
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9435693922183935e-05,
+      "loss": 1.7665,
+      "step": 655
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9433491296678712e-05,
+      "loss": 1.4511,
+      "step": 656
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9431284506151996e-05,
+      "loss": 1.7176,
+      "step": 657
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.942907355157812e-05,
+      "loss": 1.6458,
+      "step": 658
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9426858433933248e-05,
+      "loss": 1.7923,
+      "step": 659
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9424639154195386e-05,
+      "loss": 1.5881,
+      "step": 660
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9422415713344382e-05,
+      "loss": 1.5395,
+      "step": 661
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9420188112361914e-05,
+      "loss": 1.6348,
+      "step": 662
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.94179563522315e-05,
+      "loss": 1.6161,
+      "step": 663
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9415720433938495e-05,
+      "loss": 1.5854,
+      "step": 664
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9413480358470088e-05,
+      "loss": 1.4991,
+      "step": 665
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9411236126815306e-05,
+      "loss": 1.6269,
+      "step": 666
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9408987739965006e-05,
+      "loss": 1.6847,
+      "step": 667
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9406735198911885e-05,
+      "loss": 1.5164,
+      "step": 668
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9404478504650473e-05,
+      "loss": 1.6994,
+      "step": 669
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940221765817713e-05,
+      "loss": 1.7708,
+      "step": 670
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9399952660490057e-05,
+      "loss": 1.5681,
+      "step": 671
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9397683512589282e-05,
+      "loss": 1.6009,
+      "step": 672
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9395410215476662e-05,
+      "loss": 1.6308,
+      "step": 673
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939313277015589e-05,
+      "loss": 1.6737,
+      "step": 674
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9390851177632496e-05,
+      "loss": 1.6132,
+      "step": 675
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9388565438913832e-05,
+      "loss": 1.6237,
+      "step": 676
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938627555500909e-05,
+      "loss": 1.5915,
+      "step": 677
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9383981526929277e-05,
+      "loss": 1.6012,
+      "step": 678
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9381683355687245e-05,
+      "loss": 1.5484,
+      "step": 679
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937938104229767e-05,
+      "loss": 1.4984,
+      "step": 680
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937707458777705e-05,
+      "loss": 1.4849,
+      "step": 681
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9374763993143727e-05,
+      "loss": 1.6229,
+      "step": 682
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937244925941786e-05,
+      "loss": 1.4088,
+      "step": 683
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937013038762143e-05,
+      "loss": 1.555,
+      "step": 684
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9367807378778257e-05,
+      "loss": 1.6739,
+      "step": 685
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9365480233913984e-05,
+      "loss": 1.4493,
+      "step": 686
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9363148954056077e-05,
+      "loss": 1.5109,
+      "step": 687
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936081354023383e-05,
+      "loss": 1.76,
+      "step": 688
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9358473993478358e-05,
+      "loss": 1.5776,
+      "step": 689
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.935613031482261e-05,
+      "loss": 1.6592,
+      "step": 690
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9353782505301352e-05,
+      "loss": 1.6355,
+      "step": 691
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9351430565951174e-05,
+      "loss": 1.8171,
+      "step": 692
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934907449781049e-05,
+      "loss": 1.5519,
+      "step": 693
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934671430191954e-05,
+      "loss": 1.6161,
+      "step": 694
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9344349979320386e-05,
+      "loss": 1.5866,
+      "step": 695
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9341981531056905e-05,
+      "loss": 1.562,
+      "step": 696
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9339608958174806e-05,
+      "loss": 1.6922,
+      "step": 697
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933723226172161e-05,
+      "loss": 1.5617,
+      "step": 698
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9334851442746665e-05,
+      "loss": 1.608,
+      "step": 699
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9332466502301134e-05,
+      "loss": 1.7652,
+      "step": 700
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9330077441438007e-05,
+      "loss": 1.6667,
+      "step": 701
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9327684261212084e-05,
+      "loss": 1.5824,
+      "step": 702
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9325286962679993e-05,
+      "loss": 1.6185,
+      "step": 703
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932288554690017e-05,
+      "loss": 1.6518,
+      "step": 704
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932048001493288e-05,
+      "loss": 1.4915,
+      "step": 705
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9318070367840197e-05,
+      "loss": 1.7375,
+      "step": 706
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9315656606686012e-05,
+      "loss": 1.5598,
+      "step": 707
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.931323873253604e-05,
+      "loss": 1.605,
+      "step": 708
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9310816746457802e-05,
+      "loss": 1.745,
+      "step": 709
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9308390649520645e-05,
+      "loss": 1.4643,
+      "step": 710
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930596044279572e-05,
+      "loss": 1.6,
+      "step": 711
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9303526127356008e-05,
+      "loss": 1.4873,
+      "step": 712
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9301087704276283e-05,
+      "loss": 1.5556,
+      "step": 713
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9298645174633147e-05,
+      "loss": 1.6179,
+      "step": 714
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9296198539505013e-05,
+      "loss": 1.5105,
+      "step": 715
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9293747799972108e-05,
+      "loss": 1.5953,
+      "step": 716
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929129295711646e-05,
+      "loss": 1.7087,
+      "step": 717
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928883401202193e-05,
+      "loss": 1.7149,
+      "step": 718
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9286370965774166e-05,
+      "loss": 1.5106,
+      "step": 719
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9283903819460644e-05,
+      "loss": 1.6228,
+      "step": 720
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9281432574170644e-05,
+      "loss": 1.6731,
+      "step": 721
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.927895723099525e-05,
+      "loss": 1.6041,
+      "step": 722
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9276477791027374e-05,
+      "loss": 1.5067,
+      "step": 723
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9273994255361717e-05,
+      "loss": 1.7052,
+      "step": 724
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.927150662509479e-05,
+      "loss": 1.5952,
+      "step": 725
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9269014901324926e-05,
+      "loss": 1.5602,
+      "step": 726
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9266519085152254e-05,
+      "loss": 1.416,
+      "step": 727
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9264019177678712e-05,
+      "loss": 1.6633,
+      "step": 728
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9261515180008047e-05,
+      "loss": 1.4615,
+      "step": 729
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9259007093245804e-05,
+      "loss": 1.7675,
+      "step": 730
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9256494918499348e-05,
+      "loss": 1.751,
+      "step": 731
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925397865687783e-05,
+      "loss": 1.6461,
+      "step": 732
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9251458309492227e-05,
+      "loss": 1.6914,
+      "step": 733
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9248933877455296e-05,
+      "loss": 1.7653,
+      "step": 734
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9246405361881623e-05,
+      "loss": 1.5032,
+      "step": 735
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9243872763887574e-05,
+      "loss": 1.5974,
+      "step": 736
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9241336084591327e-05,
+      "loss": 1.7918,
+      "step": 737
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9238795325112867e-05,
+      "loss": 1.7279,
+      "step": 738
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9236250486573978e-05,
+      "loss": 1.7666,
+      "step": 739
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9233701570098237e-05,
+      "loss": 1.5979,
+      "step": 740
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.923114857681103e-05,
+      "loss": 1.6974,
+      "step": 741
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9228591507839538e-05,
+      "loss": 1.6683,
+      "step": 742
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9226030364312747e-05,
+      "loss": 1.3971,
+      "step": 743
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9223465147361433e-05,
+      "loss": 1.565,
+      "step": 744
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922089585811818e-05,
+      "loss": 1.7435,
+      "step": 745
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921832249771737e-05,
+      "loss": 1.7694,
+      "step": 746
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9215745067295168e-05,
+      "loss": 1.8421,
+      "step": 747
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9213163567989554e-05,
+      "loss": 1.6848,
+      "step": 748
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9210578000940297e-05,
+      "loss": 1.7941,
+      "step": 749
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9207988367288956e-05,
+      "loss": 1.6453,
+      "step": 750
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.92053946681789e-05,
+      "loss": 1.6146,
+      "step": 751
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9202796904755276e-05,
+      "loss": 1.5408,
+      "step": 752
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920019507816504e-05,
+      "loss": 1.4548,
+      "step": 753
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.919758918955693e-05,
+      "loss": 1.6623,
+      "step": 754
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.919497924008149e-05,
+      "loss": 1.6373,
+      "step": 755
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9192365230891044e-05,
+      "loss": 1.6414,
+      "step": 756
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9189747163139717e-05,
+      "loss": 1.6519,
+      "step": 757
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9187125037983424e-05,
+      "loss": 1.7046,
+      "step": 758
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918449885657987e-05,
+      "loss": 1.6493,
+      "step": 759
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918186862008855e-05,
+      "loss": 1.5609,
+      "step": 760
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9179234329670757e-05,
+      "loss": 1.6863,
+      "step": 761
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9176595986489558e-05,
+      "loss": 1.806,
+      "step": 762
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917395359170983e-05,
+      "loss": 1.6281,
+      "step": 763
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917130714649822e-05,
+      "loss": 1.7182,
+      "step": 764
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9168656652023176e-05,
+      "loss": 1.7276,
+      "step": 765
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9166002109454927e-05,
+      "loss": 1.4843,
+      "step": 766
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9163343519965494e-05,
+      "loss": 1.4563,
+      "step": 767
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916068088472868e-05,
+      "loss": 1.6524,
+      "step": 768
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9158014204920076e-05,
+      "loss": 1.5134,
+      "step": 769
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915534348171706e-05,
+      "loss": 1.542,
+      "step": 770
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.91526687162988e-05,
+      "loss": 1.617,
+      "step": 771
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9149989909846235e-05,
+      "loss": 1.5542,
+      "step": 772
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.91473070635421e-05,
+      "loss": 1.679,
+      "step": 773
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914462017857091e-05,
+      "loss": 1.578,
+      "step": 774
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914192925611896e-05,
+      "loss": 1.4474,
+      "step": 775
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913923429737434e-05,
+      "loss": 1.5909,
+      "step": 776
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9136535303526906e-05,
+      "loss": 1.5507,
+      "step": 777
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.91338322757683e-05,
+      "loss": 1.5202,
+      "step": 778
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913112521529195e-05,
+      "loss": 1.6357,
+      "step": 779
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9128414123293065e-05,
+      "loss": 1.6512,
+      "step": 780
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912569900096863e-05,
+      "loss": 1.6339,
+      "step": 781
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9122979849517406e-05,
+      "loss": 1.3471,
+      "step": 782
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9120256670139942e-05,
+      "loss": 1.4028,
+      "step": 783
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9117529464038557e-05,
+      "loss": 1.7481,
+      "step": 784
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9114798232417357e-05,
+      "loss": 1.6844,
+      "step": 785
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9112062976482213e-05,
+      "loss": 1.5684,
+      "step": 786
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9109323697440782e-05,
+      "loss": 1.4663,
+      "step": 787
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.91065803965025e-05,
+      "loss": 1.4431,
+      "step": 788
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9103833074878565e-05,
+      "loss": 1.5552,
+      "step": 789
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9101081733781966e-05,
+      "loss": 1.5772,
+      "step": 790
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909832637442746e-05,
+      "loss": 1.7017,
+      "step": 791
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909556699803157e-05,
+      "loss": 1.4459,
+      "step": 792
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9092803605812607e-05,
+      "loss": 1.6712,
+      "step": 793
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9090036198990645e-05,
+      "loss": 1.5643,
+      "step": 794
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9087264778787534e-05,
+      "loss": 1.69,
+      "step": 795
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.90844893464269e-05,
+      "loss": 1.5052,
+      "step": 796
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908170990313413e-05,
+      "loss": 1.5878,
+      "step": 797
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907892645013639e-05,
+      "loss": 1.6811,
+      "step": 798
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9076138988662615e-05,
+      "loss": 1.6881,
+      "step": 799
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9073347519943512e-05,
+      "loss": 1.6047,
+      "step": 800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9070552045211544e-05,
+      "loss": 1.6132,
+      "step": 801
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906775256570097e-05,
+      "loss": 1.5448,
+      "step": 802
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9064949082647785e-05,
+      "loss": 1.6871,
+      "step": 803
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9062141597289773e-05,
+      "loss": 1.6982,
+      "step": 804
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905933011086648e-05,
+      "loss": 1.5988,
+      "step": 805
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9056514624619217e-05,
+      "loss": 1.5698,
+      "step": 806
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9053695139791058e-05,
+      "loss": 1.4152,
+      "step": 807
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9050871657626854e-05,
+      "loss": 1.5325,
+      "step": 808
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9048044179373204e-05,
+      "loss": 1.4677,
+      "step": 809
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9045212706278485e-05,
+      "loss": 1.4929,
+      "step": 810
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904237723959283e-05,
+      "loss": 1.4778,
+      "step": 811
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903953778056814e-05,
+      "loss": 1.5147,
+      "step": 812
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9036694330458077e-05,
+      "loss": 1.572,
+      "step": 813
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9033846890518063e-05,
+      "loss": 1.399,
+      "step": 814
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903099546200529e-05,
+      "loss": 1.5031,
+      "step": 815
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9028140046178696e-05,
+      "loss": 1.3543,
+      "step": 816
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902528064429899e-05,
+      "loss": 1.5934,
+      "step": 817
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902241725762864e-05,
+      "loss": 1.4776,
+      "step": 818
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.901954988743188e-05,
+      "loss": 1.7306,
+      "step": 819
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.901667853497468e-05,
+      "loss": 1.5975,
+      "step": 820
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9013803201524794e-05,
+      "loss": 1.6231,
+      "step": 821
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9010923888351717e-05,
+      "loss": 1.6846,
+      "step": 822
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9008040596726713e-05,
+      "loss": 1.5433,
+      "step": 823
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9005153327922792e-05,
+      "loss": 1.7837,
+      "step": 824
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9002262083214728e-05,
+      "loss": 1.4585,
+      "step": 825
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899936686387904e-05,
+      "loss": 1.5634,
+      "step": 826
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8996467671194017e-05,
+      "loss": 1.502,
+      "step": 827
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8993564506439687e-05,
+      "loss": 1.6882,
+      "step": 828
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899065737089785e-05,
+      "loss": 1.5014,
+      "step": 829
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898774626585203e-05,
+      "loss": 1.5065,
+      "step": 830
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898483119258754e-05,
+      "loss": 1.755,
+      "step": 831
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8981912152391417e-05,
+      "loss": 1.577,
+      "step": 832
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8978989146552457e-05,
+      "loss": 1.4364,
+      "step": 833
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8976062176361213e-05,
+      "loss": 1.5586,
+      "step": 834
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8973131243109987e-05,
+      "loss": 1.5505,
+      "step": 835
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8970196348092824e-05,
+      "loss": 1.7757,
+      "step": 836
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8967257492605522e-05,
+      "loss": 1.611,
+      "step": 837
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.896431467794563e-05,
+      "loss": 1.74,
+      "step": 838
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.896136790541244e-05,
+      "loss": 1.5744,
+      "step": 839
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8958417176307e-05,
+      "loss": 1.5831,
+      "step": 840
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8955462491932088e-05,
+      "loss": 1.5521,
+      "step": 841
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895250385359225e-05,
+      "loss": 1.6582,
+      "step": 842
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8949541262593764e-05,
+      "loss": 1.6763,
+      "step": 843
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8946574720244656e-05,
+      "loss": 1.7142,
+      "step": 844
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8943604227854696e-05,
+      "loss": 1.7645,
+      "step": 845
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.89406297867354e-05,
+      "loss": 1.5585,
+      "step": 846
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8937651398200027e-05,
+      "loss": 1.6217,
+      "step": 847
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8934669063563577e-05,
+      "loss": 1.6735,
+      "step": 848
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8931682784142792e-05,
+      "loss": 1.6633,
+      "step": 849
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8928692561256162e-05,
+      "loss": 1.6323,
+      "step": 850
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892569839622391e-05,
+      "loss": 1.7548,
+      "step": 851
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8922700290367998e-05,
+      "loss": 1.7032,
+      "step": 852
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8919698245012143e-05,
+      "loss": 1.536,
+      "step": 853
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8916692261481784e-05,
+      "loss": 1.562,
+      "step": 854
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8913682341104107e-05,
+      "loss": 1.6445,
+      "step": 855
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8910668485208036e-05,
+      "loss": 1.6701,
+      "step": 856
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8907650695124227e-05,
+      "loss": 1.5813,
+      "step": 857
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8904628972185086e-05,
+      "loss": 1.5461,
+      "step": 858
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8901603317724742e-05,
+      "loss": 1.6134,
+      "step": 859
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8898573733079062e-05,
+      "loss": 1.5476,
+      "step": 860
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8895540219585656e-05,
+      "loss": 1.6436,
+      "step": 861
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8892502778583864e-05,
+      "loss": 1.7057,
+      "step": 862
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8889461411414754e-05,
+      "loss": 1.5111,
+      "step": 863
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.888641611942114e-05,
+      "loss": 1.6484,
+      "step": 864
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8883366903947554e-05,
+      "loss": 1.5257,
+      "step": 865
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8880313766340274e-05,
+      "loss": 1.4875,
+      "step": 866
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8877256707947308e-05,
+      "loss": 1.5076,
+      "step": 867
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887419573011838e-05,
+      "loss": 1.7335,
+      "step": 868
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8871130834204964e-05,
+      "loss": 1.6441,
+      "step": 869
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8868062021560254e-05,
+      "loss": 1.6569,
+      "step": 870
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.886498929353917e-05,
+      "loss": 1.6962,
+      "step": 871
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8861912651498372e-05,
+      "loss": 1.4508,
+      "step": 872
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885883209679624e-05,
+      "loss": 1.477,
+      "step": 873
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8855747630792878e-05,
+      "loss": 1.6184,
+      "step": 874
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8852659254850128e-05,
+      "loss": 1.7266,
+      "step": 875
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884956697033155e-05,
+      "loss": 1.665,
+      "step": 876
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8846470778602427e-05,
+      "loss": 1.6032,
+      "step": 877
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884337068102978e-05,
+      "loss": 1.6504,
+      "step": 878
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8840266678982343e-05,
+      "loss": 1.5969,
+      "step": 879
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883715877383058e-05,
+      "loss": 1.8148,
+      "step": 880
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883404696694667e-05,
+      "loss": 1.5835,
+      "step": 881
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883093125970452e-05,
+      "loss": 1.6426,
+      "step": 882
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882781165347977e-05,
+      "loss": 1.6532,
+      "step": 883
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882468814964976e-05,
+      "loss": 1.5435,
+      "step": 884
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8821560749593564e-05,
+      "loss": 1.5656,
+      "step": 885
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8818429454691974e-05,
+      "loss": 1.4943,
+      "step": 886
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8815294266327507e-05,
+      "loss": 1.5775,
+      "step": 887
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8812155185884385e-05,
+      "loss": 1.5993,
+      "step": 888
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8809012214748567e-05,
+      "loss": 1.6471,
+      "step": 889
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880586535430771e-05,
+      "loss": 1.7498,
+      "step": 890
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.88027146059512e-05,
+      "loss": 1.6645,
+      "step": 891
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879955997107014e-05,
+      "loss": 1.7022,
+      "step": 892
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8796401451057348e-05,
+      "loss": 1.6522,
+      "step": 893
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8793239047307358e-05,
+      "loss": 1.6107,
+      "step": 894
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8790072761216406e-05,
+      "loss": 1.5738,
+      "step": 895
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.878690259418246e-05,
+      "loss": 1.6301,
+      "step": 896
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8783728547605192e-05,
+      "loss": 1.5261,
+      "step": 897
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.878055062288599e-05,
+      "loss": 1.6532,
+      "step": 898
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8777368821427954e-05,
+      "loss": 1.6446,
+      "step": 899
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8774183144635894e-05,
+      "loss": 1.4935,
+      "step": 900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8770993593916332e-05,
+      "loss": 1.586,
+      "step": 901
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8767800170677505e-05,
+      "loss": 1.6432,
+      "step": 902
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8764602876329346e-05,
+      "loss": 1.7139,
+      "step": 903
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8761401712283513e-05,
+      "loss": 1.6524,
+      "step": 904
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875819667995336e-05,
+      "loss": 1.5681,
+      "step": 905
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875498778075396e-05,
+      "loss": 1.5006,
+      "step": 906
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8751775016102087e-05,
+      "loss": 1.5318,
+      "step": 907
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874855838741622e-05,
+      "loss": 1.432,
+      "step": 908
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874533789611655e-05,
+      "loss": 1.7123,
+      "step": 909
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874211354362497e-05,
+      "loss": 1.8129,
+      "step": 910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8738885331365073e-05,
+      "loss": 1.7674,
+      "step": 911
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8735653260762165e-05,
+      "loss": 1.3817,
+      "step": 912
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.873241733324325e-05,
+      "loss": 1.5101,
+      "step": 913
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8729177550237038e-05,
+      "loss": 1.5339,
+      "step": 914
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872593391317394e-05,
+      "loss": 1.5157,
+      "step": 915
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8722686423486066e-05,
+      "loss": 1.7904,
+      "step": 916
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.871943508260723e-05,
+      "loss": 1.7398,
+      "step": 917
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8716179891972942e-05,
+      "loss": 1.6023,
+      "step": 918
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.871292085302042e-05,
+      "loss": 1.721,
+      "step": 919
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8709657967188576e-05,
+      "loss": 1.5709,
+      "step": 920
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8706391235918022e-05,
+      "loss": 1.5921,
+      "step": 921
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870312066065106e-05,
+      "loss": 1.8055,
+      "step": 922
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8699846242831707e-05,
+      "loss": 1.655,
+      "step": 923
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8696567983905655e-05,
+      "loss": 1.6977,
+      "step": 924
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8693285885320305e-05,
+      "loss": 1.5009,
+      "step": 925
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8689999948524753e-05,
+      "loss": 1.5817,
+      "step": 926
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8686710174969786e-05,
+      "loss": 1.4096,
+      "step": 927
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8683416566107886e-05,
+      "loss": 1.5376,
+      "step": 928
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8680119123393225e-05,
+      "loss": 1.4173,
+      "step": 929
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8676817848281677e-05,
+      "loss": 1.7291,
+      "step": 930
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.86735127422308e-05,
+      "loss": 1.5493,
+      "step": 931
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8670203806699845e-05,
+      "loss": 1.3754,
+      "step": 932
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8666891043149757e-05,
+      "loss": 1.646,
+      "step": 933
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8663574453043162e-05,
+      "loss": 1.6188,
+      "step": 934
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.866025403784439e-05,
+      "loss": 1.5563,
+      "step": 935
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8656929799019444e-05,
+      "loss": 1.6774,
+      "step": 936
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8653601738036033e-05,
+      "loss": 1.4518,
+      "step": 937
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8650269856363532e-05,
+      "loss": 1.6938,
+      "step": 938
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8646934155473025e-05,
+      "loss": 1.5942,
+      "step": 939
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8643594636837262e-05,
+      "loss": 1.7891,
+      "step": 940
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8640251301930696e-05,
+      "loss": 1.682,
+      "step": 941
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8636904152229455e-05,
+      "loss": 1.4555,
+      "step": 942
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8633553189211353e-05,
+      "loss": 1.6025,
+      "step": 943
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8630198414355888e-05,
+      "loss": 1.6983,
+      "step": 944
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8626839829144244e-05,
+      "loss": 1.7276,
+      "step": 945
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.862347743505928e-05,
+      "loss": 1.7052,
+      "step": 946
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.862011123358554e-05,
+      "loss": 1.7705,
+      "step": 947
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8616741226209256e-05,
+      "loss": 1.6757,
+      "step": 948
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861336741441833e-05,
+      "loss": 1.6346,
+      "step": 949
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8609989799702352e-05,
+      "loss": 1.5505,
+      "step": 950
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8606608383552583e-05,
+      "loss": 1.5311,
+      "step": 951
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.860322316746197e-05,
+      "loss": 1.5595,
+      "step": 952
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.859983415292513e-05,
+      "loss": 1.4602,
+      "step": 953
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.859644134143837e-05,
+      "loss": 1.6365,
+      "step": 954
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8593044734499657e-05,
+      "loss": 1.7359,
+      "step": 955
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8589644333608645e-05,
+      "loss": 1.5032,
+      "step": 956
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8586240140266655e-05,
+      "loss": 1.5069,
+      "step": 957
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8582832155976694e-05,
+      "loss": 1.3862,
+      "step": 958
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8579420382243433e-05,
+      "loss": 1.7263,
+      "step": 959
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8576004820573217e-05,
+      "loss": 1.3741,
+      "step": 960
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8572585472474068e-05,
+      "loss": 1.5794,
+      "step": 961
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8569162339455675e-05,
+      "loss": 1.6572,
+      "step": 962
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8565735423029406e-05,
+      "loss": 1.2936,
+      "step": 963
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8562304724708285e-05,
+      "loss": 1.4861,
+      "step": 964
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8558870246007015e-05,
+      "loss": 1.6607,
+      "step": 965
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8555431988441977e-05,
+      "loss": 1.6444,
+      "step": 966
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8551989953531204e-05,
+      "loss": 1.673,
+      "step": 967
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8548544142794404e-05,
+      "loss": 1.6086,
+      "step": 968
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854509455775295e-05,
+      "loss": 1.4989,
+      "step": 969
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854164119992989e-05,
+      "loss": 1.364,
+      "step": 970
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8538184070849926e-05,
+      "loss": 1.4012,
+      "step": 971
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.853472317203943e-05,
+      "loss": 1.4886,
+      "step": 972
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.853125850502644e-05,
+      "loss": 1.7559,
+      "step": 973
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.852779007134065e-05,
+      "loss": 1.6225,
+      "step": 974
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8524317872513434e-05,
+      "loss": 1.7267,
+      "step": 975
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.852084191007781e-05,
+      "loss": 1.5998,
+      "step": 976
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8517362185568465e-05,
+      "loss": 1.6712,
+      "step": 977
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8513878700521752e-05,
+      "loss": 1.5845,
+      "step": 978
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8510391456475674e-05,
+      "loss": 1.428,
+      "step": 979
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8506900454969904e-05,
+      "loss": 1.5337,
+      "step": 980
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8503405697545767e-05,
+      "loss": 1.5918,
+      "step": 981
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.849990718574625e-05,
+      "loss": 1.4813,
+      "step": 982
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8496404921115993e-05,
+      "loss": 1.6379,
+      "step": 983
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8492898905201297e-05,
+      "loss": 1.3694,
+      "step": 984
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8489389139550114e-05,
+      "loss": 1.7517,
+      "step": 985
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8485875625712068e-05,
+      "loss": 1.6448,
+      "step": 986
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8482358365238414e-05,
+      "loss": 1.7898,
+      "step": 987
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8478837359682076e-05,
+      "loss": 1.723,
+      "step": 988
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8475312610597634e-05,
+      "loss": 1.3854,
+      "step": 989
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847178411954131e-05,
+      "loss": 1.5847,
+      "step": 990
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8468251888070982e-05,
+      "loss": 1.6017,
+      "step": 991
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8464715917746182e-05,
+      "loss": 1.6523,
+      "step": 992
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8461176210128095e-05,
+      "loss": 1.346,
+      "step": 993
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845763276677955e-05,
+      "loss": 1.6029,
+      "step": 994
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845408558926502e-05,
+      "loss": 1.4733,
+      "step": 995
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845053467915065e-05,
+      "loss": 1.5341,
+      "step": 996
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.844698003800421e-05,
+      "loss": 1.7708,
+      "step": 997
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8443421667395122e-05,
+      "loss": 1.5911,
+      "step": 998
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8439859568894464e-05,
+      "loss": 1.6238,
+      "step": 999
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.843629374407495e-05,
+      "loss": 1.2934,
+      "step": 1000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.843272419451094e-05,
+      "loss": 1.549,
+      "step": 1001
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8429150921778448e-05,
+      "loss": 1.6761,
+      "step": 1002
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.842557392745512e-05,
+      "loss": 1.7536,
+      "step": 1003
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.842199321312025e-05,
+      "loss": 1.7214,
+      "step": 1004
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8418408780354776e-05,
+      "loss": 1.5818,
+      "step": 1005
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8414820630741278e-05,
+      "loss": 1.4017,
+      "step": 1006
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8411228765863973e-05,
+      "loss": 1.6307,
+      "step": 1007
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8407633187308722e-05,
+      "loss": 1.6414,
+      "step": 1008
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8404033896663022e-05,
+      "loss": 1.5702,
+      "step": 1009
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.840043089551602e-05,
+      "loss": 1.5626,
+      "step": 1010
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839682418545848e-05,
+      "loss": 1.6596,
+      "step": 1011
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8393213768082823e-05,
+      "loss": 1.5869,
+      "step": 1012
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8389599644983102e-05,
+      "loss": 1.694,
+      "step": 1013
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8385981817754997e-05,
+      "loss": 1.4142,
+      "step": 1014
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.838236028799584e-05,
+      "loss": 1.5366,
+      "step": 1015
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.837873505730458e-05,
+      "loss": 1.6553,
+      "step": 1016
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.837510612728181e-05,
+      "loss": 1.5482,
+      "step": 1017
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.837147349952976e-05,
+      "loss": 1.5235,
+      "step": 1018
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8367837175652284e-05,
+      "loss": 1.5304,
+      "step": 1019
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8364197157254863e-05,
+      "loss": 1.4943,
+      "step": 1020
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8360553445944632e-05,
+      "loss": 1.6121,
+      "step": 1021
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8356906043330333e-05,
+      "loss": 1.5472,
+      "step": 1022
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8353254951022342e-05,
+      "loss": 1.6498,
+      "step": 1023
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.834960017063268e-05,
+      "loss": 1.6287,
+      "step": 1024
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8345941703774975e-05,
+      "loss": 1.5982,
+      "step": 1025
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.83422795520645e-05,
+      "loss": 1.721,
+      "step": 1026
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833861371711814e-05,
+      "loss": 1.4572,
+      "step": 1027
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833494420055442e-05,
+      "loss": 1.5764,
+      "step": 1028
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833127100399348e-05,
+      "loss": 1.7015,
+      "step": 1029
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8327594129057095e-05,
+      "loss": 1.6798,
+      "step": 1030
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832391357736865e-05,
+      "loss": 1.4817,
+      "step": 1031
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8320229350553165e-05,
+      "loss": 1.8062,
+      "step": 1032
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.831654145023728e-05,
+      "loss": 1.6447,
+      "step": 1033
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8312849878049253e-05,
+      "loss": 1.6315,
+      "step": 1034
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8309154635618967e-05,
+      "loss": 1.4537,
+      "step": 1035
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8305455724577926e-05,
+      "loss": 1.6629,
+      "step": 1036
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.830175314655925e-05,
+      "loss": 1.5957,
+      "step": 1037
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8298046903197685e-05,
+      "loss": 1.6808,
+      "step": 1038
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.829433699612958e-05,
+      "loss": 1.3784,
+      "step": 1039
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8290623426992928e-05,
+      "loss": 1.543,
+      "step": 1040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828690619742731e-05,
+      "loss": 1.7609,
+      "step": 1041
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8283185309073937e-05,
+      "loss": 1.4845,
+      "step": 1042
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.827946076357564e-05,
+      "loss": 1.3466,
+      "step": 1043
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8275732562576857e-05,
+      "loss": 1.4461,
+      "step": 1044
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.827200070772364e-05,
+      "loss": 1.6142,
+      "step": 1045
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8268265200663657e-05,
+      "loss": 1.5526,
+      "step": 1046
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8264526043046193e-05,
+      "loss": 1.5395,
+      "step": 1047
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8260783236522136e-05,
+      "loss": 1.5205,
+      "step": 1048
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8257036782743983e-05,
+      "loss": 1.6043,
+      "step": 1049
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8253286683365855e-05,
+      "loss": 1.5221,
+      "step": 1050
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824953294004347e-05,
+      "loss": 1.5794,
+      "step": 1051
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8245775554434163e-05,
+      "loss": 1.6388,
+      "step": 1052
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8242014528196875e-05,
+      "loss": 1.3068,
+      "step": 1053
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8238249862992144e-05,
+      "loss": 1.5119,
+      "step": 1054
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8234481560482136e-05,
+      "loss": 1.5869,
+      "step": 1055
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8230709622330596e-05,
+      "loss": 1.7378,
+      "step": 1056
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8226934050202904e-05,
+      "loss": 1.4954,
+      "step": 1057
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.822315484576602e-05,
+      "loss": 1.5793,
+      "step": 1058
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8219372010688516e-05,
+      "loss": 1.6469,
+      "step": 1059
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.821558554664057e-05,
+      "loss": 1.6086,
+      "step": 1060
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8211795455293962e-05,
+      "loss": 1.5699,
+      "step": 1061
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.820800173832207e-05,
+      "loss": 1.4399,
+      "step": 1062
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8204204397399875e-05,
+      "loss": 1.6507,
+      "step": 1063
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8200403434203957e-05,
+      "loss": 1.6534,
+      "step": 1064
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8196598850412496e-05,
+      "loss": 1.569,
+      "step": 1065
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.819279064770527e-05,
+      "loss": 1.4887,
+      "step": 1066
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8188978827763654e-05,
+      "loss": 1.7168,
+      "step": 1067
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8185163392270623e-05,
+      "loss": 1.5722,
+      "step": 1068
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8181344342910743e-05,
+      "loss": 1.4351,
+      "step": 1069
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8177521681370183e-05,
+      "loss": 1.6349,
+      "step": 1070
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8173695409336703e-05,
+      "loss": 1.527,
+      "step": 1071
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.816986552849965e-05,
+      "loss": 1.7384,
+      "step": 1072
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.816603204054998e-05,
+      "loss": 1.5194,
+      "step": 1073
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8162194947180228e-05,
+      "loss": 1.6257,
+      "step": 1074
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.815835425008453e-05,
+      "loss": 1.4522,
+      "step": 1075
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.81545099509586e-05,
+      "loss": 1.4869,
+      "step": 1076
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.815066205149976e-05,
+      "loss": 1.7322,
+      "step": 1077
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.814681055340691e-05,
+      "loss": 1.7339,
+      "step": 1078
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.814295545838054e-05,
+      "loss": 1.59,
+      "step": 1079
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8139096768122727e-05,
+      "loss": 1.5837,
+      "step": 1080
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8135234484337148e-05,
+      "loss": 1.6327,
+      "step": 1081
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8131368608729047e-05,
+      "loss": 1.7869,
+      "step": 1082
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8127499143005266e-05,
+      "loss": 1.5936,
+      "step": 1083
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8123626088874232e-05,
+      "loss": 1.7577,
+      "step": 1084
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8119749448045947e-05,
+      "loss": 1.5816,
+      "step": 1085
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8115869222232015e-05,
+      "loss": 1.6702,
+      "step": 1086
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.81119854131456e-05,
+      "loss": 1.4638,
+      "step": 1087
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.810809802250147e-05,
+      "loss": 1.6369,
+      "step": 1088
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8104207052015952e-05,
+      "loss": 1.4726,
+      "step": 1089
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.810031250340697e-05,
+      "loss": 1.4817,
+      "step": 1090
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.809641437839403e-05,
+      "loss": 1.7783,
+      "step": 1091
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.80925126786982e-05,
+      "loss": 1.608,
+      "step": 1092
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.808860740604214e-05,
+      "loss": 1.4716,
+      "step": 1093
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8084698562150086e-05,
+      "loss": 1.6736,
+      "step": 1094
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8080786148747842e-05,
+      "loss": 2.0315,
+      "step": 1095
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8076870167562803e-05,
+      "loss": 1.5921,
+      "step": 1096
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8072950620323926e-05,
+      "loss": 1.6442,
+      "step": 1097
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8069027508761747e-05,
+      "loss": 1.4141,
+      "step": 1098
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8065100834608378e-05,
+      "loss": 1.5471,
+      "step": 1099
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.80611705995975e-05,
+      "loss": 1.6842,
+      "step": 1100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.805723680546437e-05,
+      "loss": 1.6404,
+      "step": 1101
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8053299453945813e-05,
+      "loss": 1.4688,
+      "step": 1102
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8049358546780227e-05,
+      "loss": 1.4906,
+      "step": 1103
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8045414085707578e-05,
+      "loss": 1.6019,
+      "step": 1104
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.80414660724694e-05,
+      "loss": 1.581,
+      "step": 1105
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.80375145088088e-05,
+      "loss": 1.5168,
+      "step": 1106
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8033559396470455e-05,
+      "loss": 1.6971,
+      "step": 1107
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.802960073720059e-05,
+      "loss": 1.742,
+      "step": 1108
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8025638532747027e-05,
+      "loss": 1.5605,
+      "step": 1109
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.802167278485912e-05,
+      "loss": 1.6583,
+      "step": 1110
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8017703495287815e-05,
+      "loss": 1.4529,
+      "step": 1111
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8013730665785605e-05,
+      "loss": 1.512,
+      "step": 1112
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8009754298106553e-05,
+      "loss": 1.5522,
+      "step": 1113
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.800577439400628e-05,
+      "loss": 1.6583,
+      "step": 1114
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8001790955241972e-05,
+      "loss": 1.2873,
+      "step": 1115
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7997803983572372e-05,
+      "loss": 1.3498,
+      "step": 1116
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.799381348075779e-05,
+      "loss": 1.5675,
+      "step": 1117
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7989819448560087e-05,
+      "loss": 1.4491,
+      "step": 1118
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7985821888742687e-05,
+      "loss": 1.5095,
+      "step": 1119
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7981820803070564e-05,
+      "loss": 1.6472,
+      "step": 1120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7977816193310265e-05,
+      "loss": 1.6119,
+      "step": 1121
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7973808061229873e-05,
+      "loss": 1.6157,
+      "step": 1122
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7969796408599042e-05,
+      "loss": 1.5064,
+      "step": 1123
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796578123718897e-05,
+      "loss": 1.5749,
+      "step": 1124
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7961762548772413e-05,
+      "loss": 1.5516,
+      "step": 1125
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7957740345123685e-05,
+      "loss": 1.6512,
+      "step": 1126
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.795371462801864e-05,
+      "loss": 1.5789,
+      "step": 1127
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.794968539923469e-05,
+      "loss": 1.6642,
+      "step": 1128
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7945652660550802e-05,
+      "loss": 1.52,
+      "step": 1129
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7941616413747485e-05,
+      "loss": 1.4107,
+      "step": 1130
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.79375766606068e-05,
+      "loss": 1.6232,
+      "step": 1131
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7933533402912354e-05,
+      "loss": 1.5532,
+      "step": 1132
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7929486642449307e-05,
+      "loss": 1.8103,
+      "step": 1133
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7925436381004356e-05,
+      "loss": 1.7003,
+      "step": 1134
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7921382620365757e-05,
+      "loss": 1.5203,
+      "step": 1135
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.79173253623233e-05,
+      "loss": 1.515,
+      "step": 1136
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7913264608668316e-05,
+      "loss": 1.5786,
+      "step": 1137
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7909200361193698e-05,
+      "loss": 1.7758,
+      "step": 1138
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7905132621693862e-05,
+      "loss": 1.4826,
+      "step": 1139
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.790106139196477e-05,
+      "loss": 1.813,
+      "step": 1140
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.789698667380394e-05,
+      "loss": 1.5417,
+      "step": 1141
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7892908469010404e-05,
+      "loss": 1.5765,
+      "step": 1142
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.788882677938476e-05,
+      "loss": 1.4108,
+      "step": 1143
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7884741606729128e-05,
+      "loss": 1.7524,
+      "step": 1144
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7880652952847164e-05,
+      "loss": 1.482,
+      "step": 1145
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7876560819544077e-05,
+      "loss": 1.4446,
+      "step": 1146
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.78724652086266e-05,
+      "loss": 1.5877,
+      "step": 1147
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7868366121903e-05,
+      "loss": 1.5436,
+      "step": 1148
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7864263561183085e-05,
+      "loss": 1.6447,
+      "step": 1149
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.786015752827819e-05,
+      "loss": 1.4762,
+      "step": 1150
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7856048025001195e-05,
+      "loss": 1.8671,
+      "step": 1151
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7851935053166503e-05,
+      "loss": 1.5372,
+      "step": 1152
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7847818614590045e-05,
+      "loss": 1.4003,
+      "step": 1153
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.784369871108929e-05,
+      "loss": 1.4686,
+      "step": 1154
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7839575344483237e-05,
+      "loss": 1.6435,
+      "step": 1155
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7835448516592408e-05,
+      "loss": 1.554,
+      "step": 1156
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7831318229238858e-05,
+      "loss": 1.6493,
+      "step": 1157
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.782718448424617e-05,
+      "loss": 1.6545,
+      "step": 1158
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7823047283439444e-05,
+      "loss": 1.6444,
+      "step": 1159
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7818906628645322e-05,
+      "loss": 1.6671,
+      "step": 1160
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.781476252169196e-05,
+      "loss": 1.5287,
+      "step": 1161
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7810614964409037e-05,
+      "loss": 1.631,
+      "step": 1162
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7806463958627765e-05,
+      "loss": 1.611,
+      "step": 1163
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7802309506180867e-05,
+      "loss": 1.6021,
+      "step": 1164
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7798151608902595e-05,
+      "loss": 1.4556,
+      "step": 1165
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7793990268628722e-05,
+      "loss": 1.4155,
+      "step": 1166
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7789825487196538e-05,
+      "loss": 1.3908,
+      "step": 1167
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7785657266444853e-05,
+      "loss": 1.4145,
+      "step": 1168
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7781485608213995e-05,
+      "loss": 1.5121,
+      "step": 1169
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7777310514345816e-05,
+      "loss": 1.5729,
+      "step": 1170
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7773131986683675e-05,
+      "loss": 1.3965,
+      "step": 1171
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7768950027072452e-05,
+      "loss": 1.3726,
+      "step": 1172
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7764764637358545e-05,
+      "loss": 1.5704,
+      "step": 1173
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7760575819389857e-05,
+      "loss": 1.5404,
+      "step": 1174
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.775638357501582e-05,
+      "loss": 1.4881,
+      "step": 1175
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7752187906087366e-05,
+      "loss": 1.5534,
+      "step": 1176
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.774798881445694e-05,
+      "loss": 1.7527,
+      "step": 1177
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7743786301978507e-05,
+      "loss": 1.8022,
+      "step": 1178
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7739580370507533e-05,
+      "loss": 1.5887,
+      "step": 1179
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7735371021900993e-05,
+      "loss": 1.4314,
+      "step": 1180
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7731158258017382e-05,
+      "loss": 1.6803,
+      "step": 1181
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7726942080716693e-05,
+      "loss": 1.4763,
+      "step": 1182
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7722722491860427e-05,
+      "loss": 1.4335,
+      "step": 1183
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7718499493311593e-05,
+      "loss": 1.4391,
+      "step": 1184
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7714273086934705e-05,
+      "loss": 1.4964,
+      "step": 1185
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7710043274595785e-05,
+      "loss": 1.5693,
+      "step": 1186
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7705810058162354e-05,
+      "loss": 1.4437,
+      "step": 1187
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7701573439503437e-05,
+      "loss": 1.6737,
+      "step": 1188
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7697333420489565e-05,
+      "loss": 1.725,
+      "step": 1189
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769309000299276e-05,
+      "loss": 1.6491,
+      "step": 1190
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.768884318888656e-05,
+      "loss": 1.7634,
+      "step": 1191
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.768459298004599e-05,
+      "loss": 1.7404,
+      "step": 1192
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7680339378347575e-05,
+      "loss": 1.561,
+      "step": 1193
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7676082385669343e-05,
+      "loss": 1.4054,
+      "step": 1194
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7671822003890825e-05,
+      "loss": 1.6459,
+      "step": 1195
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.766755823489303e-05,
+      "loss": 1.4519,
+      "step": 1196
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7663291080558477e-05,
+      "loss": 1.7228,
+      "step": 1197
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.765902054277118e-05,
+      "loss": 1.4606,
+      "step": 1198
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7654746623416637e-05,
+      "loss": 1.7437,
+      "step": 1199
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7650469324381843e-05,
+      "loss": 1.6339,
+      "step": 1200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7646188647555293e-05,
+      "loss": 1.6092,
+      "step": 1201
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7641904594826965e-05,
+      "loss": 1.6544,
+      "step": 1202
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7637617168088327e-05,
+      "loss": 1.6722,
+      "step": 1203
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7633326369232338e-05,
+      "loss": 1.4978,
+      "step": 1204
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7629032200153456e-05,
+      "loss": 1.5489,
+      "step": 1205
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.762473466274761e-05,
+      "loss": 1.5219,
+      "step": 1206
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.762043375891223e-05,
+      "loss": 1.5032,
+      "step": 1207
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7616129490546217e-05,
+      "loss": 1.6929,
+      "step": 1208
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7611821859549977e-05,
+      "loss": 1.4355,
+      "step": 1209
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7607510867825385e-05,
+      "loss": 1.5013,
+      "step": 1210
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7603196517275808e-05,
+      "loss": 1.7467,
+      "step": 1211
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7598878809806095e-05,
+      "loss": 1.6629,
+      "step": 1212
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.759455774732257e-05,
+      "loss": 1.463,
+      "step": 1213
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7590233331733054e-05,
+      "loss": 1.5703,
+      "step": 1214
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.758590556494683e-05,
+      "loss": 1.4868,
+      "step": 1215
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7581574448874668e-05,
+      "loss": 1.4594,
+      "step": 1216
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7577239985428825e-05,
+      "loss": 1.8629,
+      "step": 1217
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7572902176523023e-05,
+      "loss": 1.5319,
+      "step": 1218
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.756856102407247e-05,
+      "loss": 1.4606,
+      "step": 1219
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7564216529993845e-05,
+      "loss": 1.8026,
+      "step": 1220
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7559868696205304e-05,
+      "loss": 1.6435,
+      "step": 1221
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7555517524626476e-05,
+      "loss": 1.4018,
+      "step": 1222
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7551163017178473e-05,
+      "loss": 1.5112,
+      "step": 1223
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7546805175783866e-05,
+      "loss": 1.695,
+      "step": 1224
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7542444002366705e-05,
+      "loss": 1.5983,
+      "step": 1225
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7538079498852512e-05,
+      "loss": 1.5056,
+      "step": 1226
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.753371166716828e-05,
+      "loss": 1.2952,
+      "step": 1227
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7529340509242464e-05,
+      "loss": 1.3561,
+      "step": 1228
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7524966027004995e-05,
+      "loss": 1.6989,
+      "step": 1229
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.752058822238727e-05,
+      "loss": 1.4933,
+      "step": 1230
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7516207097322154e-05,
+      "loss": 1.3344,
+      "step": 1231
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7511822653743973e-05,
+      "loss": 1.4907,
+      "step": 1232
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7507434893588525e-05,
+      "loss": 1.5569,
+      "step": 1233
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7503043818793068e-05,
+      "loss": 1.5218,
+      "step": 1234
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.749864943129632e-05,
+      "loss": 1.6061,
+      "step": 1235
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7494251733038473e-05,
+      "loss": 1.6357,
+      "step": 1236
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748985072596117e-05,
+      "loss": 1.6734,
+      "step": 1237
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748544641200752e-05,
+      "loss": 1.6827,
+      "step": 1238
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748103879312209e-05,
+      "loss": 1.5733,
+      "step": 1239
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7476627871250905e-05,
+      "loss": 1.6196,
+      "step": 1240
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7472213648341455e-05,
+      "loss": 1.627,
+      "step": 1241
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7467796126342677e-05,
+      "loss": 1.5575,
+      "step": 1242
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.746337530720497e-05,
+      "loss": 1.4645,
+      "step": 1243
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7458951192880196e-05,
+      "loss": 1.424,
+      "step": 1244
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.745452378532166e-05,
+      "loss": 1.703,
+      "step": 1245
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7450093086484122e-05,
+      "loss": 1.6646,
+      "step": 1246
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7445659098323807e-05,
+      "loss": 1.8234,
+      "step": 1247
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7441221822798385e-05,
+      "loss": 1.5726,
+      "step": 1248
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7436781261866965e-05,
+      "loss": 1.5187,
+      "step": 1249
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.743233741749013e-05,
+      "loss": 1.7086,
+      "step": 1250
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7427890291629895e-05,
+      "loss": 1.5691,
+      "step": 1251
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7423439886249732e-05,
+      "loss": 1.4226,
+      "step": 1252
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7418986203314557e-05,
+      "loss": 1.6106,
+      "step": 1253
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7414529244790733e-05,
+      "loss": 1.4951,
+      "step": 1254
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7410069012646076e-05,
+      "loss": 1.5656,
+      "step": 1255
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.740560550884984e-05,
+      "loss": 1.5093,
+      "step": 1256
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.740113873537273e-05,
+      "loss": 1.7528,
+      "step": 1257
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.739666869418688e-05,
+      "loss": 1.4822,
+      "step": 1258
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7392195387265888e-05,
+      "loss": 1.3372,
+      "step": 1259
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7387718816584775e-05,
+      "loss": 1.5169,
+      "step": 1260
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7383238984120016e-05,
+      "loss": 1.4587,
+      "step": 1261
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7378755891849522e-05,
+      "loss": 1.4014,
+      "step": 1262
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.737426954175264e-05,
+      "loss": 1.6253,
+      "step": 1263
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7369779935810152e-05,
+      "loss": 1.4406,
+      "step": 1264
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.73652870760043e-05,
+      "loss": 1.6697,
+      "step": 1265
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7360790964318727e-05,
+      "loss": 1.5923,
+      "step": 1266
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7356291602738542e-05,
+      "loss": 1.6754,
+      "step": 1267
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7351788993250274e-05,
+      "loss": 1.4268,
+      "step": 1268
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.734728313784189e-05,
+      "loss": 1.5145,
+      "step": 1269
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7342774038502792e-05,
+      "loss": 1.6265,
+      "step": 1270
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.733826169722381e-05,
+      "loss": 1.3404,
+      "step": 1271
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7333746115997204e-05,
+      "loss": 1.3584,
+      "step": 1272
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.732922729681667e-05,
+      "loss": 1.6537,
+      "step": 1273
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7324705241677335e-05,
+      "loss": 1.5836,
+      "step": 1274
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.732017995257575e-05,
+      "loss": 1.6354,
+      "step": 1275
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7315651431509893e-05,
+      "loss": 1.5083,
+      "step": 1276
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7311119680479168e-05,
+      "loss": 1.4178,
+      "step": 1277
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7306584701484414e-05,
+      "loss": 1.7776,
+      "step": 1278
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7302046496527882e-05,
+      "loss": 1.6633,
+      "step": 1279
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7297505067613262e-05,
+      "loss": 1.7077,
+      "step": 1280
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7292960416745657e-05,
+      "loss": 1.5804,
+      "step": 1281
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.728841254593159e-05,
+      "loss": 1.671,
+      "step": 1282
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7283861457179022e-05,
+      "loss": 1.4001,
+      "step": 1283
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7279307152497316e-05,
+      "loss": 1.6682,
+      "step": 1284
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.727474963389727e-05,
+      "loss": 1.446,
+      "step": 1285
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.727018890339108e-05,
+      "loss": 1.5421,
+      "step": 1286
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.726562496299239e-05,
+      "loss": 1.6631,
+      "step": 1287
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.726105781471623e-05,
+      "loss": 1.6309,
+      "step": 1288
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7256487460579075e-05,
+      "loss": 1.604,
+      "step": 1289
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7251913902598795e-05,
+      "loss": 1.6664,
+      "step": 1290
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.724733714279468e-05,
+      "loss": 1.4303,
+      "step": 1291
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7242757183187435e-05,
+      "loss": 1.6895,
+      "step": 1292
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7238174025799183e-05,
+      "loss": 1.6849,
+      "step": 1293
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7233587672653446e-05,
+      "loss": 1.4347,
+      "step": 1294
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7228998125775175e-05,
+      "loss": 1.4284,
+      "step": 1295
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.722440538719071e-05,
+      "loss": 1.5442,
+      "step": 1296
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7219809458927813e-05,
+      "loss": 1.6091,
+      "step": 1297
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.721521034301566e-05,
+      "loss": 1.5812,
+      "step": 1298
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.721060804148482e-05,
+      "loss": 1.5366,
+      "step": 1299
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7206002556367276e-05,
+      "loss": 1.5844,
+      "step": 1300
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7201393889696413e-05,
+      "loss": 1.2277,
+      "step": 1301
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.719678204350703e-05,
+      "loss": 1.5415,
+      "step": 1302
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7192167019835314e-05,
+      "loss": 1.7618,
+      "step": 1303
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7187548820718875e-05,
+      "loss": 1.5605,
+      "step": 1304
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7182927448196705e-05,
+      "loss": 1.6087,
+      "step": 1305
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.717830290430921e-05,
+      "loss": 1.4937,
+      "step": 1306
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.717367519109819e-05,
+      "loss": 1.6544,
+      "step": 1307
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.716904431060685e-05,
+      "loss": 1.583,
+      "step": 1308
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7164410264879788e-05,
+      "loss": 1.5388,
+      "step": 1309
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7159773055963003e-05,
+      "loss": 1.5392,
+      "step": 1310
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7155132685903888e-05,
+      "loss": 1.6567,
+      "step": 1311
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.715048915675123e-05,
+      "loss": 1.6449,
+      "step": 1312
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7145842470555218e-05,
+      "loss": 1.5239,
+      "step": 1313
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.714119262936743e-05,
+      "loss": 1.6579,
+      "step": 1314
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.713653963524084e-05,
+      "loss": 1.6511,
+      "step": 1315
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7131883490229803e-05,
+      "loss": 1.7842,
+      "step": 1316
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.712722419639008e-05,
+      "loss": 1.7427,
+      "step": 1317
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7122561755778816e-05,
+      "loss": 1.6652,
+      "step": 1318
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7117896170454542e-05,
+      "loss": 1.656,
+      "step": 1319
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7113227442477182e-05,
+      "loss": 1.5847,
+      "step": 1320
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.710855557390805e-05,
+      "loss": 1.62,
+      "step": 1321
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.710388056680984e-05,
+      "loss": 1.7254,
+      "step": 1322
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7099202423246632e-05,
+      "loss": 1.5099,
+      "step": 1323
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.70945211452839e-05,
+      "loss": 1.5864,
+      "step": 1324
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7089836734988485e-05,
+      "loss": 1.6231,
+      "step": 1325
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7085149194428628e-05,
+      "loss": 1.614,
+      "step": 1326
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7080458525673948e-05,
+      "loss": 1.6228,
+      "step": 1327
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7075764730795437e-05,
+      "loss": 1.4602,
+      "step": 1328
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7071067811865477e-05,
+      "loss": 1.4176,
+      "step": 1329
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.706636777095782e-05,
+      "loss": 1.6771,
+      "step": 1330
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7061664610147605e-05,
+      "loss": 1.6946,
+      "step": 1331
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7056958331511343e-05,
+      "loss": 1.5463,
+      "step": 1332
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7052248937126926e-05,
+      "loss": 1.6781,
+      "step": 1333
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7047536429073617e-05,
+      "loss": 1.7608,
+      "step": 1334
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7042820809432057e-05,
+      "loss": 1.6546,
+      "step": 1335
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7038102080284257e-05,
+      "loss": 1.5523,
+      "step": 1336
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7033380243713602e-05,
+      "loss": 1.5666,
+      "step": 1337
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7028655301804858e-05,
+      "loss": 1.472,
+      "step": 1338
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.702392725664415e-05,
+      "loss": 1.5898,
+      "step": 1339
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7019196110318977e-05,
+      "loss": 1.6487,
+      "step": 1340
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7014461864918205e-05,
+      "loss": 1.6211,
+      "step": 1341
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.700972452253208e-05,
+      "loss": 1.561,
+      "step": 1342
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.70049840852522e-05,
+      "loss": 1.6095,
+      "step": 1343
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7000240555171536e-05,
+      "loss": 1.6042,
+      "step": 1344
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6995493934384427e-05,
+      "loss": 1.6868,
+      "step": 1345
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.699074422498657e-05,
+      "loss": 1.4898,
+      "step": 1346
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6985991429075038e-05,
+      "loss": 1.533,
+      "step": 1347
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6981235548748247e-05,
+      "loss": 1.5295,
+      "step": 1348
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6976476586105996e-05,
+      "loss": 1.5263,
+      "step": 1349
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.697171454324943e-05,
+      "loss": 1.3823,
+      "step": 1350
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6966949422281058e-05,
+      "loss": 1.3681,
+      "step": 1351
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6962181225304756e-05,
+      "loss": 1.4484,
+      "step": 1352
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6957409954425747e-05,
+      "loss": 1.6771,
+      "step": 1353
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.695263561175061e-05,
+      "loss": 1.6545,
+      "step": 1354
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6947858199387296e-05,
+      "loss": 1.6246,
+      "step": 1355
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.694307771944509e-05,
+      "loss": 1.643,
+      "step": 1356
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6938294174034652e-05,
+      "loss": 1.4888,
+      "step": 1357
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6933507565267982e-05,
+      "loss": 1.4273,
+      "step": 1358
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.692871789525844e-05,
+      "loss": 1.4526,
+      "step": 1359
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6923925166120722e-05,
+      "loss": 1.5072,
+      "step": 1360
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.69191293799709e-05,
+      "loss": 1.7764,
+      "step": 1361
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6914330538926377e-05,
+      "loss": 1.3538,
+      "step": 1362
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6909528645105908e-05,
+      "loss": 1.5067,
+      "step": 1363
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.69047237006296e-05,
+      "loss": 1.5913,
+      "step": 1364
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6899915707618905e-05,
+      "loss": 1.602,
+      "step": 1365
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6895104668196622e-05,
+      "loss": 1.6007,
+      "step": 1366
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6890290584486892e-05,
+      "loss": 1.5227,
+      "step": 1367
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6885473458615204e-05,
+      "loss": 1.6491,
+      "step": 1368
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6880653292708383e-05,
+      "loss": 1.5674,
+      "step": 1369
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6875830088894604e-05,
+      "loss": 1.5696,
+      "step": 1370
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.687100384930338e-05,
+      "loss": 1.564,
+      "step": 1371
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.686617457606557e-05,
+      "loss": 1.775,
+      "step": 1372
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6861342271313363e-05,
+      "loss": 1.618,
+      "step": 1373
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.685650693718029e-05,
+      "loss": 1.4443,
+      "step": 1374
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.685166857580122e-05,
+      "loss": 1.6217,
+      "step": 1375
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.684682718931236e-05,
+      "loss": 1.5795,
+      "step": 1376
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6841982779851254e-05,
+      "loss": 1.5463,
+      "step": 1377
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6837135349556773e-05,
+      "loss": 1.4284,
+      "step": 1378
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.683228490056913e-05,
+      "loss": 1.542,
+      "step": 1379
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.682743143502987e-05,
+      "loss": 1.5931,
+      "step": 1380
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6822574955081862e-05,
+      "loss": 1.583,
+      "step": 1381
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6817715462869314e-05,
+      "loss": 1.6129,
+      "step": 1382
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6812852960537763e-05,
+      "loss": 1.5822,
+      "step": 1383
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.680798745023407e-05,
+      "loss": 1.8158,
+      "step": 1384
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.680311893410643e-05,
+      "loss": 1.7005,
+      "step": 1385
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6798247414304363e-05,
+      "loss": 1.6062,
+      "step": 1386
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6793372892978716e-05,
+      "loss": 1.7163,
+      "step": 1387
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6788495372281656e-05,
+      "loss": 1.4449,
+      "step": 1388
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.678361485436668e-05,
+      "loss": 1.5955,
+      "step": 1389
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6778731341388607e-05,
+      "loss": 1.604,
+      "step": 1390
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6773844835503574e-05,
+      "loss": 1.5757,
+      "step": 1391
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.676895533886905e-05,
+      "loss": 1.395,
+      "step": 1392
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6764062853643813e-05,
+      "loss": 1.6502,
+      "step": 1393
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.675916738198797e-05,
+      "loss": 1.4245,
+      "step": 1394
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6754268926062936e-05,
+      "loss": 1.6028,
+      "step": 1395
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6749367488031455e-05,
+      "loss": 1.6762,
+      "step": 1396
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6744463070057574e-05,
+      "loss": 1.5779,
+      "step": 1397
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6739555674306675e-05,
+      "loss": 1.6024,
+      "step": 1398
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6734645302945434e-05,
+      "loss": 1.5692,
+      "step": 1399
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6729731958141852e-05,
+      "loss": 1.7545,
+      "step": 1400
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6724815642065245e-05,
+      "loss": 1.5812,
+      "step": 1401
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6719896356886238e-05,
+      "loss": 1.5997,
+      "step": 1402
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.671497410477676e-05,
+      "loss": 1.4719,
+      "step": 1403
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6710048887910058e-05,
+      "loss": 1.4908,
+      "step": 1404
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6705120708460686e-05,
+      "loss": 1.5513,
+      "step": 1405
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6700189568604506e-05,
+      "loss": 1.636,
+      "step": 1406
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6695255470518687e-05,
+      "loss": 1.489,
+      "step": 1407
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6690318416381706e-05,
+      "loss": 1.5321,
+      "step": 1408
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6685378408373338e-05,
+      "loss": 1.4402,
+      "step": 1409
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6680435448674673e-05,
+      "loss": 1.6498,
+      "step": 1410
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6675489539468094e-05,
+      "loss": 1.5796,
+      "step": 1411
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.667054068293729e-05,
+      "loss": 1.5349,
+      "step": 1412
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.666558888126726e-05,
+      "loss": 1.5557,
+      "step": 1413
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6660634136644284e-05,
+      "loss": 1.5048,
+      "step": 1414
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.665567645125596e-05,
+      "loss": 1.4629,
+      "step": 1415
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6650715827291178e-05,
+      "loss": 1.7401,
+      "step": 1416
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.664575226694012e-05,
+      "loss": 1.6357,
+      "step": 1417
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6640785772394275e-05,
+      "loss": 1.5159,
+      "step": 1418
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6635816345846413e-05,
+      "loss": 1.3262,
+      "step": 1419
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6630843989490612e-05,
+      "loss": 1.3812,
+      "step": 1420
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6625868705522238e-05,
+      "loss": 1.4243,
+      "step": 1421
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6620890496137953e-05,
+      "loss": 1.6111,
+      "step": 1422
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.66159093635357e-05,
+      "loss": 1.6942,
+      "step": 1423
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.661092530991473e-05,
+      "loss": 1.6799,
+      "step": 1424
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.660593833747557e-05,
+      "loss": 1.5261,
+      "step": 1425
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6600948448420035e-05,
+      "loss": 1.619,
+      "step": 1426
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.659595564495124e-05,
+      "loss": 1.6036,
+      "step": 1427
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6590959929273574e-05,
+      "loss": 1.6133,
+      "step": 1428
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6585961303592716e-05,
+      "loss": 1.4153,
+      "step": 1429
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6580959770115637e-05,
+      "loss": 1.6207,
+      "step": 1430
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6575955331050585e-05,
+      "loss": 1.5897,
+      "step": 1431
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6570947988607084e-05,
+      "loss": 1.5924,
+      "step": 1432
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6565937744995955e-05,
+      "loss": 1.5328,
+      "step": 1433
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.656092460242929e-05,
+      "loss": 1.5606,
+      "step": 1434
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6555908563120457e-05,
+      "loss": 1.4874,
+      "step": 1435
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.655088962928412e-05,
+      "loss": 1.5164,
+      "step": 1436
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6545867803136203e-05,
+      "loss": 1.4275,
+      "step": 1437
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6540843086893915e-05,
+      "loss": 1.5823,
+      "step": 1438
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6535815482775745e-05,
+      "loss": 1.46,
+      "step": 1439
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6530784993001448e-05,
+      "loss": 1.529,
+      "step": 1440
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6525751619792056e-05,
+      "loss": 1.7576,
+      "step": 1441
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6520715365369878e-05,
+      "loss": 1.5049,
+      "step": 1442
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6515676231958488e-05,
+      "loss": 1.491,
+      "step": 1443
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6510634221782743e-05,
+      "loss": 1.4462,
+      "step": 1444
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.650558933706876e-05,
+      "loss": 1.5759,
+      "step": 1445
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.650054158004392e-05,
+      "loss": 1.6223,
+      "step": 1446
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6495490952936898e-05,
+      "loss": 1.6145,
+      "step": 1447
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.64904374579776e-05,
+      "loss": 1.4701,
+      "step": 1448
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6485381097397223e-05,
+      "loss": 1.3098,
+      "step": 1449
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6480321873428225e-05,
+      "loss": 1.4865,
+      "step": 1450
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.647525978830432e-05,
+      "loss": 1.6361,
+      "step": 1451
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6470194844260497e-05,
+      "loss": 1.5479,
+      "step": 1452
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6465127043533e-05,
+      "loss": 1.5428,
+      "step": 1453
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6460056388359336e-05,
+      "loss": 1.5231,
+      "step": 1454
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.645498288097827e-05,
+      "loss": 1.5808,
+      "step": 1455
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6449906523629825e-05,
+      "loss": 1.4403,
+      "step": 1456
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.644482731855529e-05,
+      "loss": 1.4955,
+      "step": 1457
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.643974526799721e-05,
+      "loss": 1.548,
+      "step": 1458
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6434660374199377e-05,
+      "loss": 1.4527,
+      "step": 1459
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6429572639406845e-05,
+      "loss": 1.635,
+      "step": 1460
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6424482065865925e-05,
+      "loss": 1.743,
+      "step": 1461
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6419388655824175e-05,
+      "loss": 1.5494,
+      "step": 1462
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.641429241153041e-05,
+      "loss": 1.7056,
+      "step": 1463
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6409193335234698e-05,
+      "loss": 1.6353,
+      "step": 1464
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6404091429188348e-05,
+      "loss": 1.6106,
+      "step": 1465
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6398986695643926e-05,
+      "loss": 1.6315,
+      "step": 1466
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6393879136855247e-05,
+      "loss": 1.5862,
+      "step": 1467
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6388768755077374e-05,
+      "loss": 1.6772,
+      "step": 1468
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.638365555256661e-05,
+      "loss": 1.6833,
+      "step": 1469
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6378539531580502e-05,
+      "loss": 1.7268,
+      "step": 1470
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6373420694377857e-05,
+      "loss": 1.7258,
+      "step": 1471
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.636829904321871e-05,
+      "loss": 1.5347,
+      "step": 1472
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.636317458036434e-05,
+      "loss": 1.5876,
+      "step": 1473
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.635804730807728e-05,
+      "loss": 1.4784,
+      "step": 1474
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6352917228621284e-05,
+      "loss": 1.691,
+      "step": 1475
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6347784344261365e-05,
+      "loss": 1.4505,
+      "step": 1476
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.634264865726375e-05,
+      "loss": 1.5875,
+      "step": 1477
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6337510169895935e-05,
+      "loss": 1.8371,
+      "step": 1478
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.633236888442663e-05,
+      "loss": 1.5814,
+      "step": 1479
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6327224803125782e-05,
+      "loss": 1.6374,
+      "step": 1480
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.632207792826458e-05,
+      "loss": 1.4059,
+      "step": 1481
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6316928262115443e-05,
+      "loss": 1.592,
+      "step": 1482
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.631177580695202e-05,
+      "loss": 1.6163,
+      "step": 1483
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6306620565049196e-05,
+      "loss": 1.3837,
+      "step": 1484
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6301462538683087e-05,
+      "loss": 1.6687,
+      "step": 1485
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6296301730131028e-05,
+      "loss": 1.5809,
+      "step": 1486
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6291138141671598e-05,
+      "loss": 1.4827,
+      "step": 1487
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6285971775584587e-05,
+      "loss": 1.8753,
+      "step": 1488
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.628080263415103e-05,
+      "loss": 1.5419,
+      "step": 1489
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6275630719653167e-05,
+      "loss": 1.4471,
+      "step": 1490
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6270456034374477e-05,
+      "loss": 1.5168,
+      "step": 1491
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6265278580599657e-05,
+      "loss": 1.4473,
+      "step": 1492
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6260098360614626e-05,
+      "loss": 1.5146,
+      "step": 1493
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6254915376706526e-05,
+      "loss": 1.6374,
+      "step": 1494
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6249729631163718e-05,
+      "loss": 1.4926,
+      "step": 1495
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6244541126275786e-05,
+      "loss": 1.4727,
+      "step": 1496
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6239349864333527e-05,
+      "loss": 1.6858,
+      "step": 1497
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6234155847628955e-05,
+      "loss": 1.5849,
+      "step": 1498
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6228959078455306e-05,
+      "loss": 1.4964,
+      "step": 1499
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6223759559107028e-05,
+      "loss": 1.6013,
+      "step": 1500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.621855729187978e-05,
+      "loss": 1.5135,
+      "step": 1501
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6213352279070443e-05,
+      "loss": 1.3919,
+      "step": 1502
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.62081445229771e-05,
+      "loss": 1.5336,
+      "step": 1503
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6202934025899052e-05,
+      "loss": 1.68,
+      "step": 1504
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.619772079013681e-05,
+      "loss": 1.4999,
+      "step": 1505
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6192504817992086e-05,
+      "loss": 1.4968,
+      "step": 1506
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6187286111767812e-05,
+      "loss": 1.4836,
+      "step": 1507
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.618206467376812e-05,
+      "loss": 1.6276,
+      "step": 1508
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6176840506298345e-05,
+      "loss": 1.488,
+      "step": 1509
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.617161361166504e-05,
+      "loss": 1.5407,
+      "step": 1510
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.616638399217595e-05,
+      "loss": 1.5145,
+      "step": 1511
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6161151650140027e-05,
+      "loss": 1.5094,
+      "step": 1512
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6155916587867424e-05,
+      "loss": 1.6386,
+      "step": 1513
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6150678807669494e-05,
+      "loss": 1.7514,
+      "step": 1514
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.61454383118588e-05,
+      "loss": 1.7054,
+      "step": 1515
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6140195102749086e-05,
+      "loss": 1.5409,
+      "step": 1516
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.613494918265531e-05,
+      "loss": 1.6018,
+      "step": 1517
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.612970055389362e-05,
+      "loss": 1.4723,
+      "step": 1518
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6124449218781358e-05,
+      "loss": 1.3349,
+      "step": 1519
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.611919517963707e-05,
+      "loss": 1.5824,
+      "step": 1520
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.611393843878048e-05,
+      "loss": 1.465,
+      "step": 1521
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6108678998532522e-05,
+      "loss": 1.4788,
+      "step": 1522
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6103416861215314e-05,
+      "loss": 1.7352,
+      "step": 1523
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.609815202915216e-05,
+      "loss": 1.4858,
+      "step": 1524
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6092884504667562e-05,
+      "loss": 1.727,
+      "step": 1525
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.608761429008721e-05,
+      "loss": 1.4953,
+      "step": 1526
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.608234138773797e-05,
+      "loss": 1.2879,
+      "step": 1527
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6077065799947914e-05,
+      "loss": 1.5736,
+      "step": 1528
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6071787529046288e-05,
+      "loss": 1.4467,
+      "step": 1529
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6066506577363518e-05,
+      "loss": 1.578,
+      "step": 1530
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6061222947231224e-05,
+      "loss": 1.518,
+      "step": 1531
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6055936640982207e-05,
+      "loss": 1.4645,
+      "step": 1532
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.605064766095044e-05,
+      "loss": 1.5938,
+      "step": 1533
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.604535600947109e-05,
+      "loss": 1.4999,
+      "step": 1534
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6040061688880494e-05,
+      "loss": 1.5532,
+      "step": 1535
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.603476470151617e-05,
+      "loss": 1.8662,
+      "step": 1536
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6029465049716813e-05,
+      "loss": 1.6137,
+      "step": 1537
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6024162735822296e-05,
+      "loss": 1.6793,
+      "step": 1538
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6018857762173672e-05,
+      "loss": 1.6422,
+      "step": 1539
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.601355013111315e-05,
+      "loss": 1.4664,
+      "step": 1540
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.600823984498414e-05,
+      "loss": 1.4966,
+      "step": 1541
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6002926906131202e-05,
+      "loss": 1.5487,
+      "step": 1542
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5997611316900075e-05,
+      "loss": 1.5593,
+      "step": 1543
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5992293079637665e-05,
+      "loss": 1.38,
+      "step": 1544
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5986972196692053e-05,
+      "loss": 1.4503,
+      "step": 1545
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5981648670412484e-05,
+      "loss": 1.6433,
+      "step": 1546
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5976322503149373e-05,
+      "loss": 1.4204,
+      "step": 1547
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5970993697254294e-05,
+      "loss": 1.4998,
+      "step": 1548
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5965662255080003e-05,
+      "loss": 1.3519,
+      "step": 1549
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5960328178980393e-05,
+      "loss": 1.4369,
+      "step": 1550
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5954991471310542e-05,
+      "loss": 1.5891,
+      "step": 1551
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5949652134426686e-05,
+      "loss": 1.6511,
+      "step": 1552
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5944310170686215e-05,
+      "loss": 1.6441,
+      "step": 1553
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5938965582447683e-05,
+      "loss": 1.3753,
+      "step": 1554
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5933618372070805e-05,
+      "loss": 1.461,
+      "step": 1555
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5928268541916444e-05,
+      "loss": 1.7279,
+      "step": 1556
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5922916094346633e-05,
+      "loss": 1.4245,
+      "step": 1557
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.591756103172455e-05,
+      "loss": 1.6379,
+      "step": 1558
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5912203356414532e-05,
+      "loss": 1.4998,
+      "step": 1559
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5906843070782077e-05,
+      "loss": 1.5344,
+      "step": 1560
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5901480177193816e-05,
+      "loss": 1.5145,
+      "step": 1561
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5896114678017555e-05,
+      "loss": 1.6074,
+      "step": 1562
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.589074657562223e-05,
+      "loss": 1.5691,
+      "step": 1563
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5885375872377942e-05,
+      "loss": 1.5353,
+      "step": 1564
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5880002570655925e-05,
+      "loss": 1.382,
+      "step": 1565
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5874626672828573e-05,
+      "loss": 1.7947,
+      "step": 1566
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5869248181269427e-05,
+      "loss": 1.594,
+      "step": 1567
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.586386709835316e-05,
+      "loss": 1.3782,
+      "step": 1568
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.58584834264556e-05,
+      "loss": 1.6009,
+      "step": 1569
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.585309716795372e-05,
+      "loss": 1.6168,
+      "step": 1570
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5847708325225618e-05,
+      "loss": 1.4577,
+      "step": 1571
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.584231690065056e-05,
+      "loss": 1.6535,
+      "step": 1572
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5836922896608923e-05,
+      "loss": 1.4422,
+      "step": 1573
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.583152631548225e-05,
+      "loss": 1.6535,
+      "step": 1574
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5826127159653203e-05,
+      "loss": 1.5586,
+      "step": 1575
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5820725431505584e-05,
+      "loss": 1.5447,
+      "step": 1576
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.581532113342434e-05,
+      "loss": 1.4033,
+      "step": 1577
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5809914267795542e-05,
+      "loss": 1.4737,
+      "step": 1578
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5804504837006396e-05,
+      "loss": 1.5754,
+      "step": 1579
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.579909284344524e-05,
+      "loss": 1.6443,
+      "step": 1580
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5793678289501564e-05,
+      "loss": 1.4774,
+      "step": 1581
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5788261177565952e-05,
+      "loss": 1.7905,
+      "step": 1582
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.578284151003015e-05,
+      "loss": 1.6006,
+      "step": 1583
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5777419289287008e-05,
+      "loss": 1.4872,
+      "step": 1584
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.577199451773052e-05,
+      "loss": 1.6706,
+      "step": 1585
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.57665671977558e-05,
+      "loss": 1.6523,
+      "step": 1586
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5761137331759084e-05,
+      "loss": 1.5904,
+      "step": 1587
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5755704922137742e-05,
+      "loss": 1.5417,
+      "step": 1588
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5750269971290254e-05,
+      "loss": 1.429,
+      "step": 1589
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5744832481616235e-05,
+      "loss": 1.616,
+      "step": 1590
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5739392455516408e-05,
+      "loss": 1.3763,
+      "step": 1591
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5733949895392626e-05,
+      "loss": 1.4313,
+      "step": 1592
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5728504803647857e-05,
+      "loss": 1.6783,
+      "step": 1593
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.572305718268619e-05,
+      "loss": 1.3744,
+      "step": 1594
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.571760703491282e-05,
+      "loss": 1.3904,
+      "step": 1595
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5712154362734073e-05,
+      "loss": 1.6997,
+      "step": 1596
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.570669916855738e-05,
+      "loss": 1.7354,
+      "step": 1597
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.570124145479128e-05,
+      "loss": 1.4967,
+      "step": 1598
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5695781223845442e-05,
+      "loss": 1.4931,
+      "step": 1599
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5690318478130634e-05,
+      "loss": 1.5717,
+      "step": 1600
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5684853220058727e-05,
+      "loss": 1.3101,
+      "step": 1601
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5679385452042724e-05,
+      "loss": 1.4513,
+      "step": 1602
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5673915176496716e-05,
+      "loss": 1.6303,
+      "step": 1603
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5668442395835907e-05,
+      "loss": 1.5739,
+      "step": 1604
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5662967112476613e-05,
+      "loss": 1.3594,
+      "step": 1605
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5657489328836242e-05,
+      "loss": 1.5619,
+      "step": 1606
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5652009047333322e-05,
+      "loss": 1.3912,
+      "step": 1607
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5646526270387473e-05,
+      "loss": 1.5692,
+      "step": 1608
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.564104100041942e-05,
+      "loss": 1.4883,
+      "step": 1609
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5635553239850987e-05,
+      "loss": 1.5308,
+      "step": 1610
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.56300629911051e-05,
+      "loss": 1.4778,
+      "step": 1611
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.562457025660578e-05,
+      "loss": 1.7898,
+      "step": 1612
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.561907503877815e-05,
+      "loss": 1.525,
+      "step": 1613
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5613577340048423e-05,
+      "loss": 1.6107,
+      "step": 1614
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.560807716284392e-05,
+      "loss": 1.6002,
+      "step": 1615
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5602574509593043e-05,
+      "loss": 1.7649,
+      "step": 1616
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5597069382725293e-05,
+      "loss": 1.6904,
+      "step": 1617
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5591561784671266e-05,
+      "loss": 1.5982,
+      "step": 1618
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5586051717862634e-05,
+      "loss": 1.2394,
+      "step": 1619
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5580539184732188e-05,
+      "loss": 1.5211,
+      "step": 1620
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.557502418771377e-05,
+      "loss": 1.6939,
+      "step": 1621
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5569506729242347e-05,
+      "loss": 1.7091,
+      "step": 1622
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5563986811753948e-05,
+      "loss": 1.5532,
+      "step": 1623
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5558464437685697e-05,
+      "loss": 1.494,
+      "step": 1624
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5552939609475803e-05,
+      "loss": 1.6485,
+      "step": 1625
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.554741232956355e-05,
+      "loss": 1.4488,
+      "step": 1626
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.554188260038932e-05,
+      "loss": 1.3459,
+      "step": 1627
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5536350424394565e-05,
+      "loss": 1.3798,
+      "step": 1628
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.553081580402182e-05,
+      "loss": 1.2364,
+      "step": 1629
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5525278741714692e-05,
+      "loss": 1.2505,
+      "step": 1630
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.551973923991788e-05,
+      "loss": 1.3651,
+      "step": 1631
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.551419730107716e-05,
+      "loss": 1.2561,
+      "step": 1632
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5508652927639362e-05,
+      "loss": 1.1947,
+      "step": 1633
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5503106122052417e-05,
+      "loss": 1.2109,
+      "step": 1634
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5497556886765316e-05,
+      "loss": 1.1964,
+      "step": 1635
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5492005224228128e-05,
+      "loss": 0.9979,
+      "step": 1636
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.548645113689199e-05,
+      "loss": 1.2685,
+      "step": 1637
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.548089462720911e-05,
+      "loss": 1.2953,
+      "step": 1638
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5475335697632768e-05,
+      "loss": 1.3509,
+      "step": 1639
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.546977435061731e-05,
+      "loss": 1.3724,
+      "step": 1640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5464210588618148e-05,
+      "loss": 1.2277,
+      "step": 1641
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5458644414091767e-05,
+      "loss": 1.1263,
+      "step": 1642
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.545307582949571e-05,
+      "loss": 1.1656,
+      "step": 1643
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.544750483728859e-05,
+      "loss": 1.3924,
+      "step": 1644
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.544193143993008e-05,
+      "loss": 1.2156,
+      "step": 1645
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.543635563988091e-05,
+      "loss": 1.3029,
+      "step": 1646
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5430777439602875e-05,
+      "loss": 1.1746,
+      "step": 1647
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.542519684155884e-05,
+      "loss": 1.2637,
+      "step": 1648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5419613848212708e-05,
+      "loss": 1.2551,
+      "step": 1649
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5414028462029454e-05,
+      "loss": 1.1965,
+      "step": 1650
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.540844068547511e-05,
+      "loss": 1.1899,
+      "step": 1651
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5402850521016753e-05,
+      "loss": 1.0652,
+      "step": 1652
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5397257971122527e-05,
+      "loss": 1.1278,
+      "step": 1653
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.539166303826162e-05,
+      "loss": 1.2937,
+      "step": 1654
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5386065724904273e-05,
+      "loss": 1.2144,
+      "step": 1655
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5380466033521783e-05,
+      "loss": 1.2018,
+      "step": 1656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5374863966586487e-05,
+      "loss": 1.3127,
+      "step": 1657
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.536925952657179e-05,
+      "loss": 1.1903,
+      "step": 1658
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.536365271595212e-05,
+      "loss": 1.4718,
+      "step": 1659
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5358043537202978e-05,
+      "loss": 1.2227,
+      "step": 1660
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5352431992800884e-05,
+      "loss": 1.2358,
+      "step": 1661
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5346818085223423e-05,
+      "loss": 1.2976,
+      "step": 1662
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5341201816949208e-05,
+      "loss": 1.196,
+      "step": 1663
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.533558319045791e-05,
+      "loss": 1.2092,
+      "step": 1664
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.532996220823023e-05,
+      "loss": 1.1868,
+      "step": 1665
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5324338872747908e-05,
+      "loss": 1.4195,
+      "step": 1666
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5318713186493736e-05,
+      "loss": 1.1993,
+      "step": 1667
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5313085151951527e-05,
+      "loss": 1.2816,
+      "step": 1668
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5307454771606137e-05,
+      "loss": 1.2222,
+      "step": 1669
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5301822047943474e-05,
+      "loss": 1.2851,
+      "step": 1670
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.529618698345045e-05,
+      "loss": 1.3453,
+      "step": 1671
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5290549580615033e-05,
+      "loss": 1.1555,
+      "step": 1672
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5284909841926217e-05,
+      "loss": 1.1481,
+      "step": 1673
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5279267769874034e-05,
+      "loss": 1.19,
+      "step": 1674
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5273623366949525e-05,
+      "loss": 1.4323,
+      "step": 1675
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5267976635644787e-05,
+      "loss": 1.2457,
+      "step": 1676
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5262327578452927e-05,
+      "loss": 1.2881,
+      "step": 1677
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5256676197868086e-05,
+      "loss": 1.1087,
+      "step": 1678
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5251022496385433e-05,
+      "loss": 1.378,
+      "step": 1679
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5245366476501151e-05,
+      "loss": 1.0994,
+      "step": 1680
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.523970814071246e-05,
+      "loss": 1.3414,
+      "step": 1681
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5234047491517592e-05,
+      "loss": 1.2597,
+      "step": 1682
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5228384531415809e-05,
+      "loss": 1.3623,
+      "step": 1683
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5222719262907385e-05,
+      "loss": 1.2148,
+      "step": 1684
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5217051688493616e-05,
+      "loss": 1.4465,
+      "step": 1685
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5211381810676826e-05,
+      "loss": 1.1649,
+      "step": 1686
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5205709631960337e-05,
+      "loss": 1.3015,
+      "step": 1687
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5200035154848505e-05,
+      "loss": 1.3601,
+      "step": 1688
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5194358381846686e-05,
+      "loss": 1.2119,
+      "step": 1689
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5188679315461258e-05,
+      "loss": 1.3131,
+      "step": 1690
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5182997958199617e-05,
+      "loss": 1.3659,
+      "step": 1691
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5177314312570155e-05,
+      "loss": 1.3037,
+      "step": 1692
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.517162838108229e-05,
+      "loss": 1.2587,
+      "step": 1693
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5165940166246436e-05,
+      "loss": 1.2876,
+      "step": 1694
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5160249670574026e-05,
+      "loss": 1.2768,
+      "step": 1695
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.515455689657749e-05,
+      "loss": 1.2766,
+      "step": 1696
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5148861846770275e-05,
+      "loss": 1.1377,
+      "step": 1697
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5143164523666824e-05,
+      "loss": 1.1019,
+      "step": 1698
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5137464929782586e-05,
+      "loss": 1.2767,
+      "step": 1699
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5131763067634017e-05,
+      "loss": 1.3808,
+      "step": 1700
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5126058939738566e-05,
+      "loss": 1.1652,
+      "step": 1701
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5120352548614692e-05,
+      "loss": 1.2808,
+      "step": 1702
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5114643896781844e-05,
+      "loss": 1.3239,
+      "step": 1703
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.510893298676048e-05,
+      "loss": 1.289,
+      "step": 1704
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5103219821072038e-05,
+      "loss": 1.307,
+      "step": 1705
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5097504402238969e-05,
+      "loss": 1.2171,
+      "step": 1706
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5091786732784717e-05,
+      "loss": 1.3042,
+      "step": 1707
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5086066815233706e-05,
+      "loss": 1.2139,
+      "step": 1708
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5080344652111365e-05,
+      "loss": 1.2259,
+      "step": 1709
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5074620245944112e-05,
+      "loss": 1.1309,
+      "step": 1710
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5068893599259354e-05,
+      "loss": 1.2083,
+      "step": 1711
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5063164714585485e-05,
+      "loss": 1.2487,
+      "step": 1712
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5057433594451891e-05,
+      "loss": 1.0346,
+      "step": 1713
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5051700241388947e-05,
+      "loss": 1.2015,
+      "step": 1714
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5045964657928006e-05,
+      "loss": 1.1677,
+      "step": 1715
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5040226846601413e-05,
+      "loss": 1.1429,
+      "step": 1716
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5034486809942492e-05,
+      "loss": 1.3142,
+      "step": 1717
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5028744550485553e-05,
+      "loss": 1.091,
+      "step": 1718
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5023000070765886e-05,
+      "loss": 1.1781,
+      "step": 1719
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5017253373319759e-05,
+      "loss": 1.3143,
+      "step": 1720
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5011504460684418e-05,
+      "loss": 1.2259,
+      "step": 1721
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.50057533353981e-05,
+      "loss": 1.2461,
+      "step": 1722
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 1.0911,
+      "step": 1723
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4994244457030304e-05,
+      "loss": 1.2391,
+      "step": 1724
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4988486709030166e-05,
+      "loss": 1.2421,
+      "step": 1725
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4982726758541708e-05,
+      "loss": 1.1649,
+      "step": 1726
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4976964608108038e-05,
+      "loss": 1.2082,
+      "step": 1727
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4971200260273223e-05,
+      "loss": 1.2235,
+      "step": 1728
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4965433717582307e-05,
+      "loss": 1.1487,
+      "step": 1729
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4959664982581304e-05,
+      "loss": 1.3723,
+      "step": 1730
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.495389405781719e-05,
+      "loss": 1.1733,
+      "step": 1731
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4948120945837907e-05,
+      "loss": 1.2048,
+      "step": 1732
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4942345649192376e-05,
+      "loss": 1.3527,
+      "step": 1733
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4936568170430465e-05,
+      "loss": 1.3322,
+      "step": 1734
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4930788512103018e-05,
+      "loss": 1.2819,
+      "step": 1735
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4925006676761835e-05,
+      "loss": 1.4829,
+      "step": 1736
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.491922266695968e-05,
+      "loss": 1.3326,
+      "step": 1737
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4913436485250275e-05,
+      "loss": 1.2453,
+      "step": 1738
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4907648134188304e-05,
+      "loss": 1.2797,
+      "step": 1739
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4901857616329405e-05,
+      "loss": 1.1617,
+      "step": 1740
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4896064934230174e-05,
+      "loss": 1.1823,
+      "step": 1741
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4890270090448167e-05,
+      "loss": 1.3559,
+      "step": 1742
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.488447308754189e-05,
+      "loss": 1.2228,
+      "step": 1743
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4878673928070795e-05,
+      "loss": 1.1083,
+      "step": 1744
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4872872614595304e-05,
+      "loss": 1.0395,
+      "step": 1745
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4867069149676774e-05,
+      "loss": 1.4386,
+      "step": 1746
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.486126353587752e-05,
+      "loss": 1.299,
+      "step": 1747
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4855455775760804e-05,
+      "loss": 1.0643,
+      "step": 1748
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4849645871890832e-05,
+      "loss": 1.24,
+      "step": 1749
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.484383382683276e-05,
+      "loss": 1.2561,
+      "step": 1750
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4838019643152699e-05,
+      "loss": 1.479,
+      "step": 1751
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4832203323417679e-05,
+      "loss": 1.1286,
+      "step": 1752
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4826384870195702e-05,
+      "loss": 1.2997,
+      "step": 1753
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.482056428605569e-05,
+      "loss": 1.2939,
+      "step": 1754
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4814741573567514e-05,
+      "loss": 1.2442,
+      "step": 1755
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4808916735301992e-05,
+      "loss": 1.0359,
+      "step": 1756
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4803089773830868e-05,
+      "loss": 1.2784,
+      "step": 1757
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4797260691726831e-05,
+      "loss": 1.4609,
+      "step": 1758
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.47914294915635e-05,
+      "loss": 1.3909,
+      "step": 1759
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.478559617591544e-05,
+      "loss": 1.2002,
+      "step": 1760
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4779760747358141e-05,
+      "loss": 1.2577,
+      "step": 1761
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4773923208468023e-05,
+      "loss": 1.2283,
+      "step": 1762
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4768083561822451e-05,
+      "loss": 1.3137,
+      "step": 1763
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4762241809999704e-05,
+      "loss": 1.2856,
+      "step": 1764
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4756397955579012e-05,
+      "loss": 1.1662,
+      "step": 1765
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4750552001140506e-05,
+      "loss": 1.3342,
+      "step": 1766
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4744703949265268e-05,
+      "loss": 1.2858,
+      "step": 1767
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4738853802535293e-05,
+      "loss": 1.296,
+      "step": 1768
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4733001563533507e-05,
+      "loss": 1.1682,
+      "step": 1769
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4727147234843752e-05,
+      "loss": 1.0843,
+      "step": 1770
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4721290819050804e-05,
+      "loss": 1.1913,
+      "step": 1771
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4715432318740352e-05,
+      "loss": 1.2978,
+      "step": 1772
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4709571736499008e-05,
+      "loss": 1.1692,
+      "step": 1773
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.47037090749143e-05,
+      "loss": 1.2155,
+      "step": 1774
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4697844336574685e-05,
+      "loss": 1.1616,
+      "step": 1775
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4691977524069518e-05,
+      "loss": 1.2433,
+      "step": 1776
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4686108639989091e-05,
+      "loss": 1.1045,
+      "step": 1777
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4680237686924594e-05,
+      "loss": 1.2726,
+      "step": 1778
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.467436466746814e-05,
+      "loss": 1.2804,
+      "step": 1779
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4668489584212746e-05,
+      "loss": 1.1509,
+      "step": 1780
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4662612439752356e-05,
+      "loss": 1.3874,
+      "step": 1781
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4656733236681802e-05,
+      "loss": 1.1121,
+      "step": 1782
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.465085197759684e-05,
+      "loss": 1.1295,
+      "step": 1783
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4644968665094133e-05,
+      "loss": 1.2849,
+      "step": 1784
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4639083301771246e-05,
+      "loss": 1.2766,
+      "step": 1785
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4633195890226653e-05,
+      "loss": 1.3135,
+      "step": 1786
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4627306433059724e-05,
+      "loss": 1.2934,
+      "step": 1787
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4621414932870746e-05,
+      "loss": 1.2696,
+      "step": 1788
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4615521392260899e-05,
+      "loss": 1.299,
+      "step": 1789
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4609625813832263e-05,
+      "loss": 1.2035,
+      "step": 1790
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4603728200187823e-05,
+      "loss": 1.1285,
+      "step": 1791
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4597828553931454e-05,
+      "loss": 1.296,
+      "step": 1792
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4591926877667946e-05,
+      "loss": 1.2006,
+      "step": 1793
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4586023174002961e-05,
+      "loss": 1.2732,
+      "step": 1794
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4580117445543077e-05,
+      "loss": 1.3373,
+      "step": 1795
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4574209694895755e-05,
+      "loss": 1.1622,
+      "step": 1796
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4568299924669351e-05,
+      "loss": 1.2178,
+      "step": 1797
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4562388137473114e-05,
+      "loss": 1.1591,
+      "step": 1798
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4556474335917181e-05,
+      "loss": 1.2254,
+      "step": 1799
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4550558522612584e-05,
+      "loss": 1.2477,
+      "step": 1800
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4544640700171235e-05,
+      "loss": 1.2576,
+      "step": 1801
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4538720871205938e-05,
+      "loss": 1.2171,
+      "step": 1802
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4532799038330385e-05,
+      "loss": 1.2671,
+      "step": 1803
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4526875204159148e-05,
+      "loss": 1.3604,
+      "step": 1804
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4520949371307686e-05,
+      "loss": 1.3543,
+      "step": 1805
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4515021542392335e-05,
+      "loss": 1.2134,
+      "step": 1806
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4509091720030323e-05,
+      "loss": 1.1049,
+      "step": 1807
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4503159906839743e-05,
+      "loss": 1.2808,
+      "step": 1808
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4497226105439586e-05,
+      "loss": 1.0637,
+      "step": 1809
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.44912903184497e-05,
+      "loss": 1.4093,
+      "step": 1810
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4485352548490827e-05,
+      "loss": 1.1221,
+      "step": 1811
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4479412798184576e-05,
+      "loss": 1.1876,
+      "step": 1812
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4473471070153429e-05,
+      "loss": 1.2551,
+      "step": 1813
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4467527367020745e-05,
+      "loss": 1.2783,
+      "step": 1814
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4461581691410757e-05,
+      "loss": 1.1804,
+      "step": 1815
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4455634045948562e-05,
+      "loss": 1.373,
+      "step": 1816
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4449684433260134e-05,
+      "loss": 1.2696,
+      "step": 1817
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4443732855972307e-05,
+      "loss": 1.287,
+      "step": 1818
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4437779316712797e-05,
+      "loss": 1.4263,
+      "step": 1819
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4431823818110166e-05,
+      "loss": 1.2531,
+      "step": 1820
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4425866362793863e-05,
+      "loss": 1.0988,
+      "step": 1821
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4419906953394179e-05,
+      "loss": 1.1745,
+      "step": 1822
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4413945592542282e-05,
+      "loss": 1.2617,
+      "step": 1823
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4407982282870203e-05,
+      "loss": 1.3464,
+      "step": 1824
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4402017027010824e-05,
+      "loss": 1.1411,
+      "step": 1825
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4396049827597893e-05,
+      "loss": 1.2598,
+      "step": 1826
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4390080687266013e-05,
+      "loss": 1.2396,
+      "step": 1827
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4384109608650645e-05,
+      "loss": 1.1944,
+      "step": 1828
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4378136594388105e-05,
+      "loss": 1.2699,
+      "step": 1829
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4372161647115563e-05,
+      "loss": 1.1512,
+      "step": 1830
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4366184769471053e-05,
+      "loss": 1.2825,
+      "step": 1831
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4360205964093436e-05,
+      "loss": 1.2745,
+      "step": 1832
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4354225233622457e-05,
+      "loss": 1.3422,
+      "step": 1833
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.434824258069868e-05,
+      "loss": 1.1692,
+      "step": 1834
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4342258007963541e-05,
+      "loss": 1.326,
+      "step": 1835
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.433627151805931e-05,
+      "loss": 1.1509,
+      "step": 1836
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4330283113629112e-05,
+      "loss": 1.2575,
+      "step": 1837
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4324292797316908e-05,
+      "loss": 1.1634,
+      "step": 1838
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4318300571767514e-05,
+      "loss": 1.3396,
+      "step": 1839
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4312306439626583e-05,
+      "loss": 1.2634,
+      "step": 1840
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4306310403540602e-05,
+      "loss": 1.2905,
+      "step": 1841
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4300312466156917e-05,
+      "loss": 1.2857,
+      "step": 1842
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4294312630123699e-05,
+      "loss": 1.2931,
+      "step": 1843
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.428831089808996e-05,
+      "loss": 1.3712,
+      "step": 1844
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4282307272705555e-05,
+      "loss": 1.3513,
+      "step": 1845
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4276301756621162e-05,
+      "loss": 1.183,
+      "step": 1846
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4270294352488316e-05,
+      "loss": 1.1106,
+      "step": 1847
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4264285062959358e-05,
+      "loss": 1.1466,
+      "step": 1848
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4258273890687487e-05,
+      "loss": 1.2344,
+      "step": 1849
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.425226083832671e-05,
+      "loss": 1.3053,
+      "step": 1850
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4246245908531883e-05,
+      "loss": 1.2316,
+      "step": 1851
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4240229103958687e-05,
+      "loss": 1.2236,
+      "step": 1852
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.423421042726362e-05,
+      "loss": 1.1471,
+      "step": 1853
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4228189881104016e-05,
+      "loss": 1.2515,
+      "step": 1854
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4222167468138034e-05,
+      "loss": 1.1032,
+      "step": 1855
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4216143191024656e-05,
+      "loss": 1.4317,
+      "step": 1856
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4210117052423683e-05,
+      "loss": 1.3779,
+      "step": 1857
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4204089054995745e-05,
+      "loss": 1.2488,
+      "step": 1858
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4198059201402288e-05,
+      "loss": 1.2876,
+      "step": 1859
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4192027494305578e-05,
+      "loss": 1.2179,
+      "step": 1860
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4185993936368703e-05,
+      "loss": 1.2288,
+      "step": 1861
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4179958530255559e-05,
+      "loss": 1.2247,
+      "step": 1862
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4173921278630874e-05,
+      "loss": 1.3355,
+      "step": 1863
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4167882184160168e-05,
+      "loss": 1.4152,
+      "step": 1864
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.41618412495098e-05,
+      "loss": 1.429,
+      "step": 1865
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4155798477346922e-05,
+      "loss": 1.1643,
+      "step": 1866
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4149753870339509e-05,
+      "loss": 1.2737,
+      "step": 1867
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4143707431156337e-05,
+      "loss": 1.2764,
+      "step": 1868
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4137659162466999e-05,
+      "loss": 1.2087,
+      "step": 1869
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.413160906694189e-05,
+      "loss": 1.2916,
+      "step": 1870
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4125557147252215e-05,
+      "loss": 1.3883,
+      "step": 1871
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.411950340606998e-05,
+      "loss": 1.2127,
+      "step": 1872
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4113447846068007e-05,
+      "loss": 1.4096,
+      "step": 1873
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4107390469919899e-05,
+      "loss": 1.2539,
+      "step": 1874
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.410133128030009e-05,
+      "loss": 1.0244,
+      "step": 1875
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4095270279883786e-05,
+      "loss": 1.1662,
+      "step": 1876
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4089207471347013e-05,
+      "loss": 1.0866,
+      "step": 1877
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.408314285736658e-05,
+      "loss": 1.2116,
+      "step": 1878
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4077076440620112e-05,
+      "loss": 1.2614,
+      "step": 1879
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4071008223786011e-05,
+      "loss": 1.3049,
+      "step": 1880
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4064938209543483e-05,
+      "loss": 1.133,
+      "step": 1881
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4058866400572528e-05,
+      "loss": 1.3317,
+      "step": 1882
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4052792799553935e-05,
+      "loss": 1.2114,
+      "step": 1883
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4046717409169288e-05,
+      "loss": 1.2979,
+      "step": 1884
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4040640232100955e-05,
+      "loss": 1.3034,
+      "step": 1885
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4034561271032096e-05,
+      "loss": 1.1055,
+      "step": 1886
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4028480528646669e-05,
+      "loss": 1.1613,
+      "step": 1887
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4022398007629396e-05,
+      "loss": 1.393,
+      "step": 1888
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.401631371066581e-05,
+      "loss": 1.1077,
+      "step": 1889
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4010227640442202e-05,
+      "loss": 1.2558,
+      "step": 1890
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4004139799645669e-05,
+      "loss": 1.1262,
+      "step": 1891
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3998050190964075e-05,
+      "loss": 1.1516,
+      "step": 1892
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3991958817086068e-05,
+      "loss": 1.3072,
+      "step": 1893
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3985865680701084e-05,
+      "loss": 1.0818,
+      "step": 1894
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3979770784499325e-05,
+      "loss": 1.2984,
+      "step": 1895
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3973674131171775e-05,
+      "loss": 1.2379,
+      "step": 1896
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3967575723410194e-05,
+      "loss": 1.2455,
+      "step": 1897
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.396147556390712e-05,
+      "loss": 1.0673,
+      "step": 1898
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3955373655355852e-05,
+      "loss": 1.2748,
+      "step": 1899
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3949270000450476e-05,
+      "loss": 1.2563,
+      "step": 1900
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3943164601885846e-05,
+      "loss": 1.2746,
+      "step": 1901
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3937057462357578e-05,
+      "loss": 1.4008,
+      "step": 1902
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3930948584562062e-05,
+      "loss": 1.1156,
+      "step": 1903
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3924837971196456e-05,
+      "loss": 1.0892,
+      "step": 1904
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3918725624958684e-05,
+      "loss": 1.1384,
+      "step": 1905
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3912611548547435e-05,
+      "loss": 1.3494,
+      "step": 1906
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3906495744662159e-05,
+      "loss": 1.1687,
+      "step": 1907
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3900378216003073e-05,
+      "loss": 1.1702,
+      "step": 1908
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.389425896527115e-05,
+      "loss": 1.2658,
+      "step": 1909
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.388813799516813e-05,
+      "loss": 1.2672,
+      "step": 1910
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3882015308396508e-05,
+      "loss": 1.32,
+      "step": 1911
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3875890907659539e-05,
+      "loss": 1.3336,
+      "step": 1912
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3869764795661233e-05,
+      "loss": 1.17,
+      "step": 1913
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3863636975106354e-05,
+      "loss": 1.1981,
+      "step": 1914
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3857507448700425e-05,
+      "loss": 1.393,
+      "step": 1915
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3851376219149721e-05,
+      "loss": 1.112,
+      "step": 1916
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3845243289161263e-05,
+      "loss": 1.2511,
+      "step": 1917
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3839108661442832e-05,
+      "loss": 1.3086,
+      "step": 1918
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3832972338702954e-05,
+      "loss": 1.0865,
+      "step": 1919
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3826834323650899e-05,
+      "loss": 1.1762,
+      "step": 1920
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3820694618996692e-05,
+      "loss": 1.4107,
+      "step": 1921
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3814553227451099e-05,
+      "loss": 1.3201,
+      "step": 1922
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3808410151725633e-05,
+      "loss": 1.2656,
+      "step": 1923
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.380226539453255e-05,
+      "loss": 1.2625,
+      "step": 1924
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.379611895858485e-05,
+      "loss": 1.1873,
+      "step": 1925
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3789970846596268e-05,
+      "loss": 1.2438,
+      "step": 1926
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3783821061281285e-05,
+      "loss": 1.3288,
+      "step": 1927
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.377766960535512e-05,
+      "loss": 1.2936,
+      "step": 1928
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3771516481533733e-05,
+      "loss": 1.1771,
+      "step": 1929
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.376536169253381e-05,
+      "loss": 1.2232,
+      "step": 1930
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3759205241072782e-05,
+      "loss": 1.1922,
+      "step": 1931
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3753047129868808e-05,
+      "loss": 1.3563,
+      "step": 1932
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3746887361640786e-05,
+      "loss": 1.3063,
+      "step": 1933
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3740725939108336e-05,
+      "loss": 1.3191,
+      "step": 1934
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3734562864991823e-05,
+      "loss": 1.191,
+      "step": 1935
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3728398142012323e-05,
+      "loss": 1.27,
+      "step": 1936
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3722231772891657e-05,
+      "loss": 1.4293,
+      "step": 1937
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3716063760352359e-05,
+      "loss": 1.1148,
+      "step": 1938
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3709894107117698e-05,
+      "loss": 1.3147,
+      "step": 1939
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3703722815911667e-05,
+      "loss": 1.322,
+      "step": 1940
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3697549889458974e-05,
+      "loss": 1.2163,
+      "step": 1941
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3691375330485057e-05,
+      "loss": 1.2735,
+      "step": 1942
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3685199141716074e-05,
+      "loss": 1.1337,
+      "step": 1943
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3679021325878901e-05,
+      "loss": 1.1051,
+      "step": 1944
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3672841885701127e-05,
+      "loss": 1.3592,
+      "step": 1945
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3666660823911067e-05,
+      "loss": 1.2504,
+      "step": 1946
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3660478143237748e-05,
+      "loss": 1.1531,
+      "step": 1947
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3654293846410913e-05,
+      "loss": 1.0567,
+      "step": 1948
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3648107936161015e-05,
+      "loss": 1.2061,
+      "step": 1949
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3641920415219224e-05,
+      "loss": 1.2596,
+      "step": 1950
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3635731286317415e-05,
+      "loss": 1.127,
+      "step": 1951
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3629540552188188e-05,
+      "loss": 1.2367,
+      "step": 1952
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3623348215564827e-05,
+      "loss": 1.2123,
+      "step": 1953
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3617154279181348e-05,
+      "loss": 1.0858,
+      "step": 1954
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3610958745772456e-05,
+      "loss": 1.3434,
+      "step": 1955
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3604761618073567e-05,
+      "loss": 1.1642,
+      "step": 1956
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.359856289882081e-05,
+      "loss": 1.2926,
+      "step": 1957
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3592362590751001e-05,
+      "loss": 1.3659,
+      "step": 1958
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3586160696601667e-05,
+      "loss": 1.3593,
+      "step": 1959
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3579957219111031e-05,
+      "loss": 1.3118,
+      "step": 1960
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3573752161018021e-05,
+      "loss": 1.2434,
+      "step": 1961
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3567545525062256e-05,
+      "loss": 1.1639,
+      "step": 1962
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3561337313984053e-05,
+      "loss": 1.1567,
+      "step": 1963
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3555127530524432e-05,
+      "loss": 1.2477,
+      "step": 1964
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3548916177425095e-05,
+      "loss": 1.3386,
+      "step": 1965
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3542703257428454e-05,
+      "loss": 1.4531,
+      "step": 1966
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3536488773277586e-05,
+      "loss": 1.1322,
+      "step": 1967
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3530272727716286e-05,
+      "loss": 1.2674,
+      "step": 1968
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3524055123489023e-05,
+      "loss": 1.2432,
+      "step": 1969
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3517835963340961e-05,
+      "loss": 1.1722,
+      "step": 1970
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3511615250017948e-05,
+      "loss": 1.2487,
+      "step": 1971
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3505392986266516e-05,
+      "loss": 1.2716,
+      "step": 1972
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3499169174833883e-05,
+      "loss": 1.0699,
+      "step": 1973
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3492943818467956e-05,
+      "loss": 1.1245,
+      "step": 1974
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3486716919917313e-05,
+      "loss": 1.3676,
+      "step": 1975
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.348048848193122e-05,
+      "loss": 1.2978,
+      "step": 1976
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3474258507259623e-05,
+      "loss": 1.1877,
+      "step": 1977
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3468026998653147e-05,
+      "loss": 1.2105,
+      "step": 1978
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3461793958863087e-05,
+      "loss": 1.2021,
+      "step": 1979
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3455559390641426e-05,
+      "loss": 1.2573,
+      "step": 1980
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3449323296740808e-05,
+      "loss": 1.413,
+      "step": 1981
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3443085679914566e-05,
+      "loss": 1.2162,
+      "step": 1982
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3436846542916686e-05,
+      "loss": 1.3706,
+      "step": 1983
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3430605888501846e-05,
+      "loss": 1.3335,
+      "step": 1984
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3424363719425382e-05,
+      "loss": 1.1419,
+      "step": 1985
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.34181200384433e-05,
+      "loss": 1.3706,
+      "step": 1986
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3411874848312274e-05,
+      "loss": 1.1509,
+      "step": 1987
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3405628151789645e-05,
+      "loss": 1.368,
+      "step": 1988
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.339937995163342e-05,
+      "loss": 1.1817,
+      "step": 1989
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3393130250602266e-05,
+      "loss": 1.3249,
+      "step": 1990
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.338687905145552e-05,
+      "loss": 1.1407,
+      "step": 1991
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3380626356953175e-05,
+      "loss": 1.221,
+      "step": 1992
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.337437216985588e-05,
+      "loss": 1.2487,
+      "step": 1993
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.336811649292496e-05,
+      "loss": 1.2144,
+      "step": 1994
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3361859328922368e-05,
+      "loss": 1.1989,
+      "step": 1995
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3355600680610748e-05,
+      "loss": 1.2599,
+      "step": 1996
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3349340550753374e-05,
+      "loss": 1.3629,
+      "step": 1997
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3343078942114185e-05,
+      "loss": 1.3272,
+      "step": 1998
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3336815857457772e-05,
+      "loss": 1.4123,
+      "step": 1999
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3330551299549377e-05,
+      "loss": 1.2456,
+      "step": 2000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3324285271154889e-05,
+      "loss": 1.1974,
+      "step": 2001
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3318017775040852e-05,
+      "loss": 1.4648,
+      "step": 2002
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3311748813974454e-05,
+      "loss": 1.2842,
+      "step": 2003
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3305478390723532e-05,
+      "loss": 1.2756,
+      "step": 2004
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3299206508056563e-05,
+      "loss": 1.2815,
+      "step": 2005
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3292933168742685e-05,
+      "loss": 1.3172,
+      "step": 2006
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3286658375551654e-05,
+      "loss": 1.2199,
+      "step": 2007
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3280382131253888e-05,
+      "loss": 1.1619,
+      "step": 2008
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3274104438620435e-05,
+      "loss": 1.1417,
+      "step": 2009
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3267825300422991e-05,
+      "loss": 1.429,
+      "step": 2010
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3261544719433878e-05,
+      "loss": 1.2002,
+      "step": 2011
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.325526269842607e-05,
+      "loss": 1.2777,
+      "step": 2012
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3248979240173167e-05,
+      "loss": 1.2557,
+      "step": 2013
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3242694347449405e-05,
+      "loss": 1.1529,
+      "step": 2014
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3236408023029655e-05,
+      "loss": 1.2246,
+      "step": 2015
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3230120269689417e-05,
+      "loss": 1.2578,
+      "step": 2016
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3223831090204826e-05,
+      "loss": 1.1725,
+      "step": 2017
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3217540487352645e-05,
+      "loss": 0.9825,
+      "step": 2018
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3211248463910263e-05,
+      "loss": 1.3974,
+      "step": 2019
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3204955022655704e-05,
+      "loss": 1.1033,
+      "step": 2020
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3198660166367603e-05,
+      "loss": 1.2374,
+      "step": 2021
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.319236389782524e-05,
+      "loss": 1.2196,
+      "step": 2022
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3186066219808498e-05,
+      "loss": 1.2136,
+      "step": 2023
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3179767135097901e-05,
+      "loss": 1.3564,
+      "step": 2024
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3173466646474575e-05,
+      "loss": 1.0271,
+      "step": 2025
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3167164756720285e-05,
+      "loss": 1.1848,
+      "step": 2026
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3160861468617402e-05,
+      "loss": 1.3845,
+      "step": 2027
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3154556784948918e-05,
+      "loss": 1.1689,
+      "step": 2028
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3148250708498443e-05,
+      "loss": 1.1375,
+      "step": 2029
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3141943242050195e-05,
+      "loss": 1.1059,
+      "step": 2030
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3135634388389016e-05,
+      "loss": 1.1902,
+      "step": 2031
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.312932415030035e-05,
+      "loss": 1.2217,
+      "step": 2032
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3123012530570263e-05,
+      "loss": 1.3804,
+      "step": 2033
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3116699531985426e-05,
+      "loss": 1.2783,
+      "step": 2034
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.311038515733311e-05,
+      "loss": 1.0439,
+      "step": 2035
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3104069409401214e-05,
+      "loss": 1.3232,
+      "step": 2036
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3097752290978219e-05,
+      "loss": 1.3852,
+      "step": 2037
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.309143380485323e-05,
+      "loss": 1.3759,
+      "step": 2038
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3085113953815948e-05,
+      "loss": 1.2778,
+      "step": 2039
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3078792740656679e-05,
+      "loss": 1.3341,
+      "step": 2040
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3072470168166325e-05,
+      "loss": 1.1781,
+      "step": 2041
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3066146239136398e-05,
+      "loss": 1.1905,
+      "step": 2042
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3059820956358998e-05,
+      "loss": 1.45,
+      "step": 2043
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.305349432262683e-05,
+      "loss": 1.3016,
+      "step": 2044
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3047166340733196e-05,
+      "loss": 1.233,
+      "step": 2045
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3040837013471985e-05,
+      "loss": 1.227,
+      "step": 2046
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3034506343637687e-05,
+      "loss": 1.176,
+      "step": 2047
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3028174334025392e-05,
+      "loss": 1.2446,
+      "step": 2048
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3021840987430761e-05,
+      "loss": 1.3733,
+      "step": 2049
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3015506306650063e-05,
+      "loss": 1.2798,
+      "step": 2050
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3009170294480149e-05,
+      "loss": 1.2511,
+      "step": 2051
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3002832953718462e-05,
+      "loss": 1.1022,
+      "step": 2052
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2996494287163024e-05,
+      "loss": 1.445,
+      "step": 2053
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.299015429761245e-05,
+      "loss": 1.2099,
+      "step": 2054
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2983812987865936e-05,
+      "loss": 1.3478,
+      "step": 2055
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2977470360723258e-05,
+      "loss": 1.1628,
+      "step": 2056
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2971126418984783e-05,
+      "loss": 1.238,
+      "step": 2057
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2964781165451448e-05,
+      "loss": 1.1873,
+      "step": 2058
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2958434602924771e-05,
+      "loss": 1.2215,
+      "step": 2059
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2952086734206857e-05,
+      "loss": 1.2369,
+      "step": 2060
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2945737562100373e-05,
+      "loss": 1.2392,
+      "step": 2061
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2939387089408578e-05,
+      "loss": 1.4617,
+      "step": 2062
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2933035318935285e-05,
+      "loss": 0.9645,
+      "step": 2063
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2926682253484906e-05,
+      "loss": 1.2101,
+      "step": 2064
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2920327895862397e-05,
+      "loss": 1.2208,
+      "step": 2065
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.291397224887331e-05,
+      "loss": 1.0927,
+      "step": 2066
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.290761531532374e-05,
+      "loss": 1.2681,
+      "step": 2067
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2901257098020376e-05,
+      "loss": 1.2648,
+      "step": 2068
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.289489759977046e-05,
+      "loss": 1.2718,
+      "step": 2069
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2888536823381799e-05,
+      "loss": 1.1761,
+      "step": 2070
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2882174771662765e-05,
+      "loss": 1.3121,
+      "step": 2071
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2875811447422301e-05,
+      "loss": 1.3022,
+      "step": 2072
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2869446853469898e-05,
+      "loss": 1.1216,
+      "step": 2073
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2863080992615625e-05,
+      "loss": 1.2688,
+      "step": 2074
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2856713867670089e-05,
+      "loss": 1.1901,
+      "step": 2075
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.285034548144448e-05,
+      "loss": 1.2284,
+      "step": 2076
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2843975836750519e-05,
+      "loss": 1.1461,
+      "step": 2077
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2837604936400507e-05,
+      "loss": 1.1588,
+      "step": 2078
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2831232783207278e-05,
+      "loss": 1.2376,
+      "step": 2079
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2824859379984231e-05,
+      "loss": 1.081,
+      "step": 2080
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2818484729545319e-05,
+      "loss": 1.2076,
+      "step": 2081
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.281210883470504e-05,
+      "loss": 1.3346,
+      "step": 2082
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2805731698278442e-05,
+      "loss": 1.3243,
+      "step": 2083
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.279935332308112e-05,
+      "loss": 1.396,
+      "step": 2084
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2792973711929222e-05,
+      "loss": 1.2432,
+      "step": 2085
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2786592867639436e-05,
+      "loss": 1.1933,
+      "step": 2086
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2780210793028994e-05,
+      "loss": 1.2956,
+      "step": 2087
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2773827490915678e-05,
+      "loss": 1.3077,
+      "step": 2088
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.27674429641178e-05,
+      "loss": 1.1036,
+      "step": 2089
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2761057215454229e-05,
+      "loss": 1.213,
+      "step": 2090
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2754670247744353e-05,
+      "loss": 1.3735,
+      "step": 2091
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2748282063808124e-05,
+      "loss": 1.2272,
+      "step": 2092
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2741892666466006e-05,
+      "loss": 1.218,
+      "step": 2093
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2735502058539014e-05,
+      "loss": 1.0065,
+      "step": 2094
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.272911024284869e-05,
+      "loss": 1.2563,
+      "step": 2095
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2722717222217116e-05,
+      "loss": 1.2294,
+      "step": 2096
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.27163229994669e-05,
+      "loss": 1.2446,
+      "step": 2097
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2709927577421182e-05,
+      "loss": 1.4414,
+      "step": 2098
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2703530958903631e-05,
+      "loss": 1.4832,
+      "step": 2099
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2697133146738452e-05,
+      "loss": 1.1504,
+      "step": 2100
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2690734143750361e-05,
+      "loss": 1.3627,
+      "step": 2101
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2684333952764621e-05,
+      "loss": 1.4107,
+      "step": 2102
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2677932576606998e-05,
+      "loss": 1.3179,
+      "step": 2103
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2671530018103797e-05,
+      "loss": 1.0392,
+      "step": 2104
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2665126280081837e-05,
+      "loss": 1.2748,
+      "step": 2105
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2658721365368462e-05,
+      "loss": 1.2853,
+      "step": 2106
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.265231527679153e-05,
+      "loss": 1.2699,
+      "step": 2107
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2645908017179425e-05,
+      "loss": 1.3205,
+      "step": 2108
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2639499589361041e-05,
+      "loss": 1.2578,
+      "step": 2109
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2633089996165792e-05,
+      "loss": 1.4071,
+      "step": 2110
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2626679240423605e-05,
+      "loss": 1.1199,
+      "step": 2111
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2620267324964918e-05,
+      "loss": 1.4458,
+      "step": 2112
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2613854252620687e-05,
+      "loss": 1.1814,
+      "step": 2113
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2607440026222375e-05,
+      "loss": 1.2777,
+      "step": 2114
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.260102464860195e-05,
+      "loss": 1.1166,
+      "step": 2115
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.25946081225919e-05,
+      "loss": 1.114,
+      "step": 2116
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2588190451025209e-05,
+      "loss": 1.183,
+      "step": 2117
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.258177163673537e-05,
+      "loss": 1.2601,
+      "step": 2118
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2575351682556387e-05,
+      "loss": 1.3225,
+      "step": 2119
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2568930591322757e-05,
+      "loss": 1.2029,
+      "step": 2120
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2562508365869484e-05,
+      "loss": 1.3724,
+      "step": 2121
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2556085009032072e-05,
+      "loss": 1.0388,
+      "step": 2122
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2549660523646527e-05,
+      "loss": 1.3392,
+      "step": 2123
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2543234912549352e-05,
+      "loss": 1.3165,
+      "step": 2124
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2536808178577542e-05,
+      "loss": 1.1851,
+      "step": 2125
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2530380324568599e-05,
+      "loss": 1.2053,
+      "step": 2126
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2523951353360504e-05,
+      "loss": 1.2169,
+      "step": 2127
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2517521267791746e-05,
+      "loss": 1.1467,
+      "step": 2128
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2511090070701297e-05,
+      "loss": 1.3531,
+      "step": 2129
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2504657764928626e-05,
+      "loss": 1.3994,
+      "step": 2130
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2498224353313684e-05,
+      "loss": 1.2091,
+      "step": 2131
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2491789838696918e-05,
+      "loss": 1.0596,
+      "step": 2132
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2485354223919255e-05,
+      "loss": 1.3391,
+      "step": 2133
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2478917511822113e-05,
+      "loss": 1.116,
+      "step": 2134
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2472479705247393e-05,
+      "loss": 1.1184,
+      "step": 2135
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2466040807037478e-05,
+      "loss": 1.2559,
+      "step": 2136
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2459600820035234e-05,
+      "loss": 1.2321,
+      "step": 2137
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2453159747084007e-05,
+      "loss": 1.2475,
+      "step": 2138
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2446717591027624e-05,
+      "loss": 1.4016,
+      "step": 2139
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2440274354710384e-05,
+      "loss": 1.091,
+      "step": 2140
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2433830040977077e-05,
+      "loss": 1.3957,
+      "step": 2141
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2427384652672954e-05,
+      "loss": 1.206,
+      "step": 2142
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2420938192643746e-05,
+      "loss": 1.2006,
+      "step": 2143
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2414490663735662e-05,
+      "loss": 1.3424,
+      "step": 2144
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2408042068795373e-05,
+      "loss": 1.3114,
+      "step": 2145
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2401592410670026e-05,
+      "loss": 1.4059,
+      "step": 2146
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2395141692207244e-05,
+      "loss": 1.2019,
+      "step": 2147
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2388689916255106e-05,
+      "loss": 1.116,
+      "step": 2148
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2382237085662166e-05,
+      "loss": 1.2229,
+      "step": 2149
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2375783203277436e-05,
+      "loss": 1.0737,
+      "step": 2150
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2369328271950404e-05,
+      "loss": 1.3282,
+      "step": 2151
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.236287229453101e-05,
+      "loss": 1.2165,
+      "step": 2152
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2356415273869664e-05,
+      "loss": 1.24,
+      "step": 2153
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2349957212817229e-05,
+      "loss": 1.3263,
+      "step": 2154
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2343498114225038e-05,
+      "loss": 1.2732,
+      "step": 2155
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2337037980944869e-05,
+      "loss": 1.2716,
+      "step": 2156
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2330576815828964e-05,
+      "loss": 1.028,
+      "step": 2157
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2324114621730026e-05,
+      "loss": 1.2768,
+      "step": 2158
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2317651401501201e-05,
+      "loss": 1.2024,
+      "step": 2159
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2311187157996091e-05,
+      "loss": 1.2481,
+      "step": 2160
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2304721894068758e-05,
+      "loss": 1.3137,
+      "step": 2161
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2298255612573706e-05,
+      "loss": 1.2627,
+      "step": 2162
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.229178831636589e-05,
+      "loss": 1.0627,
+      "step": 2163
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2285320008300712e-05,
+      "loss": 1.192,
+      "step": 2164
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.227885069123403e-05,
+      "loss": 1.2963,
+      "step": 2165
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2272380368022133e-05,
+      "loss": 1.0288,
+      "step": 2166
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2265909041521764e-05,
+      "loss": 1.3284,
+      "step": 2167
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2259436714590105e-05,
+      "loss": 1.252,
+      "step": 2168
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2252963390084784e-05,
+      "loss": 1.1595,
+      "step": 2169
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2246489070863863e-05,
+      "loss": 1.3244,
+      "step": 2170
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2240013759785849e-05,
+      "loss": 1.2442,
+      "step": 2171
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2233537459709683e-05,
+      "loss": 1.1341,
+      "step": 2172
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2227060173494747e-05,
+      "loss": 1.1026,
+      "step": 2173
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2220581904000852e-05,
+      "loss": 1.1104,
+      "step": 2174
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2214102654088248e-05,
+      "loss": 1.1967,
+      "step": 2175
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2207622426617619e-05,
+      "loss": 1.2443,
+      "step": 2176
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2201141224450073e-05,
+      "loss": 1.0731,
+      "step": 2177
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2194659050447158e-05,
+      "loss": 1.3355,
+      "step": 2178
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2188175907470847e-05,
+      "loss": 1.0327,
+      "step": 2179
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2181691798383536e-05,
+      "loss": 1.2014,
+      "step": 2180
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2175206726048058e-05,
+      "loss": 1.2214,
+      "step": 2181
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.216872069332766e-05,
+      "loss": 1.288,
+      "step": 2182
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2162233703086024e-05,
+      "loss": 1.0964,
+      "step": 2183
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.215574575818724e-05,
+      "loss": 1.3113,
+      "step": 2184
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2149256861495837e-05,
+      "loss": 1.3686,
+      "step": 2185
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2142767015876755e-05,
+      "loss": 1.3298,
+      "step": 2186
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2136276224195349e-05,
+      "loss": 1.2394,
+      "step": 2187
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.21297844893174e-05,
+      "loss": 1.2333,
+      "step": 2188
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2123291814109102e-05,
+      "loss": 1.2302,
+      "step": 2189
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2116798201437061e-05,
+      "loss": 1.4421,
+      "step": 2190
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2110303654168305e-05,
+      "loss": 1.177,
+      "step": 2191
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2103808175170264e-05,
+      "loss": 1.1863,
+      "step": 2192
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2097311767310791e-05,
+      "loss": 1.3977,
+      "step": 2193
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2090814433458136e-05,
+      "loss": 1.1856,
+      "step": 2194
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2084316176480972e-05,
+      "loss": 1.2215,
+      "step": 2195
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2077816999248366e-05,
+      "loss": 1.0202,
+      "step": 2196
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2071316904629807e-05,
+      "loss": 1.2779,
+      "step": 2197
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2064815895495168e-05,
+      "loss": 1.3145,
+      "step": 2198
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2058313974714746e-05,
+      "loss": 1.3263,
+      "step": 2199
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2051811145159232e-05,
+      "loss": 1.2205,
+      "step": 2200
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2045307409699713e-05,
+      "loss": 1.0167,
+      "step": 2201
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2038802771207686e-05,
+      "loss": 1.3203,
+      "step": 2202
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2032297232555039e-05,
+      "loss": 1.2396,
+      "step": 2203
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2025790796614064e-05,
+      "loss": 1.1123,
+      "step": 2204
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2019283466257441e-05,
+      "loss": 1.2535,
+      "step": 2205
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2012775244358253e-05,
+      "loss": 1.2618,
+      "step": 2206
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2006266133789975e-05,
+      "loss": 1.2914,
+      "step": 2207
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1999756137426465e-05,
+      "loss": 1.2802,
+      "step": 2208
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.199324525814199e-05,
+      "loss": 1.1718,
+      "step": 2209
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1986733498811186e-05,
+      "loss": 1.2291,
+      "step": 2210
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1980220862309097e-05,
+      "loss": 1.2041,
+      "step": 2211
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1973707351511137e-05,
+      "loss": 1.3499,
+      "step": 2212
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.196719296929312e-05,
+      "loss": 1.1717,
+      "step": 2213
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.196067771853124e-05,
+      "loss": 1.1849,
+      "step": 2214
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1954161602102068e-05,
+      "loss": 1.2836,
+      "step": 2215
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1947644622882569e-05,
+      "loss": 1.2146,
+      "step": 2216
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1941126783750075e-05,
+      "loss": 1.1262,
+      "step": 2217
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1934608087582312e-05,
+      "loss": 1.2322,
+      "step": 2218
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1928088537257376e-05,
+      "loss": 1.2848,
+      "step": 2219
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1921568135653738e-05,
+      "loss": 1.0831,
+      "step": 2220
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1915046885650256e-05,
+      "loss": 1.3375,
+      "step": 2221
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1908524790126149e-05,
+      "loss": 1.291,
+      "step": 2222
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1902001851961019e-05,
+      "loss": 1.304,
+      "step": 2223
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1895478074034831e-05,
+      "loss": 1.2657,
+      "step": 2224
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1888953459227937e-05,
+      "loss": 1.2709,
+      "step": 2225
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1882428010421036e-05,
+      "loss": 1.1547,
+      "step": 2226
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1875901730495215e-05,
+      "loss": 1.3101,
+      "step": 2227
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1869374622331919e-05,
+      "loss": 1.1227,
+      "step": 2228
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1862846688812956e-05,
+      "loss": 1.2205,
+      "step": 2229
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1856317932820507e-05,
+      "loss": 1.2206,
+      "step": 2230
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1849788357237108e-05,
+      "loss": 1.3613,
+      "step": 2231
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.184325796494566e-05,
+      "loss": 1.2467,
+      "step": 2232
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1836726758829427e-05,
+      "loss": 1.0949,
+      "step": 2233
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1830194741772025e-05,
+      "loss": 1.2728,
+      "step": 2234
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1823661916657441e-05,
+      "loss": 1.0544,
+      "step": 2235
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1817128286370004e-05,
+      "loss": 1.1554,
+      "step": 2236
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1810593853794409e-05,
+      "loss": 1.2098,
+      "step": 2237
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1804058621815699e-05,
+      "loss": 1.1831,
+      "step": 2238
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1797522593319277e-05,
+      "loss": 1.2132,
+      "step": 2239
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1790985771190886e-05,
+      "loss": 1.217,
+      "step": 2240
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1784448158316633e-05,
+      "loss": 1.3076,
+      "step": 2241
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1777909757582965e-05,
+      "loss": 1.1576,
+      "step": 2242
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1771370571876681e-05,
+      "loss": 1.4007,
+      "step": 2243
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1764830604084923e-05,
+      "loss": 1.3241,
+      "step": 2244
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1758289857095182e-05,
+      "loss": 1.1249,
+      "step": 2245
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1751748333795292e-05,
+      "loss": 1.3003,
+      "step": 2246
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1745206037073426e-05,
+      "loss": 1.1625,
+      "step": 2247
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1738662969818102e-05,
+      "loss": 1.1653,
+      "step": 2248
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1732119134918185e-05,
+      "loss": 1.2037,
+      "step": 2249
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1725574535262863e-05,
+      "loss": 1.1888,
+      "step": 2250
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1719029173741675e-05,
+      "loss": 1.2619,
+      "step": 2251
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.171248305324449e-05,
+      "loss": 1.3092,
+      "step": 2252
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1705936176661516e-05,
+      "loss": 1.2907,
+      "step": 2253
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1699388546883291e-05,
+      "loss": 1.131,
+      "step": 2254
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1692840166800688e-05,
+      "loss": 1.2642,
+      "step": 2255
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.168629103930491e-05,
+      "loss": 1.3476,
+      "step": 2256
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.167974116728749e-05,
+      "loss": 1.2024,
+      "step": 2257
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1673190553640295e-05,
+      "loss": 1.389,
+      "step": 2258
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1666639201255507e-05,
+      "loss": 1.3759,
+      "step": 2259
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1660087113025647e-05,
+      "loss": 1.3557,
+      "step": 2260
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1653534291843556e-05,
+      "loss": 1.1563,
+      "step": 2261
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1646980740602394e-05,
+      "loss": 1.1259,
+      "step": 2262
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1640426462195654e-05,
+      "loss": 1.2129,
+      "step": 2263
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1633871459517139e-05,
+      "loss": 1.2878,
+      "step": 2264
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.162731573546098e-05,
+      "loss": 1.2074,
+      "step": 2265
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1620759292921617e-05,
+      "loss": 1.2598,
+      "step": 2266
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1614202134793823e-05,
+      "loss": 1.286,
+      "step": 2267
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1607644263972669e-05,
+      "loss": 1.1251,
+      "step": 2268
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.160108568335355e-05,
+      "loss": 1.1081,
+      "step": 2269
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1594526395832176e-05,
+      "loss": 1.2744,
+      "step": 2270
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1587966404304565e-05,
+      "loss": 1.2267,
+      "step": 2271
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.158140571166705e-05,
+      "loss": 1.3703,
+      "step": 2272
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1574844320816263e-05,
+      "loss": 1.1247,
+      "step": 2273
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.156828223464916e-05,
+      "loss": 1.2533,
+      "step": 2274
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1561719456062995e-05,
+      "loss": 1.1714,
+      "step": 2275
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1555155987955322e-05,
+      "loss": 1.3492,
+      "step": 2276
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1548591833224015e-05,
+      "loss": 1.2375,
+      "step": 2277
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1542026994767235e-05,
+      "loss": 1.2142,
+      "step": 2278
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1535461475483461e-05,
+      "loss": 1.4415,
+      "step": 2279
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.152889527827145e-05,
+      "loss": 1.2881,
+      "step": 2280
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1522328406030285e-05,
+      "loss": 1.2793,
+      "step": 2281
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.151576086165933e-05,
+      "loss": 1.2582,
+      "step": 2282
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.150919264805825e-05,
+      "loss": 1.1379,
+      "step": 2283
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1502623768127004e-05,
+      "loss": 1.3455,
+      "step": 2284
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1496054224765846e-05,
+      "loss": 1.0036,
+      "step": 2285
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1489484020875325e-05,
+      "loss": 1.1449,
+      "step": 2286
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1482913159356281e-05,
+      "loss": 1.2003,
+      "step": 2287
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1476341643109842e-05,
+      "loss": 1.2074,
+      "step": 2288
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1469769475037427e-05,
+      "loss": 1.377,
+      "step": 2289
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1463196658040738e-05,
+      "loss": 1.2055,
+      "step": 2290
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1456623195021778e-05,
+      "loss": 1.3042,
+      "step": 2291
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1450049088882816e-05,
+      "loss": 1.2639,
+      "step": 2292
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1443474342526418e-05,
+      "loss": 1.2024,
+      "step": 2293
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1436898958855427e-05,
+      "loss": 1.1424,
+      "step": 2294
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.143032294077297e-05,
+      "loss": 1.1471,
+      "step": 2295
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1423746291182454e-05,
+      "loss": 1.2505,
+      "step": 2296
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1417169012987561e-05,
+      "loss": 1.2408,
+      "step": 2297
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1410591109092257e-05,
+      "loss": 1.2643,
+      "step": 2298
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1404012582400778e-05,
+      "loss": 1.2523,
+      "step": 2299
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1397433435817639e-05,
+      "loss": 1.3216,
+      "step": 2300
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1390853672247629e-05,
+      "loss": 1.3262,
+      "step": 2301
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1384273294595802e-05,
+      "loss": 1.2791,
+      "step": 2302
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1377692305767496e-05,
+      "loss": 1.3432,
+      "step": 2303
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1371110708668307e-05,
+      "loss": 1.1225,
+      "step": 2304
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1364528506204109e-05,
+      "loss": 1.2744,
+      "step": 2305
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1357945701281035e-05,
+      "loss": 1.1904,
+      "step": 2306
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1351362296805487e-05,
+      "loss": 1.3942,
+      "step": 2307
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1344778295684135e-05,
+      "loss": 1.2665,
+      "step": 2308
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.133819370082391e-05,
+      "loss": 1.2774,
+      "step": 2309
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1331608515132001e-05,
+      "loss": 1.2682,
+      "step": 2310
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1325022741515865e-05,
+      "loss": 1.2752,
+      "step": 2311
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1318436382883216e-05,
+      "loss": 1.2256,
+      "step": 2312
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1311849442142025e-05,
+      "loss": 1.2237,
+      "step": 2313
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.130526192220052e-05,
+      "loss": 1.1945,
+      "step": 2314
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1298673825967184e-05,
+      "loss": 1.0702,
+      "step": 2315
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1292085156350756e-05,
+      "loss": 1.2967,
+      "step": 2316
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.128549591626023e-05,
+      "loss": 1.2695,
+      "step": 2317
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1278906108604848e-05,
+      "loss": 1.2144,
+      "step": 2318
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1272315736294108e-05,
+      "loss": 1.1619,
+      "step": 2319
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1265724802237749e-05,
+      "loss": 1.2376,
+      "step": 2320
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1259133309345769e-05,
+      "loss": 1.4775,
+      "step": 2321
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1252541260528398e-05,
+      "loss": 1.1492,
+      "step": 2322
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1245948658696127e-05,
+      "loss": 1.2468,
+      "step": 2323
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1239355506759681e-05,
+      "loss": 1.0376,
+      "step": 2324
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1232761807630032e-05,
+      "loss": 1.3717,
+      "step": 2325
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1226167564218393e-05,
+      "loss": 1.1773,
+      "step": 2326
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1219572779436215e-05,
+      "loss": 1.3049,
+      "step": 2327
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.121297745619519e-05,
+      "loss": 1.2702,
+      "step": 2328
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1206381597407248e-05,
+      "loss": 1.1568,
+      "step": 2329
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1199785205984558e-05,
+      "loss": 1.2109,
+      "step": 2330
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1193188284839518e-05,
+      "loss": 1.1937,
+      "step": 2331
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1186590836884766e-05,
+      "loss": 1.291,
+      "step": 2332
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1179992865033164e-05,
+      "loss": 1.3505,
+      "step": 2333
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.117339437219782e-05,
+      "loss": 1.3193,
+      "step": 2334
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1166795361292055e-05,
+      "loss": 1.3567,
+      "step": 2335
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.116019583522943e-05,
+      "loss": 1.191,
+      "step": 2336
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.115359579692373e-05,
+      "loss": 1.1745,
+      "step": 2337
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1146995249288967e-05,
+      "loss": 1.2901,
+      "step": 2338
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1140394195239376e-05,
+      "loss": 1.2385,
+      "step": 2339
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.113379263768942e-05,
+      "loss": 1.2359,
+      "step": 2340
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1127190579553775e-05,
+      "loss": 1.1997,
+      "step": 2341
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1120588023747352e-05,
+      "loss": 1.3522,
+      "step": 2342
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1113984973185268e-05,
+      "loss": 1.1233,
+      "step": 2343
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1107381430782869e-05,
+      "loss": 1.1564,
+      "step": 2344
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1100777399455707e-05,
+      "loss": 1.0432,
+      "step": 2345
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1094172882119562e-05,
+      "loss": 1.3031,
+      "step": 2346
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1087567881690422e-05,
+      "loss": 1.2435,
+      "step": 2347
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1080962401084487e-05,
+      "loss": 1.1431,
+      "step": 2348
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1074356443218175e-05,
+      "loss": 1.2281,
+      "step": 2349
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1067750011008107e-05,
+      "loss": 1.2744,
+      "step": 2350
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1061143107371122e-05,
+      "loss": 1.0337,
+      "step": 2351
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1054535735224257e-05,
+      "loss": 1.283,
+      "step": 2352
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1047927897484764e-05,
+      "loss": 1.2586,
+      "step": 2353
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1041319597070102e-05,
+      "loss": 1.1102,
+      "step": 2354
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1034710836897922e-05,
+      "loss": 1.2006,
+      "step": 2355
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1028101619886095e-05,
+      "loss": 1.1963,
+      "step": 2356
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1021491948952676e-05,
+      "loss": 0.9484,
+      "step": 2357
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1014881827015939e-05,
+      "loss": 1.2487,
+      "step": 2358
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1008271256994338e-05,
+      "loss": 1.22,
+      "step": 2359
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1001660241806542e-05,
+      "loss": 1.1543,
+      "step": 2360
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0995048784371405e-05,
+      "loss": 1.2262,
+      "step": 2361
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0988436887607984e-05,
+      "loss": 1.2488,
+      "step": 2362
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0981824554435518e-05,
+      "loss": 1.1212,
+      "step": 2363
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0975211787773455e-05,
+      "loss": 1.286,
+      "step": 2364
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.096859859054142e-05,
+      "loss": 1.3362,
+      "step": 2365
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.096198496565924e-05,
+      "loss": 1.1689,
+      "step": 2366
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.095537091604692e-05,
+      "loss": 1.068,
+      "step": 2367
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0948756444624664e-05,
+      "loss": 1.3726,
+      "step": 2368
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0942141554312847e-05,
+      "loss": 1.3178,
+      "step": 2369
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0935526248032047e-05,
+      "loss": 1.3366,
+      "step": 2370
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0928910528703007e-05,
+      "loss": 1.2786,
+      "step": 2371
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0922294399246671e-05,
+      "loss": 1.0145,
+      "step": 2372
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0915677862584149e-05,
+      "loss": 1.265,
+      "step": 2373
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0909060921636739e-05,
+      "loss": 1.3197,
+      "step": 2374
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0902443579325914e-05,
+      "loss": 1.212,
+      "step": 2375
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0895825838573326e-05,
+      "loss": 1.4369,
+      "step": 2376
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0889207702300803e-05,
+      "loss": 1.2526,
+      "step": 2377
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0882589173430346e-05,
+      "loss": 1.4359,
+      "step": 2378
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.087597025488413e-05,
+      "loss": 1.2647,
+      "step": 2379
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0869350949584505e-05,
+      "loss": 1.2662,
+      "step": 2380
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0862731260453985e-05,
+      "loss": 1.2054,
+      "step": 2381
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0856111190415262e-05,
+      "loss": 1.2654,
+      "step": 2382
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0849490742391184e-05,
+      "loss": 1.2452,
+      "step": 2383
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0842869919304784e-05,
+      "loss": 1.1724,
+      "step": 2384
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0836248724079241e-05,
+      "loss": 1.2844,
+      "step": 2385
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0829627159637914e-05,
+      "loss": 1.2352,
+      "step": 2386
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0823005228904315e-05,
+      "loss": 1.2216,
+      "step": 2387
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0816382934802123e-05,
+      "loss": 1.1742,
+      "step": 2388
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0809760280255176e-05,
+      "loss": 1.1802,
+      "step": 2389
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.080313726818747e-05,
+      "loss": 1.3033,
+      "step": 2390
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0796513901523156e-05,
+      "loss": 1.2479,
+      "step": 2391
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.078989018318655e-05,
+      "loss": 1.2241,
+      "step": 2392
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0783266116102116e-05,
+      "loss": 1.1109,
+      "step": 2393
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.077664170319448e-05,
+      "loss": 1.321,
+      "step": 2394
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0770016947388407e-05,
+      "loss": 1.17,
+      "step": 2395
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0763391851608827e-05,
+      "loss": 1.2589,
+      "step": 2396
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0756766418780811e-05,
+      "loss": 1.4106,
+      "step": 2397
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0750140651829588e-05,
+      "loss": 1.1022,
+      "step": 2398
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0743514553680521e-05,
+      "loss": 1.2832,
+      "step": 2399
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0736888127259132e-05,
+      "loss": 1.2928,
+      "step": 2400
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0730261375491087e-05,
+      "loss": 1.2382,
+      "step": 2401
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0723634301302186e-05,
+      "loss": 1.3611,
+      "step": 2402
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0717006907618377e-05,
+      "loss": 1.2629,
+      "step": 2403
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0710379197365752e-05,
+      "loss": 1.3241,
+      "step": 2404
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.070375117347054e-05,
+      "loss": 1.2401,
+      "step": 2405
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0697122838859106e-05,
+      "loss": 1.1714,
+      "step": 2406
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0690494196457954e-05,
+      "loss": 1.379,
+      "step": 2407
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0683865249193732e-05,
+      "loss": 1.3516,
+      "step": 2408
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0677235999993205e-05,
+      "loss": 1.138,
+      "step": 2409
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.067060645178329e-05,
+      "loss": 1.3604,
+      "step": 2410
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.066397660749102e-05,
+      "loss": 1.342,
+      "step": 2411
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0657346470043575e-05,
+      "loss": 1.2746,
+      "step": 2412
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0650716042368242e-05,
+      "loss": 1.2123,
+      "step": 2413
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0644085327392465e-05,
+      "loss": 1.1604,
+      "step": 2414
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0637454328043792e-05,
+      "loss": 1.2107,
+      "step": 2415
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0630823047249903e-05,
+      "loss": 1.3038,
+      "step": 2416
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0624191487938607e-05,
+      "loss": 1.3382,
+      "step": 2417
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0617559653037833e-05,
+      "loss": 1.3196,
+      "step": 2418
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0610927545475624e-05,
+      "loss": 1.2409,
+      "step": 2419
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.060429516818016e-05,
+      "loss": 1.2419,
+      "step": 2420
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0597662524079721e-05,
+      "loss": 1.1576,
+      "step": 2421
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0591029616102725e-05,
+      "loss": 1.352,
+      "step": 2422
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0584396447177682e-05,
+      "loss": 1.3195,
+      "step": 2423
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.057776302023325e-05,
+      "loss": 1.2727,
+      "step": 2424
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0571129338198162e-05,
+      "loss": 1.1232,
+      "step": 2425
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0564495404001299e-05,
+      "loss": 1.1632,
+      "step": 2426
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0557861220571626e-05,
+      "loss": 1.2776,
+      "step": 2427
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.055122679083824e-05,
+      "loss": 1.1132,
+      "step": 2428
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.054459211773033e-05,
+      "loss": 1.3757,
+      "step": 2429
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0537957204177204e-05,
+      "loss": 1.2746,
+      "step": 2430
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0531322053108268e-05,
+      "loss": 1.1269,
+      "step": 2431
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0524686667453038e-05,
+      "loss": 1.4041,
+      "step": 2432
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.051805105014113e-05,
+      "loss": 1.0861,
+      "step": 2433
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0511415204102264e-05,
+      "loss": 1.3504,
+      "step": 2434
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.050477913226626e-05,
+      "loss": 1.3396,
+      "step": 2435
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0498142837563048e-05,
+      "loss": 1.114,
+      "step": 2436
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0491506322922633e-05,
+      "loss": 1.2471,
+      "step": 2437
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0484869591275144e-05,
+      "loss": 1.1791,
+      "step": 2438
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0478232645550784e-05,
+      "loss": 1.1025,
+      "step": 2439
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0471595488679865e-05,
+      "loss": 1.2797,
+      "step": 2440
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0464958123592779e-05,
+      "loss": 1.249,
+      "step": 2441
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.045832055322003e-05,
+      "loss": 1.2701,
+      "step": 2442
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.045168278049219e-05,
+      "loss": 1.2873,
+      "step": 2443
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0445044808339935e-05,
+      "loss": 1.3274,
+      "step": 2444
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0438406639694025e-05,
+      "loss": 1.2452,
+      "step": 2445
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0431768277485306e-05,
+      "loss": 1.2234,
+      "step": 2446
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.042512972464471e-05,
+      "loss": 1.3913,
+      "step": 2447
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.041849098410325e-05,
+      "loss": 1.2435,
+      "step": 2448
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.041185205879203e-05,
+      "loss": 1.1515,
+      "step": 2449
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0405212951642229e-05,
+      "loss": 1.2448,
+      "step": 2450
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0398573665585105e-05,
+      "loss": 1.3387,
+      "step": 2451
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0391934203552003e-05,
+      "loss": 1.3005,
+      "step": 2452
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0385294568474335e-05,
+      "loss": 1.2441,
+      "step": 2453
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0378654763283602e-05,
+      "loss": 1.2033,
+      "step": 2454
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0372014790911365e-05,
+      "loss": 1.2559,
+      "step": 2455
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0365374654289272e-05,
+      "loss": 1.1973,
+      "step": 2456
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0358734356349035e-05,
+      "loss": 1.1586,
+      "step": 2457
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0352093900022447e-05,
+      "loss": 1.1585,
+      "step": 2458
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0345453288241356e-05,
+      "loss": 1.3484,
+      "step": 2459
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0338812523937694e-05,
+      "loss": 1.2555,
+      "step": 2460
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.033217161004345e-05,
+      "loss": 1.3045,
+      "step": 2461
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0325530549490685e-05,
+      "loss": 0.959,
+      "step": 2462
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0318889345211519e-05,
+      "loss": 1.15,
+      "step": 2463
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0312248000138143e-05,
+      "loss": 1.3336,
+      "step": 2464
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.03056065172028e-05,
+      "loss": 1.0718,
+      "step": 2465
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0298964899337807e-05,
+      "loss": 1.34,
+      "step": 2466
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0292323149475527e-05,
+      "loss": 0.965,
+      "step": 2467
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0285681270548393e-05,
+      "loss": 1.0955,
+      "step": 2468
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0279039265488885e-05,
+      "loss": 1.2313,
+      "step": 2469
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0272397137229548e-05,
+      "loss": 1.0952,
+      "step": 2470
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0265754888702972e-05,
+      "loss": 1.1289,
+      "step": 2471
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0259112522841808e-05,
+      "loss": 1.317,
+      "step": 2472
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0252470042578756e-05,
+      "loss": 1.164,
+      "step": 2473
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0245827450846564e-05,
+      "loss": 1.1607,
+      "step": 2474
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.023918475057803e-05,
+      "loss": 1.4302,
+      "step": 2475
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0232541944706007e-05,
+      "loss": 1.2126,
+      "step": 2476
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0225899036163383e-05,
+      "loss": 1.2462,
+      "step": 2477
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0219256027883102e-05,
+      "loss": 1.1299,
+      "step": 2478
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0212612922798143e-05,
+      "loss": 1.4365,
+      "step": 2479
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0205969723841536e-05,
+      "loss": 1.2411,
+      "step": 2480
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0199326433946345e-05,
+      "loss": 1.0644,
+      "step": 2481
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0192683056045678e-05,
+      "loss": 1.2497,
+      "step": 2482
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0186039593072685e-05,
+      "loss": 1.2059,
+      "step": 2483
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0179396047960546e-05,
+      "loss": 1.2325,
+      "step": 2484
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0172752423642484e-05,
+      "loss": 1.1252,
+      "step": 2485
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0166108723051753e-05,
+      "loss": 0.9204,
+      "step": 2486
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0159464949121642e-05,
+      "loss": 1.2303,
+      "step": 2487
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0152821104785473e-05,
+      "loss": 1.1447,
+      "step": 2488
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0146177192976599e-05,
+      "loss": 1.3783,
+      "step": 2489
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0139533216628399e-05,
+      "loss": 1.2714,
+      "step": 2490
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0132889178674283e-05,
+      "loss": 1.3612,
+      "step": 2491
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0126245082047699e-05,
+      "loss": 1.1866,
+      "step": 2492
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0119600929682096e-05,
+      "loss": 1.1693,
+      "step": 2493
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0112956724510974e-05,
+      "loss": 1.0463,
+      "step": 2494
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0106312469467841e-05,
+      "loss": 1.1381,
+      "step": 2495
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0099668167486227e-05,
+      "loss": 1.2538,
+      "step": 2496
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0093023821499692e-05,
+      "loss": 1.3574,
+      "step": 2497
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.008637943444181e-05,
+      "loss": 1.1931,
+      "step": 2498
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0079735009246168e-05,
+      "loss": 1.2232,
+      "step": 2499
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0073090548846378e-05,
+      "loss": 1.2774,
+      "step": 2500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0066446056176065e-05,
+      "loss": 1.211,
+      "step": 2501
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0059801534168868e-05,
+      "loss": 1.3409,
+      "step": 2502
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0053156985758435e-05,
+      "loss": 1.265,
+      "step": 2503
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.004651241387843e-05,
+      "loss": 1.4048,
+      "step": 2504
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0039867821462529e-05,
+      "loss": 1.2547,
+      "step": 2505
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0033223211444417e-05,
+      "loss": 1.3338,
+      "step": 2506
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0026578586757778e-05,
+      "loss": 1.2049,
+      "step": 2507
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0019933950336314e-05,
+      "loss": 1.2922,
+      "step": 2508
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0013289305113725e-05,
+      "loss": 1.0741,
+      "step": 2509
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0006644654023719e-05,
+      "loss": 1.244,
+      "step": 2510
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1e-05,
+      "loss": 1.1634,
+      "step": 2511
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.993355345976284e-06,
+      "loss": 1.2073,
+      "step": 2512
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.986710694886276e-06,
+      "loss": 1.2083,
+      "step": 2513
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.98006604966369e-06,
+      "loss": 1.0704,
+      "step": 2514
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.973421413242224e-06,
+      "loss": 1.2341,
+      "step": 2515
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.966776788555588e-06,
+      "loss": 1.2602,
+      "step": 2516
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.960132178537471e-06,
+      "loss": 1.2055,
+      "step": 2517
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.953487586121571e-06,
+      "loss": 1.2988,
+      "step": 2518
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.946843014241572e-06,
+      "loss": 1.3952,
+      "step": 2519
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.940198465831137e-06,
+      "loss": 1.0866,
+      "step": 2520
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.933553943823938e-06,
+      "loss": 1.2323,
+      "step": 2521
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.926909451153625e-06,
+      "loss": 1.4708,
+      "step": 2522
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.920264990753837e-06,
+      "loss": 1.2124,
+      "step": 2523
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.913620565558194e-06,
+      "loss": 1.2431,
+      "step": 2524
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.90697617850031e-06,
+      "loss": 1.332,
+      "step": 2525
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.900331832513775e-06,
+      "loss": 1.292,
+      "step": 2526
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.89368753053216e-06,
+      "loss": 1.1333,
+      "step": 2527
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.88704327548903e-06,
+      "loss": 1.3979,
+      "step": 2528
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.880399070317907e-06,
+      "loss": 1.3593,
+      "step": 2529
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.873754917952307e-06,
+      "loss": 1.2046,
+      "step": 2530
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.867110821325717e-06,
+      "loss": 1.1478,
+      "step": 2531
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.860466783371605e-06,
+      "loss": 1.246,
+      "step": 2532
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.853822807023408e-06,
+      "loss": 1.3474,
+      "step": 2533
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.84717889521453e-06,
+      "loss": 1.3129,
+      "step": 2534
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.84053505087836e-06,
+      "loss": 1.2997,
+      "step": 2535
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.83389127694825e-06,
+      "loss": 1.2607,
+      "step": 2536
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.827247576357521e-06,
+      "loss": 1.3741,
+      "step": 2537
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.820603952039456e-06,
+      "loss": 1.0747,
+      "step": 2538
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.813960406927318e-06,
+      "loss": 1.3188,
+      "step": 2539
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.807316943954325e-06,
+      "loss": 1.278,
+      "step": 2540
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.800673566053657e-06,
+      "loss": 1.2832,
+      "step": 2541
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.794030276158465e-06,
+      "loss": 1.2916,
+      "step": 2542
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.78738707720186e-06,
+      "loss": 1.3855,
+      "step": 2543
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.780743972116903e-06,
+      "loss": 1.2588,
+      "step": 2544
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.774100963836617e-06,
+      "loss": 1.1956,
+      "step": 2545
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.767458055293995e-06,
+      "loss": 1.3228,
+      "step": 2546
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.760815249421973e-06,
+      "loss": 1.3516,
+      "step": 2547
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.754172549153439e-06,
+      "loss": 1.2354,
+      "step": 2548
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.747529957421246e-06,
+      "loss": 1.2291,
+      "step": 2549
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.740887477158194e-06,
+      "loss": 1.2329,
+      "step": 2550
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.734245111297033e-06,
+      "loss": 1.2363,
+      "step": 2551
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.727602862770456e-06,
+      "loss": 1.2971,
+      "step": 2552
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.720960734511118e-06,
+      "loss": 1.3212,
+      "step": 2553
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.71431872945161e-06,
+      "loss": 1.4043,
+      "step": 2554
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.707676850524473e-06,
+      "loss": 1.3001,
+      "step": 2555
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.701035100662195e-06,
+      "loss": 1.197,
+      "step": 2556
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.694393482797203e-06,
+      "loss": 1.2476,
+      "step": 2557
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.687751999861864e-06,
+      "loss": 1.3236,
+      "step": 2558
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.681110654788483e-06,
+      "loss": 1.1591,
+      "step": 2559
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.674469450509319e-06,
+      "loss": 1.2593,
+      "step": 2560
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.667828389956555e-06,
+      "loss": 1.3164,
+      "step": 2561
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.661187476062308e-06,
+      "loss": 1.1326,
+      "step": 2562
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.654546711758646e-06,
+      "loss": 1.406,
+      "step": 2563
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.647906099977556e-06,
+      "loss": 1.2704,
+      "step": 2564
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.641265643650968e-06,
+      "loss": 1.2453,
+      "step": 2565
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.63462534571073e-06,
+      "loss": 1.3102,
+      "step": 2566
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.627985209088639e-06,
+      "loss": 1.2716,
+      "step": 2567
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.621345236716403e-06,
+      "loss": 1.3473,
+      "step": 2568
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.614705431525665e-06,
+      "loss": 1.2105,
+      "step": 2569
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.608065796447999e-06,
+      "loss": 1.1008,
+      "step": 2570
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.601426334414898e-06,
+      "loss": 1.1675,
+      "step": 2571
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.594787048357776e-06,
+      "loss": 0.9034,
+      "step": 2572
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.588147941207972e-06,
+      "loss": 1.2888,
+      "step": 2573
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.581509015896751e-06,
+      "loss": 1.2966,
+      "step": 2574
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.574870275355295e-06,
+      "loss": 1.209,
+      "step": 2575
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.568231722514697e-06,
+      "loss": 1.1755,
+      "step": 2576
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.561593360305977e-06,
+      "loss": 1.3806,
+      "step": 2577
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.554955191660069e-06,
+      "loss": 1.1436,
+      "step": 2578
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.548317219507815e-06,
+      "loss": 1.3203,
+      "step": 2579
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.541679446779972e-06,
+      "loss": 1.293,
+      "step": 2580
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.535041876407223e-06,
+      "loss": 1.2021,
+      "step": 2581
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.52840451132014e-06,
+      "loss": 1.209,
+      "step": 2582
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.521767354449218e-06,
+      "loss": 1.2117,
+      "step": 2583
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.51513040872486e-06,
+      "loss": 1.2525,
+      "step": 2584
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.50849367707737e-06,
+      "loss": 1.1185,
+      "step": 2585
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.501857162436957e-06,
+      "loss": 1.1053,
+      "step": 2586
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.49522086773374e-06,
+      "loss": 1.3237,
+      "step": 2587
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.488584795897738e-06,
+      "loss": 1.2439,
+      "step": 2588
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.481948949858876e-06,
+      "loss": 1.2176,
+      "step": 2589
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.475313332546964e-06,
+      "loss": 0.9896,
+      "step": 2590
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.468677946891735e-06,
+      "loss": 1.3731,
+      "step": 2591
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.462042795822799e-06,
+      "loss": 1.128,
+      "step": 2592
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.455407882269674e-06,
+      "loss": 1.2368,
+      "step": 2593
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.448773209161762e-06,
+      "loss": 1.062,
+      "step": 2594
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.442138779428376e-06,
+      "loss": 1.2697,
+      "step": 2595
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.435504595998706e-06,
+      "loss": 1.192,
+      "step": 2596
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.42887066180184e-06,
+      "loss": 1.2175,
+      "step": 2597
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.422236979766755e-06,
+      "loss": 1.092,
+      "step": 2598
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.41560355282232e-06,
+      "loss": 1.063,
+      "step": 2599
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.40897038389728e-06,
+      "loss": 1.2951,
+      "step": 2600
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.40233747592028e-06,
+      "loss": 1.3124,
+      "step": 2601
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.395704831819844e-06,
+      "loss": 1.3339,
+      "step": 2602
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.389072454524381e-06,
+      "loss": 1.1463,
+      "step": 2603
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.382440346962169e-06,
+      "loss": 1.0609,
+      "step": 2604
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.375808512061394e-06,
+      "loss": 1.2384,
+      "step": 2605
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.369176952750099e-06,
+      "loss": 1.2777,
+      "step": 2606
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.362545671956213e-06,
+      "loss": 1.1982,
+      "step": 2607
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.355914672607536e-06,
+      "loss": 1.2845,
+      "step": 2608
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.34928395763176e-06,
+      "loss": 1.2293,
+      "step": 2609
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.342653529956432e-06,
+      "loss": 1.2109,
+      "step": 2610
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.33602339250898e-06,
+      "loss": 1.2876,
+      "step": 2611
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.329393548216712e-06,
+      "loss": 1.1726,
+      "step": 2612
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.322764000006798e-06,
+      "loss": 1.1896,
+      "step": 2613
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.316134750806273e-06,
+      "loss": 1.3652,
+      "step": 2614
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.309505803542046e-06,
+      "loss": 1.0929,
+      "step": 2615
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.302877161140897e-06,
+      "loss": 1.2418,
+      "step": 2616
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.296248826529467e-06,
+      "loss": 1.2309,
+      "step": 2617
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.289620802634248e-06,
+      "loss": 1.3184,
+      "step": 2618
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.282993092381626e-06,
+      "loss": 1.3066,
+      "step": 2619
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.276365698697818e-06,
+      "loss": 1.2299,
+      "step": 2620
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.269738624508918e-06,
+      "loss": 1.153,
+      "step": 2621
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.263111872740866e-06,
+      "loss": 1.268,
+      "step": 2622
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.256485446319482e-06,
+      "loss": 1.2275,
+      "step": 2623
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.249859348170417e-06,
+      "loss": 1.1009,
+      "step": 2624
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.24323358121919e-06,
+      "loss": 1.2268,
+      "step": 2625
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.236608148391174e-06,
+      "loss": 0.9954,
+      "step": 2626
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.229983052611598e-06,
+      "loss": 1.3056,
+      "step": 2627
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.223358296805526e-06,
+      "loss": 1.3669,
+      "step": 2628
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.216733883897884e-06,
+      "loss": 1.1895,
+      "step": 2629
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.210109816813452e-06,
+      "loss": 1.3082,
+      "step": 2630
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.20348609847685e-06,
+      "loss": 1.3111,
+      "step": 2631
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.196862731812532e-06,
+      "loss": 1.3483,
+      "step": 2632
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.190239719744828e-06,
+      "loss": 1.1786,
+      "step": 2633
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.183617065197879e-06,
+      "loss": 1.1556,
+      "step": 2634
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.176994771095687e-06,
+      "loss": 1.413,
+      "step": 2635
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.170372840362086e-06,
+      "loss": 1.0855,
+      "step": 2636
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.163751275920762e-06,
+      "loss": 1.1916,
+      "step": 2637
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.15713008069522e-06,
+      "loss": 1.3907,
+      "step": 2638
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.150509257608816e-06,
+      "loss": 1.1006,
+      "step": 2639
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.143888809584741e-06,
+      "loss": 1.2108,
+      "step": 2640
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.137268739546019e-06,
+      "loss": 1.2329,
+      "step": 2641
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.130649050415499e-06,
+      "loss": 1.2254,
+      "step": 2642
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.124029745115872e-06,
+      "loss": 1.2804,
+      "step": 2643
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.117410826569655e-06,
+      "loss": 1.2606,
+      "step": 2644
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.110792297699202e-06,
+      "loss": 1.3446,
+      "step": 2645
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.104174161426674e-06,
+      "loss": 1.1353,
+      "step": 2646
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.097556420674089e-06,
+      "loss": 1.2468,
+      "step": 2647
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.090939078363265e-06,
+      "loss": 1.1284,
+      "step": 2648
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.084322137415855e-06,
+      "loss": 1.2185,
+      "step": 2649
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.07770560075333e-06,
+      "loss": 1.3007,
+      "step": 2650
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.071089471296996e-06,
+      "loss": 1.1858,
+      "step": 2651
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.064473751967958e-06,
+      "loss": 1.1914,
+      "step": 2652
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.057858445687155e-06,
+      "loss": 1.1856,
+      "step": 2653
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.05124355537534e-06,
+      "loss": 1.2149,
+      "step": 2654
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.044629083953082e-06,
+      "loss": 1.3537,
+      "step": 2655
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.038015034340765e-06,
+      "loss": 1.2875,
+      "step": 2656
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.031401409458582e-06,
+      "loss": 1.2426,
+      "step": 2657
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.024788212226549e-06,
+      "loss": 1.3863,
+      "step": 2658
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.018175445564485e-06,
+      "loss": 1.3829,
+      "step": 2659
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.011563112392018e-06,
+      "loss": 1.3553,
+      "step": 2660
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.004951215628597e-06,
+      "loss": 1.3049,
+      "step": 2661
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.99833975819346e-06,
+      "loss": 1.2981,
+      "step": 2662
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.991728743005664e-06,
+      "loss": 1.1604,
+      "step": 2663
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.985118172984063e-06,
+      "loss": 1.2409,
+      "step": 2664
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.978508051047326e-06,
+      "loss": 1.2294,
+      "step": 2665
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.97189838011391e-06,
+      "loss": 1.2416,
+      "step": 2666
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.96528916310208e-06,
+      "loss": 1.2347,
+      "step": 2667
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.958680402929902e-06,
+      "loss": 1.266,
+      "step": 2668
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.95207210251524e-06,
+      "loss": 1.1572,
+      "step": 2669
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.945464264775748e-06,
+      "loss": 1.1027,
+      "step": 2670
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.938856892628881e-06,
+      "loss": 1.2344,
+      "step": 2671
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.932249988991894e-06,
+      "loss": 1.1598,
+      "step": 2672
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.925643556781828e-06,
+      "loss": 1.3899,
+      "step": 2673
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.919037598915513e-06,
+      "loss": 1.2453,
+      "step": 2674
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.912432118309581e-06,
+      "loss": 1.2569,
+      "step": 2675
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.905827117880441e-06,
+      "loss": 1.2363,
+      "step": 2676
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.899222600544298e-06,
+      "loss": 0.98,
+      "step": 2677
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.892618569217134e-06,
+      "loss": 1.2975,
+      "step": 2678
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.886015026814736e-06,
+      "loss": 1.1748,
+      "step": 2679
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.879411976252652e-06,
+      "loss": 1.2805,
+      "step": 2680
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.872809420446225e-06,
+      "loss": 1.1822,
+      "step": 2681
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.866207362310583e-06,
+      "loss": 1.0791,
+      "step": 2682
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.859605804760626e-06,
+      "loss": 1.3375,
+      "step": 2683
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.853004750711038e-06,
+      "loss": 1.2964,
+      "step": 2684
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.846404203076273e-06,
+      "loss": 1.1777,
+      "step": 2685
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.839804164770574e-06,
+      "loss": 1.0691,
+      "step": 2686
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.833204638707948e-06,
+      "loss": 1.1577,
+      "step": 2687
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.826605627802182e-06,
+      "loss": 1.0639,
+      "step": 2688
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.820007134966837e-06,
+      "loss": 1.186,
+      "step": 2689
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.81340916311524e-06,
+      "loss": 1.1502,
+      "step": 2690
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.806811715160485e-06,
+      "loss": 1.3539,
+      "step": 2691
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.800214794015443e-06,
+      "loss": 1.3028,
+      "step": 2692
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.793618402592753e-06,
+      "loss": 1.2329,
+      "step": 2693
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.787022543804814e-06,
+      "loss": 1.1907,
+      "step": 2694
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.780427220563788e-06,
+      "loss": 1.1694,
+      "step": 2695
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.77383243578161e-06,
+      "loss": 1.1247,
+      "step": 2696
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.76723819236997e-06,
+      "loss": 1.2644,
+      "step": 2697
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.760644493240324e-06,
+      "loss": 1.3697,
+      "step": 2698
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.754051341303875e-06,
+      "loss": 1.2928,
+      "step": 2699
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.747458739471603e-06,
+      "loss": 1.3337,
+      "step": 2700
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.740866690654234e-06,
+      "loss": 1.3028,
+      "step": 2701
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.734275197762251e-06,
+      "loss": 1.2258,
+      "step": 2702
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.727684263705896e-06,
+      "loss": 1.2245,
+      "step": 2703
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.721093891395155e-06,
+      "loss": 1.332,
+      "step": 2704
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.714504083739775e-06,
+      "loss": 1.1785,
+      "step": 2705
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.707914843649247e-06,
+      "loss": 1.2153,
+      "step": 2706
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.701326174032821e-06,
+      "loss": 1.3346,
+      "step": 2707
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.694738077799487e-06,
+      "loss": 1.3058,
+      "step": 2708
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.688150557857979e-06,
+      "loss": 1.1847,
+      "step": 2709
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.681563617116786e-06,
+      "loss": 1.1282,
+      "step": 2710
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.674977258484136e-06,
+      "loss": 1.1785,
+      "step": 2711
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.668391484868004e-06,
+      "loss": 1.2085,
+      "step": 2712
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.661806299176095e-06,
+      "loss": 1.1481,
+      "step": 2713
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.655221704315868e-06,
+      "loss": 1.1803,
+      "step": 2714
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.648637703194515e-06,
+      "loss": 1.2135,
+      "step": 2715
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.642054298718966e-06,
+      "loss": 1.3462,
+      "step": 2716
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.635471493795895e-06,
+      "loss": 1.2759,
+      "step": 2717
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.628889291331696e-06,
+      "loss": 1.1836,
+      "step": 2718
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.622307694232508e-06,
+      "loss": 1.1136,
+      "step": 2719
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.6157267054042e-06,
+      "loss": 1.2749,
+      "step": 2720
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.609146327752376e-06,
+      "loss": 1.248,
+      "step": 2721
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.602566564182366e-06,
+      "loss": 1.2897,
+      "step": 2722
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.595987417599225e-06,
+      "loss": 1.347,
+      "step": 2723
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.589408890907746e-06,
+      "loss": 1.1125,
+      "step": 2724
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.582830987012442e-06,
+      "loss": 1.3633,
+      "step": 2725
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.576253708817553e-06,
+      "loss": 1.2476,
+      "step": 2726
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.569677059227033e-06,
+      "loss": 1.3306,
+      "step": 2727
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.563101041144575e-06,
+      "loss": 1.2804,
+      "step": 2728
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.556525657473586e-06,
+      "loss": 1.2069,
+      "step": 2729
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.549950911117186e-06,
+      "loss": 1.127,
+      "step": 2730
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.543376804978223e-06,
+      "loss": 1.2222,
+      "step": 2731
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.536803341959263e-06,
+      "loss": 1.117,
+      "step": 2732
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.530230524962578e-06,
+      "loss": 1.2575,
+      "step": 2733
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.52365835689016e-06,
+      "loss": 1.143,
+      "step": 2734
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.517086840643722e-06,
+      "loss": 1.1809,
+      "step": 2735
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.51051597912468e-06,
+      "loss": 1.0164,
+      "step": 2736
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.503945775234157e-06,
+      "loss": 1.1162,
+      "step": 2737
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.497376231873e-06,
+      "loss": 1.2667,
+      "step": 2738
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.490807351941753e-06,
+      "loss": 1.1698,
+      "step": 2739
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.484239138340674e-06,
+      "loss": 1.2744,
+      "step": 2740
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.477671593969717e-06,
+      "loss": 1.1083,
+      "step": 2741
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.471104721728551e-06,
+      "loss": 1.3262,
+      "step": 2742
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.464538524516546e-06,
+      "loss": 1.1445,
+      "step": 2743
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.457973005232765e-06,
+      "loss": 1.2086,
+      "step": 2744
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.451408166775987e-06,
+      "loss": 1.2323,
+      "step": 2745
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.444844012044681e-06,
+      "loss": 1.1741,
+      "step": 2746
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.438280543937012e-06,
+      "loss": 1.1908,
+      "step": 2747
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.431717765350842e-06,
+      "loss": 1.2897,
+      "step": 2748
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.425155679183738e-06,
+      "loss": 1.1715,
+      "step": 2749
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.418594288332957e-06,
+      "loss": 1.3528,
+      "step": 2750
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.412033595695436e-06,
+      "loss": 1.2707,
+      "step": 2751
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.405473604167826e-06,
+      "loss": 1.1992,
+      "step": 2752
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.398914316646452e-06,
+      "loss": 1.3241,
+      "step": 2753
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.392355736027338e-06,
+      "loss": 1.4286,
+      "step": 2754
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.385797865206178e-06,
+      "loss": 1.1344,
+      "step": 2755
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.379240707078385e-06,
+      "loss": 1.2732,
+      "step": 2756
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.372684264539025e-06,
+      "loss": 1.1888,
+      "step": 2757
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.366128540482861e-06,
+      "loss": 1.378,
+      "step": 2758
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.359573537804347e-06,
+      "loss": 1.1247,
+      "step": 2759
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.353019259397609e-06,
+      "loss": 1.1456,
+      "step": 2760
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.346465708156449e-06,
+      "loss": 1.1445,
+      "step": 2761
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.339912886974355e-06,
+      "loss": 1.22,
+      "step": 2762
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.333360798744496e-06,
+      "loss": 1.202,
+      "step": 2763
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.326809446359711e-06,
+      "loss": 1.3407,
+      "step": 2764
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.320258832712513e-06,
+      "loss": 1.2587,
+      "step": 2765
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.313708960695094e-06,
+      "loss": 1.4483,
+      "step": 2766
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.307159833199317e-06,
+      "loss": 1.4281,
+      "step": 2767
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.300611453116714e-06,
+      "loss": 1.0554,
+      "step": 2768
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.294063823338486e-06,
+      "loss": 1.2257,
+      "step": 2769
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.287516946755514e-06,
+      "loss": 1.1578,
+      "step": 2770
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.28097082625833e-06,
+      "loss": 1.234,
+      "step": 2771
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.274425464737139e-06,
+      "loss": 1.1932,
+      "step": 2772
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.26788086508182e-06,
+      "loss": 1.1787,
+      "step": 2773
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.261337030181901e-06,
+      "loss": 1.2876,
+      "step": 2774
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.254793962926579e-06,
+      "loss": 1.2697,
+      "step": 2775
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.248251666204711e-06,
+      "loss": 1.347,
+      "step": 2776
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.241710142904821e-06,
+      "loss": 1.0467,
+      "step": 2777
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.23516939591508e-06,
+      "loss": 1.1199,
+      "step": 2778
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.228629428123319e-06,
+      "loss": 1.184,
+      "step": 2779
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.222090242417036e-06,
+      "loss": 1.2906,
+      "step": 2780
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.215551841683369e-06,
+      "loss": 1.2454,
+      "step": 2781
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.209014228809117e-06,
+      "loss": 1.1697,
+      "step": 2782
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.202477406680725e-06,
+      "loss": 1.1635,
+      "step": 2783
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.195941378184303e-06,
+      "loss": 1.2213,
+      "step": 2784
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.189406146205594e-06,
+      "loss": 1.1817,
+      "step": 2785
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.182871713629998e-06,
+      "loss": 1.2535,
+      "step": 2786
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.17633808334256e-06,
+      "loss": 1.1494,
+      "step": 2787
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.169805258227978e-06,
+      "loss": 1.2079,
+      "step": 2788
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.16327324117058e-06,
+      "loss": 1.1865,
+      "step": 2789
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.156742035054343e-06,
+      "loss": 1.3215,
+      "step": 2790
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.150211642762895e-06,
+      "loss": 1.3105,
+      "step": 2791
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.143682067179498e-06,
+      "loss": 1.2464,
+      "step": 2792
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.137153311187044e-06,
+      "loss": 0.9995,
+      "step": 2793
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.130625377668084e-06,
+      "loss": 1.3771,
+      "step": 2794
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.124098269504786e-06,
+      "loss": 1.1707,
+      "step": 2795
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.117571989578969e-06,
+      "loss": 1.2211,
+      "step": 2796
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.111046540772064e-06,
+      "loss": 1.2496,
+      "step": 2797
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.10452192596517e-06,
+      "loss": 1.3008,
+      "step": 2798
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.097998148038986e-06,
+      "loss": 0.9738,
+      "step": 2799
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.091475209873853e-06,
+      "loss": 1.2004,
+      "step": 2800
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.084953114349745e-06,
+      "loss": 1.2186,
+      "step": 2801
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.078431864346265e-06,
+      "loss": 1.2057,
+      "step": 2802
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.07191146274263e-06,
+      "loss": 1.3492,
+      "step": 2803
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.06539191241769e-06,
+      "loss": 1.2724,
+      "step": 2804
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.058873216249927e-06,
+      "loss": 1.0831,
+      "step": 2805
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.052355377117438e-06,
+      "loss": 1.2457,
+      "step": 2806
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.045838397897932e-06,
+      "loss": 1.2208,
+      "step": 2807
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.039322281468765e-06,
+      "loss": 1.1256,
+      "step": 2808
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.032807030706882e-06,
+      "loss": 1.1192,
+      "step": 2809
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.026292648488864e-06,
+      "loss": 1.2403,
+      "step": 2810
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.019779137690906e-06,
+      "loss": 1.2192,
+      "step": 2811
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.013266501188816e-06,
+      "loss": 1.1547,
+      "step": 2812
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.006754741858017e-06,
+      "loss": 1.2959,
+      "step": 2813
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.000243862573536e-06,
+      "loss": 1.3132,
+      "step": 2814
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.993733866210029e-06,
+      "loss": 1.3543,
+      "step": 2815
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.98722475564175e-06,
+      "loss": 1.0682,
+      "step": 2816
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.980716533742558e-06,
+      "loss": 1.0957,
+      "step": 2817
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.97420920338594e-06,
+      "loss": 1.253,
+      "step": 2818
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.967702767444963e-06,
+      "loss": 1.224,
+      "step": 2819
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.96119722879232e-06,
+      "loss": 1.1378,
+      "step": 2820
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.954692590300288e-06,
+      "loss": 1.1586,
+      "step": 2821
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.948188854840773e-06,
+      "loss": 1.0884,
+      "step": 2822
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.941686025285257e-06,
+      "loss": 1.2531,
+      "step": 2823
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.935184104504832e-06,
+      "loss": 1.1699,
+      "step": 2824
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.928683095370195e-06,
+      "loss": 1.3299,
+      "step": 2825
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.922183000751635e-06,
+      "loss": 1.1287,
+      "step": 2826
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.915683823519031e-06,
+      "loss": 1.1826,
+      "step": 2827
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.909185566541866e-06,
+      "loss": 1.3555,
+      "step": 2828
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.902688232689212e-06,
+      "loss": 1.2069,
+      "step": 2829
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.89619182482974e-06,
+      "loss": 1.1787,
+      "step": 2830
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.889696345831696e-06,
+      "loss": 1.0652,
+      "step": 2831
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.88320179856294e-06,
+      "loss": 1.2135,
+      "step": 2832
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.876708185890901e-06,
+      "loss": 1.3137,
+      "step": 2833
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.870215510682604e-06,
+      "loss": 1.3093,
+      "step": 2834
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.863723775804651e-06,
+      "loss": 1.1619,
+      "step": 2835
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.857232984123249e-06,
+      "loss": 1.1335,
+      "step": 2836
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.850743138504165e-06,
+      "loss": 1.2676,
+      "step": 2837
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.84425424181276e-06,
+      "loss": 1.3829,
+      "step": 2838
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.83776629691398e-06,
+      "loss": 1.1842,
+      "step": 2839
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.831279306672343e-06,
+      "loss": 1.4014,
+      "step": 2840
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.824793273951945e-06,
+      "loss": 1.2043,
+      "step": 2841
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.818308201616465e-06,
+      "loss": 1.169,
+      "step": 2842
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.811824092529155e-06,
+      "loss": 1.2577,
+      "step": 2843
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.805340949552846e-06,
+      "loss": 1.2658,
+      "step": 2844
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.798858775549927e-06,
+      "loss": 1.1956,
+      "step": 2845
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.792377573382385e-06,
+      "loss": 1.2504,
+      "step": 2846
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.785897345911755e-06,
+      "loss": 1.2312,
+      "step": 2847
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.779418095999151e-06,
+      "loss": 1.1751,
+      "step": 2848
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.772939826505254e-06,
+      "loss": 1.2383,
+      "step": 2849
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.76646254029032e-06,
+      "loss": 1.0334,
+      "step": 2850
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.759986240214155e-06,
+      "loss": 1.1136,
+      "step": 2851
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.753510929136137e-06,
+      "loss": 1.2685,
+      "step": 2852
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.747036609915217e-06,
+      "loss": 1.1207,
+      "step": 2853
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.740563285409898e-06,
+      "loss": 1.3093,
+      "step": 2854
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.734090958478241e-06,
+      "loss": 1.4008,
+      "step": 2855
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.727619631977869e-06,
+      "loss": 1.1483,
+      "step": 2856
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.721149308765973e-06,
+      "loss": 1.1278,
+      "step": 2857
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.714679991699292e-06,
+      "loss": 1.2478,
+      "step": 2858
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.708211683634112e-06,
+      "loss": 1.4012,
+      "step": 2859
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.701744387426297e-06,
+      "loss": 1.4497,
+      "step": 2860
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.695278105931244e-06,
+      "loss": 1.3307,
+      "step": 2861
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.68881284200391e-06,
+      "loss": 1.0848,
+      "step": 2862
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.6823485984988e-06,
+      "loss": 1.2978,
+      "step": 2863
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.675885378269977e-06,
+      "loss": 1.2089,
+      "step": 2864
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.669423184171037e-06,
+      "loss": 1.2858,
+      "step": 2865
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.662962019055133e-06,
+      "loss": 1.0612,
+      "step": 2866
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.656501885774963e-06,
+      "loss": 1.3854,
+      "step": 2867
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.650042787182773e-06,
+      "loss": 1.1552,
+      "step": 2868
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.64358472613034e-06,
+      "loss": 1.1784,
+      "step": 2869
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.637127705468992e-06,
+      "loss": 1.2618,
+      "step": 2870
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.6306717280496e-06,
+      "loss": 1.3558,
+      "step": 2871
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.624216796722567e-06,
+      "loss": 1.321,
+      "step": 2872
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.617762914337837e-06,
+      "loss": 1.3213,
+      "step": 2873
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.611310083744897e-06,
+      "loss": 1.1186,
+      "step": 2874
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.604858307792759e-06,
+      "loss": 1.394,
+      "step": 2875
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.598407589329975e-06,
+      "loss": 1.2227,
+      "step": 2876
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.591957931204629e-06,
+      "loss": 1.1006,
+      "step": 2877
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5855093362643425e-06,
+      "loss": 1.1695,
+      "step": 2878
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.579061807356256e-06,
+      "loss": 1.2698,
+      "step": 2879
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.572615347327048e-06,
+      "loss": 1.1663,
+      "step": 2880
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.566169959022925e-06,
+      "loss": 1.358,
+      "step": 2881
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.55972564528962e-06,
+      "loss": 1.1597,
+      "step": 2882
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.553282408972382e-06,
+      "loss": 1.1989,
+      "step": 2883
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.546840252915996e-06,
+      "loss": 1.2774,
+      "step": 2884
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.540399179964769e-06,
+      "loss": 1.3449,
+      "step": 2885
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.533959192962525e-06,
+      "loss": 1.457,
+      "step": 2886
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.527520294752607e-06,
+      "loss": 1.0745,
+      "step": 2887
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.521082488177889e-06,
+      "loss": 1.1999,
+      "step": 2888
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.514645776080747e-06,
+      "loss": 1.2357,
+      "step": 2889
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.508210161303085e-06,
+      "loss": 1.2114,
+      "step": 2890
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.501775646686316e-06,
+      "loss": 1.3164,
+      "step": 2891
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.495342235071377e-06,
+      "loss": 1.3166,
+      "step": 2892
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.488909929298706e-06,
+      "loss": 1.1279,
+      "step": 2893
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.482478732208256e-06,
+      "loss": 1.253,
+      "step": 2894
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.476048646639498e-06,
+      "loss": 1.2941,
+      "step": 2895
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.469619675431406e-06,
+      "loss": 1.3033,
+      "step": 2896
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.463191821422462e-06,
+      "loss": 1.1463,
+      "step": 2897
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.456765087450652e-06,
+      "loss": 1.2359,
+      "step": 2898
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.450339476353475e-06,
+      "loss": 1.2852,
+      "step": 2899
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.443914990967931e-06,
+      "loss": 1.1757,
+      "step": 2900
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.437491634130518e-06,
+      "loss": 1.2645,
+      "step": 2901
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.431069408677248e-06,
+      "loss": 1.274,
+      "step": 2902
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.424648317443616e-06,
+      "loss": 1.1227,
+      "step": 2903
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.418228363264631e-06,
+      "loss": 1.2976,
+      "step": 2904
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.411809548974792e-06,
+      "loss": 1.2207,
+      "step": 2905
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.405391877408103e-06,
+      "loss": 1.266,
+      "step": 2906
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.398975351398054e-06,
+      "loss": 1.2147,
+      "step": 2907
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3925599737776285e-06,
+      "loss": 1.1505,
+      "step": 2908
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.386145747379316e-06,
+      "loss": 1.19,
+      "step": 2909
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.379732675035085e-06,
+      "loss": 1.172,
+      "step": 2910
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3733207595764015e-06,
+      "loss": 1.0521,
+      "step": 2911
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.366910003834211e-06,
+      "loss": 1.1811,
+      "step": 2912
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.360500410638962e-06,
+      "loss": 1.1627,
+      "step": 2913
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.354091982820578e-06,
+      "loss": 1.2792,
+      "step": 2914
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.347684723208471e-06,
+      "loss": 1.2557,
+      "step": 2915
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3412786346315415e-06,
+      "loss": 1.1789,
+      "step": 2916
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.334873719918165e-06,
+      "loss": 1.3499,
+      "step": 2917
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.328469981896207e-06,
+      "loss": 1.2987,
+      "step": 2918
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.322067423393002e-06,
+      "loss": 1.2166,
+      "step": 2919
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.3156660472353814e-06,
+      "loss": 1.2152,
+      "step": 2920
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.309265856249641e-06,
+      "loss": 1.2961,
+      "step": 2921
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.302866853261552e-06,
+      "loss": 1.3708,
+      "step": 2922
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2964690410963705e-06,
+      "loss": 1.2708,
+      "step": 2923
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2900724225788225e-06,
+      "loss": 1.351,
+      "step": 2924
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.283677000533106e-06,
+      "loss": 1.1894,
+      "step": 2925
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.277282777782886e-06,
+      "loss": 1.1922,
+      "step": 2926
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2708897571513116e-06,
+      "loss": 1.0869,
+      "step": 2927
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.264497941460988e-06,
+      "loss": 1.2893,
+      "step": 2928
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.258107333533994e-06,
+      "loss": 1.2565,
+      "step": 2929
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.251717936191879e-06,
+      "loss": 1.2063,
+      "step": 2930
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2453297522556475e-06,
+      "loss": 1.1493,
+      "step": 2931
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.238942784545777e-06,
+      "loss": 1.2779,
+      "step": 2932
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.232557035882202e-06,
+      "loss": 1.0251,
+      "step": 2933
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.226172509084326e-06,
+      "loss": 1.0917,
+      "step": 2934
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.2197892069710105e-06,
+      "loss": 1.158,
+      "step": 2935
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.213407132360567e-06,
+      "loss": 1.2059,
+      "step": 2936
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.20702628807078e-06,
+      "loss": 1.3261,
+      "step": 2937
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.200646676918881e-06,
+      "loss": 1.2429,
+      "step": 2938
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.194268301721563e-06,
+      "loss": 1.3238,
+      "step": 2939
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1878911652949626e-06,
+      "loss": 1.1851,
+      "step": 2940
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.181515270454683e-06,
+      "loss": 1.2025,
+      "step": 2941
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1751406200157714e-06,
+      "loss": 1.194,
+      "step": 2942
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.168767216792725e-06,
+      "loss": 0.9941,
+      "step": 2943
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.162395063599497e-06,
+      "loss": 1.3881,
+      "step": 2944
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1560241632494845e-06,
+      "loss": 1.3128,
+      "step": 2945
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.149654518555526e-06,
+      "loss": 1.1922,
+      "step": 2946
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.143286132329912e-06,
+      "loss": 1.1684,
+      "step": 2947
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.136919007384379e-06,
+      "loss": 1.0854,
+      "step": 2948
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.130553146530105e-06,
+      "loss": 1.2073,
+      "step": 2949
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.124188552577702e-06,
+      "loss": 1.2219,
+      "step": 2950
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.117825228337236e-06,
+      "loss": 1.2286,
+      "step": 2951
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.111463176618205e-06,
+      "loss": 1.2843,
+      "step": 2952
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.1051024002295444e-06,
+      "loss": 1.1799,
+      "step": 2953
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.098742901979625e-06,
+      "loss": 1.1449,
+      "step": 2954
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.092384684676263e-06,
+      "loss": 1.1528,
+      "step": 2955
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.086027751126696e-06,
+      "loss": 1.0723,
+      "step": 2956
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.079672104137604e-06,
+      "loss": 1.3355,
+      "step": 2957
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.073317746515097e-06,
+      "loss": 1.1193,
+      "step": 2958
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.066964681064716e-06,
+      "loss": 1.0779,
+      "step": 2959
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0606129105914274e-06,
+      "loss": 1.1559,
+      "step": 2960
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.054262437899628e-06,
+      "loss": 1.0991,
+      "step": 2961
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.047913265793146e-06,
+      "loss": 1.4268,
+      "step": 2962
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.041565397075232e-06,
+      "loss": 1.0654,
+      "step": 2963
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.035218834548554e-06,
+      "loss": 1.3243,
+      "step": 2964
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.028873581015219e-06,
+      "loss": 1.23,
+      "step": 2965
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.022529639276743e-06,
+      "loss": 1.0725,
+      "step": 2966
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.016187012134069e-06,
+      "loss": 1.1924,
+      "step": 2967
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.009845702387551e-06,
+      "loss": 1.1393,
+      "step": 2968
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.00350571283698e-06,
+      "loss": 1.1225,
+      "step": 2969
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9971670462815425e-06,
+      "loss": 1.2235,
+      "step": 2970
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.990829705519853e-06,
+      "loss": 1.3524,
+      "step": 2971
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.984493693349939e-06,
+      "loss": 1.2313,
+      "step": 2972
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.978159012569243e-06,
+      "loss": 1.2088,
+      "step": 2973
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.971825665974614e-06,
+      "loss": 1.2967,
+      "step": 2974
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.965493656362314e-06,
+      "loss": 1.2626,
+      "step": 2975
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.959162986528019e-06,
+      "loss": 1.2066,
+      "step": 2976
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.952833659266811e-06,
+      "loss": 1.3015,
+      "step": 2977
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.946505677373172e-06,
+      "loss": 1.1385,
+      "step": 2978
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.940179043641005e-06,
+      "loss": 1.004,
+      "step": 2979
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.933853760863607e-06,
+      "loss": 1.1988,
+      "step": 2980
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.927529831833679e-06,
+      "loss": 1.3904,
+      "step": 2981
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.921207259343323e-06,
+      "loss": 1.2878,
+      "step": 2982
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.914886046184055e-06,
+      "loss": 1.3005,
+      "step": 2983
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.908566195146773e-06,
+      "loss": 1.1902,
+      "step": 2984
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.902247709021783e-06,
+      "loss": 1.2889,
+      "step": 2985
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.895930590598789e-06,
+      "loss": 1.2666,
+      "step": 2986
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.889614842666892e-06,
+      "loss": 1.3229,
+      "step": 2987
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.88330046801458e-06,
+      "loss": 1.2321,
+      "step": 2988
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.876987469429738e-06,
+      "loss": 1.3007,
+      "step": 2989
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.870675849699651e-06,
+      "loss": 1.1943,
+      "step": 2990
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.86436561161099e-06,
+      "loss": 1.1063,
+      "step": 2991
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.858056757949807e-06,
+      "loss": 1.1839,
+      "step": 2992
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.851749291501561e-06,
+      "loss": 1.0745,
+      "step": 2993
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.845443215051084e-06,
+      "loss": 1.2877,
+      "step": 2994
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8391385313826035e-06,
+      "loss": 1.2686,
+      "step": 2995
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.832835243279717e-06,
+      "loss": 1.1899,
+      "step": 2996
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.826533353525427e-06,
+      "loss": 1.3902,
+      "step": 2997
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8202328649021055e-06,
+      "loss": 1.1806,
+      "step": 2998
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.813933780191504e-06,
+      "loss": 1.2948,
+      "step": 2999
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.807636102174764e-06,
+      "loss": 1.1321,
+      "step": 3000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.801339833632401e-06,
+      "loss": 1.2385,
+      "step": 3001
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.795044977344303e-06,
+      "loss": 1.3925,
+      "step": 3002
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.788751536089739e-06,
+      "loss": 1.2031,
+      "step": 3003
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.782459512647358e-06,
+      "loss": 1.3125,
+      "step": 3004
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.776168909795178e-06,
+      "loss": 1.3596,
+      "step": 3005
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.769879730310583e-06,
+      "loss": 1.1317,
+      "step": 3006
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.763591976970347e-06,
+      "loss": 1.3449,
+      "step": 3007
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.757305652550596e-06,
+      "loss": 1.0799,
+      "step": 3008
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.751020759826836e-06,
+      "loss": 1.3843,
+      "step": 3009
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.744737301573929e-06,
+      "loss": 1.2709,
+      "step": 3010
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.738455280566124e-06,
+      "loss": 1.3795,
+      "step": 3011
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7321746995770144e-06,
+      "loss": 1.1806,
+      "step": 3012
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.725895561379567e-06,
+      "loss": 1.1576,
+      "step": 3013
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.719617868746116e-06,
+      "loss": 1.2625,
+      "step": 3014
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.713341624448351e-06,
+      "loss": 1.2405,
+      "step": 3015
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.70706683125732e-06,
+      "loss": 1.2748,
+      "step": 3016
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.700793491943436e-06,
+      "loss": 1.1901,
+      "step": 3017
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.694521609276471e-06,
+      "loss": 1.2138,
+      "step": 3018
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6882511860255505e-06,
+      "loss": 1.2476,
+      "step": 3019
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.681982224959149e-06,
+      "loss": 1.1342,
+      "step": 3020
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.675714728845114e-06,
+      "loss": 1.2942,
+      "step": 3021
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.669448700450627e-06,
+      "loss": 1.2897,
+      "step": 3022
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.663184142542233e-06,
+      "loss": 1.1017,
+      "step": 3023
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.656921057885815e-06,
+      "loss": 1.4481,
+      "step": 3024
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.65065944924663e-06,
+      "loss": 1.1895,
+      "step": 3025
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6443993193892566e-06,
+      "loss": 1.3758,
+      "step": 3026
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.638140671077633e-06,
+      "loss": 1.3779,
+      "step": 3027
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.631883507075045e-06,
+      "loss": 1.0059,
+      "step": 3028
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.625627830144121e-06,
+      "loss": 1.1189,
+      "step": 3029
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.619373643046829e-06,
+      "loss": 1.1435,
+      "step": 3030
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.613120948544482e-06,
+      "loss": 1.3319,
+      "step": 3031
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.606869749397735e-06,
+      "loss": 1.2026,
+      "step": 3032
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.600620048366586e-06,
+      "loss": 1.2233,
+      "step": 3033
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.594371848210357e-06,
+      "loss": 1.1201,
+      "step": 3034
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.58812515168773e-06,
+      "loss": 1.2551,
+      "step": 3035
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.581879961556703e-06,
+      "loss": 1.1169,
+      "step": 3036
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.57563628057462e-06,
+      "loss": 1.2952,
+      "step": 3037
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.569394111498153e-06,
+      "loss": 1.3411,
+      "step": 3038
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.563153457083315e-06,
+      "loss": 1.2582,
+      "step": 3039
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.556914320085439e-06,
+      "loss": 1.2529,
+      "step": 3040
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.550676703259193e-06,
+      "loss": 1.1444,
+      "step": 3041
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.544440609358575e-06,
+      "loss": 1.2701,
+      "step": 3042
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.538206041136915e-06,
+      "loss": 1.2749,
+      "step": 3043
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.531973001346857e-06,
+      "loss": 1.1584,
+      "step": 3044
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.525741492740378e-06,
+      "loss": 1.1376,
+      "step": 3045
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.519511518068782e-06,
+      "loss": 1.2822,
+      "step": 3046
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.513283080082692e-06,
+      "loss": 1.1845,
+      "step": 3047
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.507056181532045e-06,
+      "loss": 1.0274,
+      "step": 3048
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.5008308251661175e-06,
+      "loss": 1.3019,
+      "step": 3049
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.494607013733486e-06,
+      "loss": 1.2161,
+      "step": 3050
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.488384749982054e-06,
+      "loss": 1.3315,
+      "step": 3051
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.482164036659037e-06,
+      "loss": 1.3374,
+      "step": 3052
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4759448765109776e-06,
+      "loss": 1.2623,
+      "step": 3053
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.469727272283717e-06,
+      "loss": 1.1589,
+      "step": 3054
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.463511226722416e-06,
+      "loss": 1.1198,
+      "step": 3055
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.45729674257155e-06,
+      "loss": 1.3226,
+      "step": 3056
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.451083822574907e-06,
+      "loss": 1.2133,
+      "step": 3057
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.444872469475572e-06,
+      "loss": 1.2218,
+      "step": 3058
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.438662686015947e-06,
+      "loss": 1.1097,
+      "step": 3059
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.432454474937747e-06,
+      "loss": 1.1947,
+      "step": 3060
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.426247838981983e-06,
+      "loss": 1.296,
+      "step": 3061
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.420042780888969e-06,
+      "loss": 1.132,
+      "step": 3062
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.413839303398337e-06,
+      "loss": 1.1681,
+      "step": 3063
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.407637409249002e-06,
+      "loss": 1.1032,
+      "step": 3064
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.401437101179193e-06,
+      "loss": 0.9001,
+      "step": 3065
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.395238381926432e-06,
+      "loss": 1.1597,
+      "step": 3066
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.389041254227548e-06,
+      "loss": 1.4187,
+      "step": 3067
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3828457208186564e-06,
+      "loss": 1.2821,
+      "step": 3068
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.376651784435174e-06,
+      "loss": 1.3673,
+      "step": 3069
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.370459447811815e-06,
+      "loss": 1.2084,
+      "step": 3070
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.364268713682586e-06,
+      "loss": 1.2651,
+      "step": 3071
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.358079584780781e-06,
+      "loss": 1.3693,
+      "step": 3072
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.351892063838988e-06,
+      "loss": 1.1778,
+      "step": 3073
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.345706153589091e-06,
+      "loss": 1.1247,
+      "step": 3074
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.339521856762254e-06,
+      "loss": 1.4585,
+      "step": 3075
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.333339176088934e-06,
+      "loss": 1.2706,
+      "step": 3076
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.327158114298877e-06,
+      "loss": 1.1358,
+      "step": 3077
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.320978674121103e-06,
+      "loss": 1.2426,
+      "step": 3078
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3148008582839284e-06,
+      "loss": 1.3212,
+      "step": 3079
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.308624669514942e-06,
+      "loss": 1.1746,
+      "step": 3080
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.3024501105410295e-06,
+      "loss": 1.0435,
+      "step": 3081
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2962771840883375e-06,
+      "loss": 1.2943,
+      "step": 3082
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.290105892882303e-06,
+      "loss": 1.1666,
+      "step": 3083
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.283936239647643e-06,
+      "loss": 1.2722,
+      "step": 3084
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.277768227108348e-06,
+      "loss": 1.178,
+      "step": 3085
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.271601857987682e-06,
+      "loss": 1.2538,
+      "step": 3086
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.265437135008181e-06,
+      "loss": 1.3091,
+      "step": 3087
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.259274060891665e-06,
+      "loss": 1.1906,
+      "step": 3088
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.253112638359218e-06,
+      "loss": 1.3112,
+      "step": 3089
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2469528701311935e-06,
+      "loss": 1.1012,
+      "step": 3090
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.240794758927222e-06,
+      "loss": 1.3035,
+      "step": 3091
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.234638307466193e-06,
+      "loss": 1.2485,
+      "step": 3092
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.228483518466271e-06,
+      "loss": 1.5009,
+      "step": 3093
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.22233039464488e-06,
+      "loss": 1.2708,
+      "step": 3094
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.21617893871872e-06,
+      "loss": 1.2424,
+      "step": 3095
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.210029153403738e-06,
+      "loss": 1.1488,
+      "step": 3096
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.203881041415154e-06,
+      "loss": 1.3873,
+      "step": 3097
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.197734605467453e-06,
+      "loss": 1.3308,
+      "step": 3098
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.191589848274369e-06,
+      "loss": 1.1287,
+      "step": 3099
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.185446772548905e-06,
+      "loss": 1.1539,
+      "step": 3100
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.179305381003311e-06,
+      "loss": 1.1266,
+      "step": 3101
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.173165676349103e-06,
+      "loss": 1.1997,
+      "step": 3102
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.167027661297049e-06,
+      "loss": 1.2376,
+      "step": 3103
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.160891338557167e-06,
+      "loss": 1.0749,
+      "step": 3104
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.154756710838739e-06,
+      "loss": 1.2696,
+      "step": 3105
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.148623780850282e-06,
+      "loss": 1.2579,
+      "step": 3106
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.142492551299577e-06,
+      "loss": 1.2637,
+      "step": 3107
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.136363024893647e-06,
+      "loss": 1.3041,
+      "step": 3108
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.13023520433877e-06,
+      "loss": 1.2839,
+      "step": 3109
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.124109092340465e-06,
+      "loss": 1.2853,
+      "step": 3110
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.117984691603493e-06,
+      "loss": 1.2745,
+      "step": 3111
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.1118620048318724e-06,
+      "loss": 1.2592,
+      "step": 3112
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.1057410347288535e-06,
+      "loss": 1.1845,
+      "step": 3113
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.099621783996934e-06,
+      "loss": 1.2793,
+      "step": 3114
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.093504255337844e-06,
+      "loss": 1.1784,
+      "step": 3115
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.087388451452567e-06,
+      "loss": 1.1503,
+      "step": 3116
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.081274375041317e-06,
+      "loss": 0.9678,
+      "step": 3117
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.075162028803544e-06,
+      "loss": 1.3078,
+      "step": 3118
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.069051415437941e-06,
+      "loss": 0.9762,
+      "step": 3119
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.062942537642427e-06,
+      "loss": 1.2539,
+      "step": 3120
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.056835398114158e-06,
+      "loss": 1.1238,
+      "step": 3121
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.050729999549525e-06,
+      "loss": 1.2478,
+      "step": 3122
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.044626344644151e-06,
+      "loss": 1.1814,
+      "step": 3123
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.038524436092887e-06,
+      "loss": 1.3059,
+      "step": 3124
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.032424276589808e-06,
+      "loss": 1.5643,
+      "step": 3125
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.026325868828228e-06,
+      "loss": 1.1054,
+      "step": 3126
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.020229215500677e-06,
+      "loss": 1.4116,
+      "step": 3127
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.014134319298921e-06,
+      "loss": 1.1431,
+      "step": 3128
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.008041182913933e-06,
+      "loss": 1.1358,
+      "step": 3129
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.001949809035929e-06,
+      "loss": 1.2836,
+      "step": 3130
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.995860200354335e-06,
+      "loss": 1.2656,
+      "step": 3131
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.989772359557799e-06,
+      "loss": 1.1137,
+      "step": 3132
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.983686289334194e-06,
+      "loss": 1.2772,
+      "step": 3133
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9776019923706055e-06,
+      "loss": 1.247,
+      "step": 3134
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.971519471353335e-06,
+      "loss": 1.1771,
+      "step": 3135
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.965438728967902e-06,
+      "loss": 1.2439,
+      "step": 3136
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.959359767899048e-06,
+      "loss": 1.1474,
+      "step": 3137
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.953282590830718e-06,
+      "loss": 1.3395,
+      "step": 3138
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9472072004460665e-06,
+      "loss": 1.0985,
+      "step": 3139
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.941133599427474e-06,
+      "loss": 1.321,
+      "step": 3140
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.935061790456519e-06,
+      "loss": 1.3719,
+      "step": 3141
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.928991776213994e-06,
+      "loss": 1.2798,
+      "step": 3142
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.922923559379892e-06,
+      "loss": 1.237,
+      "step": 3143
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.916857142633421e-06,
+      "loss": 1.4187,
+      "step": 3144
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.910792528652992e-06,
+      "loss": 1.4825,
+      "step": 3145
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.904729720116216e-06,
+      "loss": 1.3337,
+      "step": 3146
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.898668719699914e-06,
+      "loss": 1.2022,
+      "step": 3147
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.892609530080102e-06,
+      "loss": 1.2401,
+      "step": 3148
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.886552153931999e-06,
+      "loss": 1.0555,
+      "step": 3149
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.880496593930021e-06,
+      "loss": 1.3009,
+      "step": 3150
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.874442852747788e-06,
+      "loss": 1.0944,
+      "step": 3151
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8683909330581146e-06,
+      "loss": 1.1812,
+      "step": 3152
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8623408375330035e-06,
+      "loss": 1.3206,
+      "step": 3153
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.856292568843665e-06,
+      "loss": 1.3106,
+      "step": 3154
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.850246129660494e-06,
+      "loss": 1.3448,
+      "step": 3155
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.844201522653081e-06,
+      "loss": 1.1961,
+      "step": 3156
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8381587504902025e-06,
+      "loss": 1.073,
+      "step": 3157
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.832117815839833e-06,
+      "loss": 1.2696,
+      "step": 3158
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.82607872136913e-06,
+      "loss": 1.1681,
+      "step": 3159
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.820041469744443e-06,
+      "loss": 1.1146,
+      "step": 3160
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.8140060636313015e-06,
+      "loss": 1.1904,
+      "step": 3161
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.8079725056944245e-06,
+      "loss": 1.4044,
+      "step": 3162
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.801940798597716e-06,
+      "loss": 1.2623,
+      "step": 3163
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.795910945004255e-06,
+      "loss": 1.1909,
+      "step": 3164
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.789882947576321e-06,
+      "loss": 1.3287,
+      "step": 3165
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.783856808975349e-06,
+      "loss": 1.0969,
+      "step": 3166
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7778325318619665e-06,
+      "loss": 1.2665,
+      "step": 3167
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.771810118895984e-06,
+      "loss": 1.4068,
+      "step": 3168
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.765789572736386e-06,
+      "loss": 1.2632,
+      "step": 3169
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.759770896041319e-06,
+      "loss": 1.3172,
+      "step": 3170
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.753754091468116e-06,
+      "loss": 1.038,
+      "step": 3171
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.747739161673291e-06,
+      "loss": 1.3439,
+      "step": 3172
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.741726109312521e-06,
+      "loss": 1.0497,
+      "step": 3173
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.735714937040642e-06,
+      "loss": 1.2148,
+      "step": 3174
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.729705647511688e-06,
+      "loss": 1.1986,
+      "step": 3175
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.723698243378837e-06,
+      "loss": 1.239,
+      "step": 3176
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.717692727294453e-06,
+      "loss": 1.1884,
+      "step": 3177
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.711689101910042e-06,
+      "loss": 1.3165,
+      "step": 3178
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.7056873698763035e-06,
+      "loss": 1.1132,
+      "step": 3179
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.699687533843086e-06,
+      "loss": 1.2607,
+      "step": 3180
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6936895964593995e-06,
+      "loss": 1.4827,
+      "step": 3181
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.687693560373421e-06,
+      "loss": 1.2523,
+      "step": 3182
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.681699428232488e-06,
+      "loss": 1.0,
+      "step": 3183
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.675707202683094e-06,
+      "loss": 1.3332,
+      "step": 3184
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.669716886370893e-06,
+      "loss": 1.4018,
+      "step": 3185
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6637284819406936e-06,
+      "loss": 1.4355,
+      "step": 3186
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.657741992036463e-06,
+      "loss": 1.174,
+      "step": 3187
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.651757419301321e-06,
+      "loss": 1.2104,
+      "step": 3188
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.64577476637755e-06,
+      "loss": 1.197,
+      "step": 3189
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.639794035906567e-06,
+      "loss": 1.2991,
+      "step": 3190
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.633815230528954e-06,
+      "loss": 1.2506,
+      "step": 3191
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.627838352884435e-06,
+      "loss": 1.3153,
+      "step": 3192
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6218634056119e-06,
+      "loss": 1.1635,
+      "step": 3193
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.615890391349359e-06,
+      "loss": 1.0671,
+      "step": 3194
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.609919312733987e-06,
+      "loss": 1.2537,
+      "step": 3195
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.603950172402107e-06,
+      "loss": 1.2143,
+      "step": 3196
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5979829729891796e-06,
+      "loss": 1.2302,
+      "step": 3197
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.592017717129801e-06,
+      "loss": 1.3011,
+      "step": 3198
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.586054407457717e-06,
+      "loss": 1.2499,
+      "step": 3199
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.580093046605823e-06,
+      "loss": 1.1964,
+      "step": 3200
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.574133637206144e-06,
+      "loss": 1.1751,
+      "step": 3201
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.568176181889833e-06,
+      "loss": 1.3976,
+      "step": 3202
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.562220683287205e-06,
+      "loss": 1.2261,
+      "step": 3203
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.556267144027692e-06,
+      "loss": 1.17,
+      "step": 3204
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.550315566739872e-06,
+      "loss": 1.2595,
+      "step": 3205
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.544365954051441e-06,
+      "loss": 1.2649,
+      "step": 3206
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.538418308589246e-06,
+      "loss": 1.2313,
+      "step": 3207
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.532472632979256e-06,
+      "loss": 1.2584,
+      "step": 3208
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5265289298465744e-06,
+      "loss": 1.321,
+      "step": 3209
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.520587201815428e-06,
+      "loss": 1.1895,
+      "step": 3210
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.514647451509175e-06,
+      "loss": 1.1089,
+      "step": 3211
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.508709681550301e-06,
+      "loss": 1.2626,
+      "step": 3212
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.502773894560418e-06,
+      "loss": 1.1693,
+      "step": 3213
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.496840093160259e-06,
+      "loss": 1.218,
+      "step": 3214
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.490908279969682e-06,
+      "loss": 1.1136,
+      "step": 3215
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.484978457607664e-06,
+      "loss": 1.1618,
+      "step": 3216
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.47905062869232e-06,
+      "loss": 1.2797,
+      "step": 3217
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.473124795840856e-06,
+      "loss": 1.2204,
+      "step": 3218
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.467200961669619e-06,
+      "loss": 1.3359,
+      "step": 3219
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.461279128794061e-06,
+      "loss": 1.2823,
+      "step": 3220
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.455359299828769e-06,
+      "loss": 1.2457,
+      "step": 3221
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4494414773874205e-06,
+      "loss": 1.1819,
+      "step": 3222
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.443525664082818e-06,
+      "loss": 1.2721,
+      "step": 3223
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.437611862526887e-06,
+      "loss": 1.4169,
+      "step": 3224
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.431700075330654e-06,
+      "loss": 1.1572,
+      "step": 3225
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.425790305104251e-06,
+      "loss": 1.3091,
+      "step": 3226
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.419882554456923e-06,
+      "loss": 1.1963,
+      "step": 3227
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.4139768259970385e-06,
+      "loss": 1.1754,
+      "step": 3228
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.40807312233206e-06,
+      "loss": 1.233,
+      "step": 3229
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.402171446068545e-06,
+      "loss": 1.1716,
+      "step": 3230
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.39627179981218e-06,
+      "loss": 1.1514,
+      "step": 3231
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.390374186167739e-06,
+      "loss": 1.3846,
+      "step": 3232
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.384478607739103e-06,
+      "loss": 1.1595,
+      "step": 3233
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.378585067129256e-06,
+      "loss": 1.0379,
+      "step": 3234
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.372693566940277e-06,
+      "loss": 1.219,
+      "step": 3235
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.366804109773352e-06,
+      "loss": 1.2875,
+      "step": 3236
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.360916698228756e-06,
+      "loss": 1.3817,
+      "step": 3237
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3550313349058695e-06,
+      "loss": 1.3308,
+      "step": 3238
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.349148022403162e-06,
+      "loss": 1.0493,
+      "step": 3239
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.343266763318202e-06,
+      "loss": 1.2418,
+      "step": 3240
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.337387560247648e-06,
+      "loss": 1.2157,
+      "step": 3241
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.331510415787255e-06,
+      "loss": 1.2374,
+      "step": 3242
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.325635332531864e-06,
+      "loss": 1.342,
+      "step": 3243
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.319762313075405e-06,
+      "loss": 1.1669,
+      "step": 3244
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.313891360010912e-06,
+      "loss": 1.1894,
+      "step": 3245
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.308022475930484e-06,
+      "loss": 1.0371,
+      "step": 3246
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.302155663425319e-06,
+      "loss": 1.2563,
+      "step": 3247
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.296290925085699e-06,
+      "loss": 1.296,
+      "step": 3248
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.290428263500996e-06,
+      "loss": 1.3595,
+      "step": 3249
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.284567681259651e-06,
+      "loss": 1.1532,
+      "step": 3250
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.278709180949195e-06,
+      "loss": 1.2538,
+      "step": 3251
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2728527651562475e-06,
+      "loss": 0.9161,
+      "step": 3252
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2669984364664996e-06,
+      "loss": 0.851,
+      "step": 3253
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.261146197464711e-06,
+      "loss": 1.0538,
+      "step": 3254
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.255296050734733e-06,
+      "loss": 1.0393,
+      "step": 3255
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2494479988594936e-06,
+      "loss": 1.0657,
+      "step": 3256
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.243602044420991e-06,
+      "loss": 1.1161,
+      "step": 3257
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.237758190000295e-06,
+      "loss": 1.0682,
+      "step": 3258
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.231916438177552e-06,
+      "loss": 0.9533,
+      "step": 3259
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.226076791531979e-06,
+      "loss": 1.1286,
+      "step": 3260
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.220239252641863e-06,
+      "loss": 0.7898,
+      "step": 3261
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.214403824084562e-06,
+      "loss": 0.8499,
+      "step": 3262
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.208570508436501e-06,
+      "loss": 1.1641,
+      "step": 3263
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.202739308273172e-06,
+      "loss": 0.9969,
+      "step": 3264
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.196910226169134e-06,
+      "loss": 1.0465,
+      "step": 3265
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.191083264698011e-06,
+      "loss": 1.0357,
+      "step": 3266
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1852584264324864e-06,
+      "loss": 0.8781,
+      "step": 3267
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.179435713944314e-06,
+      "loss": 0.8952,
+      "step": 3268
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.173615129804301e-06,
+      "loss": 0.9593,
+      "step": 3269
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1677966765823226e-06,
+      "loss": 0.9196,
+      "step": 3270
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.161980356847307e-06,
+      "loss": 0.8832,
+      "step": 3271
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.156166173167237e-06,
+      "loss": 1.09,
+      "step": 3272
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.150354128109173e-06,
+      "loss": 1.0717,
+      "step": 3273
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.144544224239203e-06,
+      "loss": 0.996,
+      "step": 3274
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.138736464122485e-06,
+      "loss": 0.928,
+      "step": 3275
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.132930850323228e-06,
+      "loss": 1.1147,
+      "step": 3276
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.127127385404702e-06,
+      "loss": 0.9354,
+      "step": 3277
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.121326071929209e-06,
+      "loss": 1.0017,
+      "step": 3278
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.115526912458113e-06,
+      "loss": 0.9602,
+      "step": 3279
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.109729909551833e-06,
+      "loss": 0.9572,
+      "step": 3280
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.103935065769825e-06,
+      "loss": 1.148,
+      "step": 3281
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0981423836706e-06,
+      "loss": 1.0449,
+      "step": 3282
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.092351865811698e-06,
+      "loss": 1.1064,
+      "step": 3283
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.086563514749727e-06,
+      "loss": 0.9376,
+      "step": 3284
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.080777333040323e-06,
+      "loss": 1.0485,
+      "step": 3285
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.074993323238168e-06,
+      "loss": 0.7706,
+      "step": 3286
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0692114878969845e-06,
+      "loss": 0.888,
+      "step": 3287
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.063431829569537e-06,
+      "loss": 1.0907,
+      "step": 3288
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.057654350807628e-06,
+      "loss": 0.9053,
+      "step": 3289
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0518790541620945e-06,
+      "loss": 0.9464,
+      "step": 3290
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.046105942182815e-06,
+      "loss": 1.0732,
+      "step": 3291
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0403350174186995e-06,
+      "loss": 1.0492,
+      "step": 3292
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.034566282417694e-06,
+      "loss": 0.865,
+      "step": 3293
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.02879973972678e-06,
+      "loss": 0.952,
+      "step": 3294
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.023035391891966e-06,
+      "loss": 1.0741,
+      "step": 3295
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.017273241458296e-06,
+      "loss": 0.9639,
+      "step": 3296
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.011513290969839e-06,
+      "loss": 0.9889,
+      "step": 3297
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.005755542969699e-06,
+      "loss": 0.9278,
+      "step": 3298
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.9664,
+      "step": 3299
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9942466646019e-06,
+      "loss": 0.9461,
+      "step": 3300
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.988495539315584e-06,
+      "loss": 0.9925,
+      "step": 3301
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.982746626680247e-06,
+      "loss": 0.9908,
+      "step": 3302
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.97699992923412e-06,
+      "loss": 1.0607,
+      "step": 3303
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.971255449514447e-06,
+      "loss": 1.1335,
+      "step": 3304
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.965513190057508e-06,
+      "loss": 1.197,
+      "step": 3305
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9597731533985915e-06,
+      "loss": 1.0085,
+      "step": 3306
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.954035342071994e-06,
+      "loss": 0.9954,
+      "step": 3307
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9482997586110545e-06,
+      "loss": 1.0195,
+      "step": 3308
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.942566405548109e-06,
+      "loss": 0.9449,
+      "step": 3309
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.936835285414522e-06,
+      "loss": 0.8665,
+      "step": 3310
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9311064007406494e-06,
+      "loss": 0.9855,
+      "step": 3311
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.925379754055891e-06,
+      "loss": 0.8872,
+      "step": 3312
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.919655347888638e-06,
+      "loss": 0.967,
+      "step": 3313
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.913933184766298e-06,
+      "loss": 0.9564,
+      "step": 3314
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.908213267215287e-06,
+      "loss": 0.9667,
+      "step": 3315
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.902495597761032e-06,
+      "loss": 0.9116,
+      "step": 3316
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.896780178927965e-06,
+      "loss": 0.9577,
+      "step": 3317
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.891067013239526e-06,
+      "loss": 1.0256,
+      "step": 3318
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.885356103218159e-06,
+      "loss": 1.1266,
+      "step": 3319
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.879647451385311e-06,
+      "loss": 1.065,
+      "step": 3320
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.873941060261437e-06,
+      "loss": 0.9753,
+      "step": 3321
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.868236932365988e-06,
+      "loss": 0.9313,
+      "step": 3322
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.862535070217416e-06,
+      "loss": 1.0341,
+      "step": 3323
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.856835476333181e-06,
+      "loss": 1.1211,
+      "step": 3324
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8511381532297296e-06,
+      "loss": 1.0878,
+      "step": 3325
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.845443103422513e-06,
+      "loss": 1.1223,
+      "step": 3326
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.839750329425979e-06,
+      "loss": 1.0205,
+      "step": 3327
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.834059833753565e-06,
+      "loss": 1.0491,
+      "step": 3328
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.828371618917711e-06,
+      "loss": 0.8911,
+      "step": 3329
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.822685687429848e-06,
+      "loss": 1.07,
+      "step": 3330
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8170020418003884e-06,
+      "loss": 0.9286,
+      "step": 3331
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.811320684538741e-06,
+      "loss": 0.9824,
+      "step": 3332
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.805641618153316e-06,
+      "loss": 0.9689,
+      "step": 3333
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.799964845151501e-06,
+      "loss": 1.184,
+      "step": 3334
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.794290368039664e-06,
+      "loss": 1.0928,
+      "step": 3335
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.788618189323176e-06,
+      "loss": 0.8862,
+      "step": 3336
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.782948311506383e-06,
+      "loss": 0.942,
+      "step": 3337
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.777280737092621e-06,
+      "loss": 1.1321,
+      "step": 3338
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.771615468584194e-06,
+      "loss": 0.9402,
+      "step": 3339
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.76595250848241e-06,
+      "loss": 0.9658,
+      "step": 3340
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.760291859287543e-06,
+      "loss": 0.9864,
+      "step": 3341
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7546335234988515e-06,
+      "loss": 0.9086,
+      "step": 3342
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7489775036145715e-06,
+      "loss": 0.9073,
+      "step": 3343
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.743323802131916e-06,
+      "loss": 1.089,
+      "step": 3344
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.737672421547076e-06,
+      "loss": 1.0275,
+      "step": 3345
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.732023364355217e-06,
+      "loss": 0.9789,
+      "step": 3346
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.726376633050479e-06,
+      "loss": 0.9247,
+      "step": 3347
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.720732230125973e-06,
+      "loss": 1.0137,
+      "step": 3348
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.715090158073785e-06,
+      "loss": 1.0454,
+      "step": 3349
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.709450419384971e-06,
+      "loss": 0.8735,
+      "step": 3350
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.703813016549555e-06,
+      "loss": 0.8885,
+      "step": 3351
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.698177952056533e-06,
+      "loss": 0.9174,
+      "step": 3352
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.692545228393861e-06,
+      "loss": 0.9713,
+      "step": 3353
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.6869148480484795e-06,
+      "loss": 0.9223,
+      "step": 3354
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.681286813506271e-06,
+      "loss": 1.0787,
+      "step": 3355
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.675661127252092e-06,
+      "loss": 0.7894,
+      "step": 3356
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.670037791769772e-06,
+      "loss": 1.131,
+      "step": 3357
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.664416809542095e-06,
+      "loss": 1.0638,
+      "step": 3358
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.658798183050796e-06,
+      "loss": 0.9831,
+      "step": 3359
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.65318191477658e-06,
+      "loss": 1.0159,
+      "step": 3360
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.647568007199117e-06,
+      "loss": 1.0116,
+      "step": 3361
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.641956462797027e-06,
+      "loss": 0.8812,
+      "step": 3362
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.636347284047878e-06,
+      "loss": 1.0684,
+      "step": 3363
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.630740473428212e-06,
+      "loss": 1.0431,
+      "step": 3364
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.625136033413512e-06,
+      "loss": 0.7739,
+      "step": 3365
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.619533966478225e-06,
+      "loss": 1.0195,
+      "step": 3366
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.613934275095729e-06,
+      "loss": 1.0277,
+      "step": 3367
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.608336961738383e-06,
+      "loss": 0.9561,
+      "step": 3368
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.602742028877475e-06,
+      "loss": 0.9171,
+      "step": 3369
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.597149478983247e-06,
+      "loss": 0.8587,
+      "step": 3370
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.5915593145248926e-06,
+      "loss": 0.8524,
+      "step": 3371
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5859715379705475e-06,
+      "loss": 0.9913,
+      "step": 3372
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.580386151787295e-06,
+      "loss": 1.0046,
+      "step": 3373
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.574803158441165e-06,
+      "loss": 1.0345,
+      "step": 3374
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.569222560397126e-06,
+      "loss": 0.8621,
+      "step": 3375
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5636443601190935e-06,
+      "loss": 0.9974,
+      "step": 3376
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.558068560069921e-06,
+      "loss": 1.0008,
+      "step": 3377
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.552495162711412e-06,
+      "loss": 1.1245,
+      "step": 3378
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.546924170504292e-06,
+      "loss": 0.8608,
+      "step": 3379
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.541355585908237e-06,
+      "loss": 0.9928,
+      "step": 3380
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.535789411381852e-06,
+      "loss": 1.2997,
+      "step": 3381
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.530225649382697e-06,
+      "loss": 1.0933,
+      "step": 3382
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.524664302367239e-06,
+      "loss": 1.0417,
+      "step": 3383
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.519105372790892e-06,
+      "loss": 0.8694,
+      "step": 3384
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.513548863108012e-06,
+      "loss": 0.8842,
+      "step": 3385
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.507994775771877e-06,
+      "loss": 0.9466,
+      "step": 3386
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.502443113234688e-06,
+      "loss": 1.1095,
+      "step": 3387
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.496893877947583e-06,
+      "loss": 0.8841,
+      "step": 3388
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.49134707236064e-06,
+      "loss": 1.0326,
+      "step": 3389
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.485802698922849e-06,
+      "loss": 1.1392,
+      "step": 3390
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.48026076008212e-06,
+      "loss": 0.9866,
+      "step": 3391
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.474721258285311e-06,
+      "loss": 0.8819,
+      "step": 3392
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4691841959781865e-06,
+      "loss": 1.0311,
+      "step": 3393
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.463649575605442e-06,
+      "loss": 0.8271,
+      "step": 3394
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.458117399610682e-06,
+      "loss": 1.1958,
+      "step": 3395
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.452587670436451e-06,
+      "loss": 1.0464,
+      "step": 3396
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4470603905242014e-06,
+      "loss": 1.0885,
+      "step": 3397
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.441535562314307e-06,
+      "loss": 1.0131,
+      "step": 3398
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.436013188246056e-06,
+      "loss": 1.0085,
+      "step": 3399
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.430493270757656e-06,
+      "loss": 0.9509,
+      "step": 3400
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.424975812286228e-06,
+      "loss": 0.9623,
+      "step": 3401
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.419460815267818e-06,
+      "loss": 0.9393,
+      "step": 3402
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4139482821373665e-06,
+      "loss": 0.9649,
+      "step": 3403
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.408438215328739e-06,
+      "loss": 0.993,
+      "step": 3404
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.402930617274704e-06,
+      "loss": 0.9835,
+      "step": 3405
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.397425490406959e-06,
+      "loss": 0.8205,
+      "step": 3406
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.391922837156082e-06,
+      "loss": 1.0003,
+      "step": 3407
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.386422659951575e-06,
+      "loss": 0.9784,
+      "step": 3408
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3809249612218516e-06,
+      "loss": 1.0068,
+      "step": 3409
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.375429743394226e-06,
+      "loss": 0.8003,
+      "step": 3410
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.369937008894906e-06,
+      "loss": 0.9396,
+      "step": 3411
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.364446760149014e-06,
+      "loss": 0.9901,
+      "step": 3412
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.35895899958058e-06,
+      "loss": 1.0225,
+      "step": 3413
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.353473729612531e-06,
+      "loss": 0.8475,
+      "step": 3414
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3479909526666775e-06,
+      "loss": 0.9508,
+      "step": 3415
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3425106711637575e-06,
+      "loss": 0.9584,
+      "step": 3416
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.33703288752339e-06,
+      "loss": 1.0369,
+      "step": 3417
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.331557604164098e-06,
+      "loss": 1.1033,
+      "step": 3418
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.326084823503287e-06,
+      "loss": 1.1843,
+      "step": 3419
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.320614547957278e-06,
+      "loss": 0.8953,
+      "step": 3420
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.315146779941274e-06,
+      "loss": 1.0242,
+      "step": 3421
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.309681521869371e-06,
+      "loss": 1.1448,
+      "step": 3422
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.30421877615456e-06,
+      "loss": 0.9629,
+      "step": 3423
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.298758545208722e-06,
+      "loss": 0.872,
+      "step": 3424
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.293300831442625e-06,
+      "loss": 0.8683,
+      "step": 3425
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2878456372659295e-06,
+      "loss": 1.0688,
+      "step": 3426
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.282392965087182e-06,
+      "loss": 1.0599,
+      "step": 3427
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.276942817313814e-06,
+      "loss": 1.1563,
+      "step": 3428
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.271495196352141e-06,
+      "loss": 0.7655,
+      "step": 3429
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.266050104607377e-06,
+      "loss": 0.931,
+      "step": 3430
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.260607544483596e-06,
+      "loss": 0.9245,
+      "step": 3431
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.255167518383771e-06,
+      "loss": 0.8281,
+      "step": 3432
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.249730028709746e-06,
+      "loss": 1.1185,
+      "step": 3433
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.244295077862263e-06,
+      "loss": 1.0229,
+      "step": 3434
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.23886266824092e-06,
+      "loss": 0.734,
+      "step": 3435
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.233432802244203e-06,
+      "loss": 0.8341,
+      "step": 3436
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.228005482269481e-06,
+      "loss": 1.0583,
+      "step": 3437
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.222580710712998e-06,
+      "loss": 0.9825,
+      "step": 3438
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.217158489969857e-06,
+      "loss": 1.0008,
+      "step": 3439
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.211738822434047e-06,
+      "loss": 0.8498,
+      "step": 3440
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.2063217104984374e-06,
+      "loss": 1.0138,
+      "step": 3441
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.20090715655476e-06,
+      "loss": 0.9713,
+      "step": 3442
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.195495162993607e-06,
+      "loss": 0.8748,
+      "step": 3443
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.190085732204462e-06,
+      "loss": 0.8952,
+      "step": 3444
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.184678866575661e-06,
+      "loss": 0.9506,
+      "step": 3445
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.179274568494416e-06,
+      "loss": 0.954,
+      "step": 3446
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.173872840346798e-06,
+      "loss": 1.0151,
+      "step": 3447
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1684736845177496e-06,
+      "loss": 1.1186,
+      "step": 3448
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1630771033910764e-06,
+      "loss": 0.9234,
+      "step": 3449
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.157683099349444e-06,
+      "loss": 1.0104,
+      "step": 3450
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.152291674774384e-06,
+      "loss": 0.8385,
+      "step": 3451
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1469028320462854e-06,
+      "loss": 0.9853,
+      "step": 3452
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1415165735444016e-06,
+      "loss": 1.0992,
+      "step": 3453
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.136132901646843e-06,
+      "loss": 0.8536,
+      "step": 3454
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.130751818730578e-06,
+      "loss": 0.9419,
+      "step": 3455
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.125373327171428e-06,
+      "loss": 1.1142,
+      "step": 3456
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.119997429344074e-06,
+      "loss": 1.0436,
+      "step": 3457
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.114624127622063e-06,
+      "loss": 1.1142,
+      "step": 3458
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.109253424377773e-06,
+      "loss": 1.0186,
+      "step": 3459
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.103885321982448e-06,
+      "loss": 1.0818,
+      "step": 3460
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.098519822806182e-06,
+      "loss": 1.0088,
+      "step": 3461
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.093156929217929e-06,
+      "loss": 1.1876,
+      "step": 3462
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.08779664358547e-06,
+      "loss": 1.0758,
+      "step": 3463
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.082438968275452e-06,
+      "loss": 0.9793,
+      "step": 3464
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.07708390565337e-06,
+      "loss": 0.9982,
+      "step": 3465
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.071731458083562e-06,
+      "loss": 1.0855,
+      "step": 3466
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.066381627929202e-06,
+      "loss": 1.0347,
+      "step": 3467
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0610344175523175e-06,
+      "loss": 1.1204,
+      "step": 3468
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0556898293137845e-06,
+      "loss": 1.0003,
+      "step": 3469
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0503478655733146e-06,
+      "loss": 1.0636,
+      "step": 3470
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.045008528689457e-06,
+      "loss": 0.9801,
+      "step": 3471
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.039671821019609e-06,
+      "loss": 0.8997,
+      "step": 3472
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.034337744920002e-06,
+      "loss": 0.7505,
+      "step": 3473
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.029006302745706e-06,
+      "loss": 0.8971,
+      "step": 3474
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.02367749685063e-06,
+      "loss": 0.9301,
+      "step": 3475
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0183513295875184e-06,
+      "loss": 0.9581,
+      "step": 3476
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.01302780330795e-06,
+      "loss": 0.9195,
+      "step": 3477
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.00770692036234e-06,
+      "loss": 1.1516,
+      "step": 3478
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.00238868309993e-06,
+      "loss": 0.913,
+      "step": 3479
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.997073093868802e-06,
+      "loss": 1.0942,
+      "step": 3480
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9917601550158616e-06,
+      "loss": 1.0957,
+      "step": 3481
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.986449868886849e-06,
+      "loss": 0.8364,
+      "step": 3482
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.981142237826332e-06,
+      "loss": 0.9722,
+      "step": 3483
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.975837264177706e-06,
+      "loss": 1.0155,
+      "step": 3484
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.970534950283187e-06,
+      "loss": 0.8499,
+      "step": 3485
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.965235298483835e-06,
+      "loss": 0.992,
+      "step": 3486
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.959938311119511e-06,
+      "loss": 0.9626,
+      "step": 3487
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.954643990528914e-06,
+      "loss": 1.104,
+      "step": 3488
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.949352339049561e-06,
+      "loss": 1.0452,
+      "step": 3489
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.944063359017799e-06,
+      "loss": 0.9452,
+      "step": 3490
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9387770527687795e-06,
+      "loss": 1.0838,
+      "step": 3491
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.933493422636483e-06,
+      "loss": 0.9741,
+      "step": 3492
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.928212470953714e-06,
+      "loss": 1.0206,
+      "step": 3493
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9229342000520855e-06,
+      "loss": 1.1177,
+      "step": 3494
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.917658612262033e-06,
+      "loss": 0.8711,
+      "step": 3495
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.912385709912794e-06,
+      "loss": 1.0639,
+      "step": 3496
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.907115495332439e-06,
+      "loss": 0.9186,
+      "step": 3497
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.901847970847842e-06,
+      "loss": 1.076,
+      "step": 3498
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.8965831387846885e-06,
+      "loss": 0.9547,
+      "step": 3499
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.891321001467478e-06,
+      "loss": 1.0037,
+      "step": 3500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.886061561219521e-06,
+      "loss": 1.1238,
+      "step": 3501
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.880804820362934e-06,
+      "loss": 1.008,
+      "step": 3502
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.875550781218644e-06,
+      "loss": 1.0024,
+      "step": 3503
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.870299446106382e-06,
+      "loss": 0.9925,
+      "step": 3504
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.865050817344692e-06,
+      "loss": 0.9621,
+      "step": 3505
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.859804897250918e-06,
+      "loss": 1.11,
+      "step": 3506
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.854561688141205e-06,
+      "loss": 1.1276,
+      "step": 3507
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8493211923305075e-06,
+      "loss": 1.1517,
+      "step": 3508
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.844083412132581e-06,
+      "loss": 1.0497,
+      "step": 3509
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.838848349859977e-06,
+      "loss": 1.0267,
+      "step": 3510
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.833616007824054e-06,
+      "loss": 0.8075,
+      "step": 3511
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.828386388334962e-06,
+      "loss": 0.951,
+      "step": 3512
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.823159493701654e-06,
+      "loss": 0.8731,
+      "step": 3513
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.817935326231886e-06,
+      "loss": 0.881,
+      "step": 3514
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8127138882321937e-06,
+      "loss": 0.9294,
+      "step": 3515
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8074951820079186e-06,
+      "loss": 0.9559,
+      "step": 3516
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.802279209863192e-06,
+      "loss": 1.0293,
+      "step": 3517
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.7970659741009473e-06,
+      "loss": 1.0714,
+      "step": 3518
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.791855477022903e-06,
+      "loss": 0.8714,
+      "step": 3519
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.786647720929557e-06,
+      "loss": 0.8455,
+      "step": 3520
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.781442708120219e-06,
+      "loss": 0.8754,
+      "step": 3521
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.776240440892973e-06,
+      "loss": 1.0903,
+      "step": 3522
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7710409215446986e-06,
+      "loss": 0.812,
+      "step": 3523
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7658441523710467e-06,
+      "loss": 1.0033,
+      "step": 3524
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.760650135666476e-06,
+      "loss": 0.9374,
+      "step": 3525
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.755458873724216e-06,
+      "loss": 0.9789,
+      "step": 3526
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.750270368836283e-06,
+      "loss": 0.8856,
+      "step": 3527
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7450846232934768e-06,
+      "loss": 0.9968,
+      "step": 3528
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7399016393853772e-06,
+      "loss": 0.9134,
+      "step": 3529
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7347214194003466e-06,
+      "loss": 0.9217,
+      "step": 3530
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.729543965625526e-06,
+      "loss": 0.9855,
+      "step": 3531
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7243692803468357e-06,
+      "loss": 0.9235,
+      "step": 3532
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7191973658489735e-06,
+      "loss": 0.9056,
+      "step": 3533
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7140282244154136e-06,
+      "loss": 1.0407,
+      "step": 3534
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7088618583284054e-06,
+      "loss": 1.0887,
+      "step": 3535
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7036982698689726e-06,
+      "loss": 1.0543,
+      "step": 3536
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6985374613169166e-06,
+      "loss": 0.9258,
+      "step": 3537
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6933794349508057e-06,
+      "loss": 0.9217,
+      "step": 3538
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6882241930479824e-06,
+      "loss": 1.1429,
+      "step": 3539
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6830717378845614e-06,
+      "loss": 1.079,
+      "step": 3540
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6779220717354204e-06,
+      "loss": 0.9603,
+      "step": 3541
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6727751968742185e-06,
+      "loss": 0.9999,
+      "step": 3542
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6676311155733745e-06,
+      "loss": 0.9345,
+      "step": 3543
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.662489830104068e-06,
+      "loss": 1.1551,
+      "step": 3544
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6573513427362485e-06,
+      "loss": 0.7792,
+      "step": 3545
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.652215655738639e-06,
+      "loss": 0.9083,
+      "step": 3546
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6470827713787195e-06,
+      "loss": 1.2325,
+      "step": 3547
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.641952691922721e-06,
+      "loss": 0.9444,
+      "step": 3548
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6368254196356576e-06,
+      "loss": 1.0492,
+      "step": 3549
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.63170095678129e-06,
+      "loss": 1.0706,
+      "step": 3550
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6265793056221465e-06,
+      "loss": 0.7473,
+      "step": 3551
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6214604684194974e-06,
+      "loss": 0.8884,
+      "step": 3552
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6163444474333933e-06,
+      "loss": 1.0051,
+      "step": 3553
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.611231244922628e-06,
+      "loss": 0.9642,
+      "step": 3554
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.606120863144753e-06,
+      "loss": 0.9344,
+      "step": 3555
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6010133043560748e-06,
+      "loss": 1.0784,
+      "step": 3556
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5959085708116546e-06,
+      "loss": 0.8244,
+      "step": 3557
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.590806664765305e-06,
+      "loss": 1.0607,
+      "step": 3558
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5857075884695915e-06,
+      "loss": 1.0647,
+      "step": 3559
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5806113441758272e-06,
+      "loss": 1.0923,
+      "step": 3560
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5755179341340772e-06,
+      "loss": 0.9879,
+      "step": 3561
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.570427360593157e-06,
+      "loss": 0.976,
+      "step": 3562
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5653396258006266e-06,
+      "loss": 0.9852,
+      "step": 3563
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5602547320027937e-06,
+      "loss": 1.1262,
+      "step": 3564
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.555172681444712e-06,
+      "loss": 1.0315,
+      "step": 3565
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5500934763701743e-06,
+      "loss": 1.1755,
+      "step": 3566
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5450171190217364e-06,
+      "loss": 1.0461,
+      "step": 3567
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.539943611640669e-06,
+      "loss": 1.0957,
+      "step": 3568
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5348729564669993e-06,
+      "loss": 0.9608,
+      "step": 3569
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.529805155739502e-06,
+      "loss": 0.9441,
+      "step": 3570
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.524740211695683e-06,
+      "loss": 0.9628,
+      "step": 3571
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5196781265717807e-06,
+      "loss": 0.8233,
+      "step": 3572
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5146189026027787e-06,
+      "loss": 0.9675,
+      "step": 3573
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.509562542022402e-06,
+      "loss": 1.0062,
+      "step": 3574
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.504509047063108e-06,
+      "loss": 0.9693,
+      "step": 3575
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4994584199560767e-06,
+      "loss": 0.9216,
+      "step": 3576
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4944106629312423e-06,
+      "loss": 0.9803,
+      "step": 3577
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4893657782172573e-06,
+      "loss": 1.0229,
+      "step": 3578
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4843237680415153e-06,
+      "loss": 0.8238,
+      "step": 3579
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.479284634630126e-06,
+      "loss": 0.9686,
+      "step": 3580
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4742483802079473e-06,
+      "loss": 0.9474,
+      "step": 3581
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4692150069985563e-06,
+      "loss": 0.9415,
+      "step": 3582
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4641845172242573e-06,
+      "loss": 0.8589,
+      "step": 3583
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4591569131060855e-06,
+      "loss": 0.8339,
+      "step": 3584
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4541321968637995e-06,
+      "loss": 1.0447,
+      "step": 3585
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4491103707158833e-06,
+      "loss": 0.9085,
+      "step": 3586
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.444091436879545e-06,
+      "loss": 0.9245,
+      "step": 3587
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.439075397570716e-06,
+      "loss": 0.9171,
+      "step": 3588
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.434062255004049e-06,
+      "loss": 0.8951,
+      "step": 3589
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.429052011392916e-06,
+      "loss": 1.0184,
+      "step": 3590
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.42404466894942e-06,
+      "loss": 1.0324,
+      "step": 3591
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4190402298843637e-06,
+      "loss": 0.924,
+      "step": 3592
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.414038696407285e-06,
+      "loss": 0.9288,
+      "step": 3593
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.409040070726427e-06,
+      "loss": 0.8701,
+      "step": 3594
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4040443550487645e-06,
+      "loss": 0.9152,
+      "step": 3595
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3990515515799684e-06,
+      "loss": 1.0961,
+      "step": 3596
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3940616625244314e-06,
+      "loss": 1.0708,
+      "step": 3597
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.389074690085269e-06,
+      "loss": 1.099,
+      "step": 3598
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.384090636464301e-06,
+      "loss": 0.979,
+      "step": 3599
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3791095038620515e-06,
+      "loss": 1.0875,
+      "step": 3600
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3741312944777616e-06,
+      "loss": 0.9996,
+      "step": 3601
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.369156010509389e-06,
+      "loss": 0.7845,
+      "step": 3602
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.364183654153592e-06,
+      "loss": 1.1196,
+      "step": 3603
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.359214227605728e-06,
+      "loss": 1.0802,
+      "step": 3604
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3542477330598812e-06,
+      "loss": 0.8263,
+      "step": 3605
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.349284172708823e-06,
+      "loss": 1.053,
+      "step": 3606
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.344323548744044e-06,
+      "loss": 0.9366,
+      "step": 3607
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.339365863355718e-06,
+      "loss": 0.9324,
+      "step": 3608
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.334411118732744e-06,
+      "loss": 1.0714,
+      "step": 3609
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.329459317062712e-06,
+      "loss": 0.8828,
+      "step": 3610
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.32451046053191e-06,
+      "loss": 0.9621,
+      "step": 3611
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3195645513253316e-06,
+      "loss": 0.9209,
+      "step": 3612
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3146215916266643e-06,
+      "loss": 0.9183,
+      "step": 3613
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.309681583618298e-06,
+      "loss": 1.025,
+      "step": 3614
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.304744529481315e-06,
+      "loss": 0.9182,
+      "step": 3615
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.299810431395497e-06,
+      "loss": 1.0946,
+      "step": 3616
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.294879291539317e-06,
+      "loss": 0.9303,
+      "step": 3617
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.289951112089943e-06,
+      "loss": 1.1704,
+      "step": 3618
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.285025895223244e-06,
+      "loss": 1.0861,
+      "step": 3619
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2801036431137658e-06,
+      "loss": 1.0486,
+      "step": 3620
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2751843579347562e-06,
+      "loss": 1.049,
+      "step": 3621
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2702680418581467e-06,
+      "loss": 0.8581,
+      "step": 3622
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.26535469705457e-06,
+      "loss": 0.9607,
+      "step": 3623
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2604443256933307e-06,
+      "loss": 0.874,
+      "step": 3624
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2555369299424254e-06,
+      "loss": 1.0199,
+      "step": 3625
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2506325119685476e-06,
+      "loss": 1.0449,
+      "step": 3626
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2457310739370684e-06,
+      "loss": 0.9583,
+      "step": 3627
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2408326180120354e-06,
+      "loss": 1.0269,
+      "step": 3628
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.235937146356187e-06,
+      "loss": 0.9634,
+      "step": 3629
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.231044661130951e-06,
+      "loss": 1.0424,
+      "step": 3630
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2261551644964305e-06,
+      "loss": 0.8826,
+      "step": 3631
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2212686586113973e-06,
+      "loss": 0.9147,
+      "step": 3632
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.216385145633324e-06,
+      "loss": 1.0248,
+      "step": 3633
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.211504627718347e-06,
+      "loss": 0.9351,
+      "step": 3634
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2066271070212873e-06,
+      "loss": 1.0996,
+      "step": 3635
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.201752585695638e-06,
+      "loss": 1.1362,
+      "step": 3636
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.196881065893571e-06,
+      "loss": 0.9494,
+      "step": 3637
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1920125497659327e-06,
+      "loss": 1.0232,
+      "step": 3638
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1871470394622407e-06,
+      "loss": 0.7767,
+      "step": 3639
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1822845371306897e-06,
+      "loss": 1.0698,
+      "step": 3640
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1774250449181423e-06,
+      "loss": 0.8812,
+      "step": 3641
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1725685649701354e-06,
+      "loss": 0.956,
+      "step": 3642
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.167715099430874e-06,
+      "loss": 1.0581,
+      "step": 3643
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1628646504432304e-06,
+      "loss": 1.1455,
+      "step": 3644
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.158017220148751e-06,
+      "loss": 0.9632,
+      "step": 3645
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1531728106876404e-06,
+      "loss": 0.9136,
+      "step": 3646
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.148331424198784e-06,
+      "loss": 1.016,
+      "step": 3647
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1434930628197145e-06,
+      "loss": 0.8515,
+      "step": 3648
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1386577286866414e-06,
+      "loss": 0.9426,
+      "step": 3649
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.13382542393443e-06,
+      "loss": 1.068,
+      "step": 3650
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1289961506966217e-06,
+      "loss": 1.0296,
+      "step": 3651
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1241699111054004e-06,
+      "loss": 0.9646,
+      "step": 3652
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1193467072916195e-06,
+      "loss": 0.9657,
+      "step": 3653
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.114526541384799e-06,
+      "loss": 1.0597,
+      "step": 3654
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1097094155131122e-06,
+      "loss": 1.0386,
+      "step": 3655
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1048953318033826e-06,
+      "loss": 1.0272,
+      "step": 3656
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.100084292381095e-06,
+      "loss": 1.1061,
+      "step": 3657
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.095276299370401e-06,
+      "loss": 0.9187,
+      "step": 3658
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0904713548940936e-06,
+      "loss": 0.9388,
+      "step": 3659
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.085669461073626e-06,
+      "loss": 0.9581,
+      "step": 3660
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0808706200291017e-06,
+      "loss": 0.8793,
+      "step": 3661
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.076074833879279e-06,
+      "loss": 1.2059,
+      "step": 3662
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0712821047415655e-06,
+      "loss": 1.2101,
+      "step": 3663
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0664924347320203e-06,
+      "loss": 0.9676,
+      "step": 3664
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.06170582596535e-06,
+      "loss": 1.047,
+      "step": 3665
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0569222805549114e-06,
+      "loss": 1.0228,
+      "step": 3666
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.052141800612709e-06,
+      "loss": 1.0701,
+      "step": 3667
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.047364388249393e-06,
+      "loss": 0.9098,
+      "step": 3668
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0425900455742584e-06,
+      "loss": 1.0912,
+      "step": 3669
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.037818774695247e-06,
+      "loss": 0.9133,
+      "step": 3670
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0330505777189433e-06,
+      "loss": 1.0384,
+      "step": 3671
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0282854567505737e-06,
+      "loss": 0.9207,
+      "step": 3672
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0235234138940073e-06,
+      "loss": 1.021,
+      "step": 3673
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0187644512517523e-06,
+      "loss": 1.0983,
+      "step": 3674
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0140085709249666e-06,
+      "loss": 0.966,
+      "step": 3675
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.009255775013431e-06,
+      "loss": 0.904,
+      "step": 3676
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0045060656155766e-06,
+      "loss": 1.217,
+      "step": 3677
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9997594448284638e-06,
+      "loss": 1.1656,
+      "step": 3678
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9950159147478043e-06,
+      "loss": 1.0678,
+      "step": 3679
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9902754774679233e-06,
+      "loss": 1.0765,
+      "step": 3680
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9855381350817937e-06,
+      "loss": 1.0207,
+      "step": 3681
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9808038896810254e-06,
+      "loss": 0.9889,
+      "step": 3682
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.976072743355852e-06,
+      "loss": 1.0459,
+      "step": 3683
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9713446981951466e-06,
+      "loss": 1.075,
+      "step": 3684
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9666197562863973e-06,
+      "loss": 1.0099,
+      "step": 3685
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.961897919715746e-06,
+      "loss": 1.061,
+      "step": 3686
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.957179190567947e-06,
+      "loss": 1.0937,
+      "step": 3687
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.952463570926385e-06,
+      "loss": 0.9028,
+      "step": 3688
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9477510628730755e-06,
+      "loss": 0.9454,
+      "step": 3689
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9430416684886575e-06,
+      "loss": 1.0567,
+      "step": 3690
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.938335389852397e-06,
+      "loss": 1.0149,
+      "step": 3691
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9336322290421815e-06,
+      "loss": 1.0919,
+      "step": 3692
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9289321881345257e-06,
+      "loss": 1.0422,
+      "step": 3693
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9242352692045638e-06,
+      "loss": 0.951,
+      "step": 3694
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9195414743260544e-06,
+      "loss": 1.0659,
+      "step": 3695
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9148508055713733e-06,
+      "loss": 0.988,
+      "step": 3696
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9101632650115185e-06,
+      "loss": 0.8791,
+      "step": 3697
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9054788547161063e-06,
+      "loss": 0.9012,
+      "step": 3698
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9007975767533714e-06,
+      "loss": 0.9114,
+      "step": 3699
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.896119433190164e-06,
+      "loss": 0.9758,
+      "step": 3700
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8914444260919526e-06,
+      "loss": 0.8758,
+      "step": 3701
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8867725575228166e-06,
+      "loss": 0.957,
+      "step": 3702
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.882103829545462e-06,
+      "loss": 0.8783,
+      "step": 3703
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.877438244221189e-06,
+      "loss": 0.9847,
+      "step": 3704
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.872775803609924e-06,
+      "loss": 1.0128,
+      "step": 3705
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.868116509770198e-06,
+      "loss": 1.0094,
+      "step": 3706
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.863460364759163e-06,
+      "loss": 0.9619,
+      "step": 3707
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8588073706325726e-06,
+      "loss": 0.9974,
+      "step": 3708
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.854157529444781e-06,
+      "loss": 1.0977,
+      "step": 3709
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8495108432487707e-06,
+      "loss": 1.0796,
+      "step": 3710
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.844867314096115e-06,
+      "loss": 0.9557,
+      "step": 3711
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8402269440370025e-06,
+      "loss": 1.0803,
+      "step": 3712
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.835589735120213e-06,
+      "loss": 1.1529,
+      "step": 3713
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8309556893931522e-06,
+      "loss": 1.0368,
+      "step": 3714
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8263248089018116e-06,
+      "loss": 0.9608,
+      "step": 3715
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.821697095690793e-06,
+      "loss": 1.0513,
+      "step": 3716
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8170725518032983e-06,
+      "loss": 1.1668,
+      "step": 3717
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.812451179281129e-06,
+      "loss": 1.234,
+      "step": 3718
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8078329801646876e-06,
+      "loss": 0.9195,
+      "step": 3719
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.803217956492974e-06,
+      "loss": 0.9965,
+      "step": 3720
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.798606110303589e-06,
+      "loss": 0.9181,
+      "step": 3721
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7939974436327277e-06,
+      "loss": 0.9537,
+      "step": 3722
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.789391958515183e-06,
+      "loss": 1.0546,
+      "step": 3723
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7847896569843422e-06,
+      "loss": 0.9354,
+      "step": 3724
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7801905410721875e-06,
+      "loss": 0.8354,
+      "step": 3725
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7755946128092935e-06,
+      "loss": 1.0038,
+      "step": 3726
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.77100187422483e-06,
+      "loss": 0.9564,
+      "step": 3727
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.766412327346556e-06,
+      "loss": 1.1745,
+      "step": 3728
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7618259742008226e-06,
+      "loss": 0.8542,
+      "step": 3729
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7572428168125654e-06,
+      "loss": 1.1635,
+      "step": 3730
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7526628572053227e-06,
+      "loss": 0.9233,
+      "step": 3731
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.74808609740121e-06,
+      "loss": 0.9554,
+      "step": 3732
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7435125394209283e-06,
+      "loss": 0.9826,
+      "step": 3733
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7389421852837685e-06,
+      "loss": 0.8895,
+      "step": 3734
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7343750370076127e-06,
+      "loss": 1.0891,
+      "step": 3735
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.729811096608922e-06,
+      "loss": 0.9837,
+      "step": 3736
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.725250366102733e-06,
+      "loss": 0.9669,
+      "step": 3737
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7206928475026828e-06,
+      "loss": 0.858,
+      "step": 3738
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7161385428209773e-06,
+      "loss": 1.1278,
+      "step": 3739
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7115874540684106e-06,
+      "loss": 0.9512,
+      "step": 3740
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.707039583254346e-06,
+      "loss": 1.1283,
+      "step": 3741
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7024949323867398e-06,
+      "loss": 1.0648,
+      "step": 3742
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6979535034721195e-06,
+      "loss": 1.0114,
+      "step": 3743
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6934152985155904e-06,
+      "loss": 1.0441,
+      "step": 3744
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6888803195208355e-06,
+      "loss": 0.9267,
+      "step": 3745
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6843485684901118e-06,
+      "loss": 0.9607,
+      "step": 3746
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.679820047424253e-06,
+      "loss": 1.0598,
+      "step": 3747
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6752947583226665e-06,
+      "loss": 1.0385,
+      "step": 3748
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6707727031833306e-06,
+      "loss": 1.2018,
+      "step": 3749
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6662538840027984e-06,
+      "loss": 0.9807,
+      "step": 3750
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6617383027761944e-06,
+      "loss": 0.9895,
+      "step": 3751
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6572259614972107e-06,
+      "loss": 1.0362,
+      "step": 3752
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6527168621581114e-06,
+      "loss": 1.046,
+      "step": 3753
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.648211006749728e-06,
+      "loss": 1.0262,
+      "step": 3754
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6437083972614575e-06,
+      "loss": 1.1114,
+      "step": 3755
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.639209035681276e-06,
+      "loss": 1.0259,
+      "step": 3756
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6347129239957058e-06,
+      "loss": 0.8619,
+      "step": 3757
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6302200641898444e-06,
+      "loss": 0.9757,
+      "step": 3758
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.625730458247362e-06,
+      "loss": 1.1311,
+      "step": 3759
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6212441081504812e-06,
+      "loss": 1.0557,
+      "step": 3760
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.616761015879986e-06,
+      "loss": 1.0417,
+      "step": 3761
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6122811834152253e-06,
+      "loss": 1.0759,
+      "step": 3762
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6078046127341138e-06,
+      "loss": 0.9405,
+      "step": 3763
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6033313058131238e-06,
+      "loss": 0.9859,
+      "step": 3764
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5988612646272735e-06,
+      "loss": 1.1147,
+      "step": 3765
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5943944911501596e-06,
+      "loss": 1.0894,
+      "step": 3766
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5899309873539245e-06,
+      "loss": 0.9627,
+      "step": 3767
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5854707552092705e-06,
+      "loss": 1.183,
+      "step": 3768
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5810137966854466e-06,
+      "loss": 0.8639,
+      "step": 3769
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.576560113750272e-06,
+      "loss": 0.7807,
+      "step": 3770
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5721097083701085e-06,
+      "loss": 1.0571,
+      "step": 3771
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.567662582509873e-06,
+      "loss": 0.8811,
+      "step": 3772
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5632187381330366e-06,
+      "loss": 0.9119,
+      "step": 3773
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5587781772016196e-06,
+      "loss": 1.0976,
+      "step": 3774
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5543409016761935e-06,
+      "loss": 0.8718,
+      "step": 3775
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.549906913515878e-06,
+      "loss": 0.9353,
+      "step": 3776
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.545476214678344e-06,
+      "loss": 0.8378,
+      "step": 3777
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.541048807119807e-06,
+      "loss": 0.9611,
+      "step": 3778
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5366246927950287e-06,
+      "loss": 0.9806,
+      "step": 3779
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5322038736573273e-06,
+      "loss": 0.9331,
+      "step": 3780
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.52778635165855e-06,
+      "loss": 0.9727,
+      "step": 3781
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5233721287490977e-06,
+      "loss": 1.133,
+      "step": 3782
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.518961206877911e-06,
+      "loss": 1.045,
+      "step": 3783
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5145535879924832e-06,
+      "loss": 1.053,
+      "step": 3784
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5101492740388324e-06,
+      "loss": 1.0412,
+      "step": 3785
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5057482669615263e-06,
+      "loss": 1.0822,
+      "step": 3786
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5013505687036787e-06,
+      "loss": 0.9788,
+      "step": 3787
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4969561812069367e-06,
+      "loss": 0.8834,
+      "step": 3788
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4925651064114788e-06,
+      "loss": 0.868,
+      "step": 3789
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4881773462560267e-06,
+      "loss": 0.9068,
+      "step": 3790
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.483792902677847e-06,
+      "loss": 0.9943,
+      "step": 3791
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4794117776127337e-06,
+      "loss": 1.0489,
+      "step": 3792
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.475033972995006e-06,
+      "loss": 0.9631,
+      "step": 3793
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4706594907575377e-06,
+      "loss": 0.9344,
+      "step": 3794
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.466288332831722e-06,
+      "loss": 1.2384,
+      "step": 3795
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.461920501147491e-06,
+      "loss": 1.1458,
+      "step": 3796
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4575559976332964e-06,
+      "loss": 0.8764,
+      "step": 3797
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4531948242161365e-06,
+      "loss": 0.9457,
+      "step": 3798
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4488369828215286e-06,
+      "loss": 1.0153,
+      "step": 3799
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4444824753735243e-06,
+      "loss": 1.0123,
+      "step": 3800
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4401313037946993e-06,
+      "loss": 1.0805,
+      "step": 3801
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4357834700061588e-06,
+      "loss": 0.8486,
+      "step": 3802
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4314389759275334e-06,
+      "loss": 0.9014,
+      "step": 3803
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4270978234769794e-06,
+      "loss": 0.9219,
+      "step": 3804
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4227600145711784e-06,
+      "loss": 0.9969,
+      "step": 3805
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4184255511253342e-06,
+      "loss": 0.8752,
+      "step": 3806
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4140944350531714e-06,
+      "loss": 1.1435,
+      "step": 3807
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4097666682669496e-06,
+      "loss": 1.1443,
+      "step": 3808
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4054422526774314e-06,
+      "loss": 1.0379,
+      "step": 3809
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.401121190193909e-06,
+      "loss": 0.9184,
+      "step": 3810
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3968034827241926e-06,
+      "loss": 1.1303,
+      "step": 3811
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.392489132174619e-06,
+      "loss": 1.105,
+      "step": 3812
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3881781404500284e-06,
+      "loss": 0.9798,
+      "step": 3813
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.383870509453785e-06,
+      "loss": 0.922,
+      "step": 3814
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.379566241087774e-06,
+      "loss": 1.0243,
+      "step": 3815
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3752653372523937e-06,
+      "loss": 0.9514,
+      "step": 3816
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3709677998465476e-06,
+      "loss": 1.0252,
+      "step": 3817
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.366673630767661e-06,
+      "loss": 0.963,
+      "step": 3818
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.362382831911675e-06,
+      "loss": 1.1937,
+      "step": 3819
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3580954051730396e-06,
+      "loss": 0.9635,
+      "step": 3820
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3538113524447083e-06,
+      "loss": 0.8373,
+      "step": 3821
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.349530675618158e-06,
+      "loss": 0.7977,
+      "step": 3822
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.345253376583366e-06,
+      "loss": 1.0244,
+      "step": 3823
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3409794572288226e-06,
+      "loss": 1.0217,
+      "step": 3824
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3367089194415226e-06,
+      "loss": 0.9876,
+      "step": 3825
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.332441765106972e-06,
+      "loss": 1.0611,
+      "step": 3826
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3281779961091777e-06,
+      "loss": 0.8469,
+      "step": 3827
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3239176143306576e-06,
+      "loss": 0.9679,
+      "step": 3828
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3196606216524286e-06,
+      "loss": 0.9183,
+      "step": 3829
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.315407019954016e-06,
+      "loss": 0.9357,
+      "step": 3830
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.311156811113444e-06,
+      "loss": 1.0267,
+      "step": 3831
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.306909997007243e-06,
+      "loss": 0.9589,
+      "step": 3832
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.30266657951044e-06,
+      "loss": 0.9295,
+      "step": 3833
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.298426560496566e-06,
+      "loss": 0.9883,
+      "step": 3834
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.294189941837647e-06,
+      "loss": 0.9805,
+      "step": 3835
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2899567254042187e-06,
+      "loss": 0.9596,
+      "step": 3836
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2857269130652983e-06,
+      "loss": 0.893,
+      "step": 3837
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.281500506688412e-06,
+      "loss": 1.0024,
+      "step": 3838
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.277277508139576e-06,
+      "loss": 0.9658,
+      "step": 3839
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2730579192833124e-06,
+      "loss": 0.8994,
+      "step": 3840
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2688417419826212e-06,
+      "loss": 1.0263,
+      "step": 3841
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2646289780990073e-06,
+      "loss": 1.0706,
+      "step": 3842
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2604196294924696e-06,
+      "loss": 1.0235,
+      "step": 3843
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2562136980214977e-06,
+      "loss": 1.0938,
+      "step": 3844
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2520111855430625e-06,
+      "loss": 0.9581,
+      "step": 3845
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2478120939126357e-06,
+      "loss": 0.9115,
+      "step": 3846
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2436164249841806e-06,
+      "loss": 0.9355,
+      "step": 3847
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2394241806101425e-06,
+      "loss": 0.9542,
+      "step": 3848
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.235235362641458e-06,
+      "loss": 0.9785,
+      "step": 3849
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.23104997292755e-06,
+      "loss": 0.9478,
+      "step": 3850
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.226868013316328e-06,
+      "loss": 0.9123,
+      "step": 3851
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.222689485654187e-06,
+      "loss": 0.9458,
+      "step": 3852
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.218514391786006e-06,
+      "loss": 0.9758,
+      "step": 3853
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2143427335551505e-06,
+      "loss": 0.9786,
+      "step": 3854
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2101745128034657e-06,
+      "loss": 1.0619,
+      "step": 3855
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.206009731371281e-06,
+      "loss": 1.0113,
+      "step": 3856
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2018483910974077e-06,
+      "loss": 1.1076,
+      "step": 3857
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1976904938191366e-06,
+      "loss": 1.0629,
+      "step": 3858
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1935360413722397e-06,
+      "loss": 0.9727,
+      "step": 3859
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.189385035590966e-06,
+      "loss": 0.9454,
+      "step": 3860
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1852374783080443e-06,
+      "loss": 0.8438,
+      "step": 3861
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.181093371354681e-06,
+      "loss": 0.9631,
+      "step": 3862
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1769527165605563e-06,
+      "loss": 0.7751,
+      "step": 3863
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1728155157538367e-06,
+      "loss": 1.0093,
+      "step": 3864
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.168681770761146e-06,
+      "loss": 0.9531,
+      "step": 3865
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1645514834075955e-06,
+      "loss": 0.771,
+      "step": 3866
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.160424655516764e-06,
+      "loss": 0.9921,
+      "step": 3867
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1563012889107125e-06,
+      "loss": 0.945,
+      "step": 3868
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.152181385409958e-06,
+      "loss": 0.9597,
+      "step": 3869
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.148064946833498e-06,
+      "loss": 0.9518,
+      "step": 3870
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1439519749988045e-06,
+      "loss": 1.111,
+      "step": 3871
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.139842471721809e-06,
+      "loss": 0.9936,
+      "step": 3872
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.135736438816921e-06,
+      "loss": 0.8671,
+      "step": 3873
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1316338780970026e-06,
+      "loss": 1.0338,
+      "step": 3874
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1275347913734023e-06,
+      "loss": 0.9261,
+      "step": 3875
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.123439180455924e-06,
+      "loss": 0.9124,
+      "step": 3876
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1193470471528364e-06,
+      "loss": 1.0115,
+      "step": 3877
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.115258393270876e-06,
+      "loss": 0.9236,
+      "step": 3878
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1111732206152424e-06,
+      "loss": 1.2379,
+      "step": 3879
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1070915309895967e-06,
+      "loss": 1.0399,
+      "step": 3880
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.103013326196064e-06,
+      "loss": 0.9858,
+      "step": 3881
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0989386080352304e-06,
+      "loss": 0.9052,
+      "step": 3882
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0948673783061424e-06,
+      "loss": 1.1097,
+      "step": 3883
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0907996388063056e-06,
+      "loss": 0.8988,
+      "step": 3884
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.086735391331686e-06,
+      "loss": 1.1413,
+      "step": 3885
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.082674637676706e-06,
+      "loss": 1.0299,
+      "step": 3886
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0786173796342468e-06,
+      "loss": 0.8897,
+      "step": 3887
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0745636189956465e-06,
+      "loss": 1.0238,
+      "step": 3888
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0705133575506976e-06,
+      "loss": 1.0352,
+      "step": 3889
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0664665970876496e-06,
+      "loss": 0.976,
+      "step": 3890
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0624233393932024e-06,
+      "loss": 1.0873,
+      "step": 3891
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.058383586252519e-06,
+      "loss": 1.0535,
+      "step": 3892
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0543473394492017e-06,
+      "loss": 0.9311,
+      "step": 3893
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.050314600765313e-06,
+      "loss": 0.9661,
+      "step": 3894
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0462853719813624e-06,
+      "loss": 1.0452,
+      "step": 3895
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0422596548763173e-06,
+      "loss": 0.9266,
+      "step": 3896
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.038237451227589e-06,
+      "loss": 1.003,
+      "step": 3897
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.034218762811032e-06,
+      "loss": 1.0283,
+      "step": 3898
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.03020359140096e-06,
+      "loss": 0.7872,
+      "step": 3899
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.026191938770128e-06,
+      "loss": 0.9269,
+      "step": 3900
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0221838066897404e-06,
+      "loss": 1.141,
+      "step": 3901
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.018179196929436e-06,
+      "loss": 0.9717,
+      "step": 3902
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0141781112573155e-06,
+      "loss": 0.9713,
+      "step": 3903
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0101805514399143e-06,
+      "loss": 0.7495,
+      "step": 3904
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0061865192422115e-06,
+      "loss": 1.0494,
+      "step": 3905
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.002196016427629e-06,
+      "loss": 1.0055,
+      "step": 3906
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9982090447580305e-06,
+      "loss": 1.051,
+      "step": 3907
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9942256059937225e-06,
+      "loss": 0.971,
+      "step": 3908
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9902457018934496e-06,
+      "loss": 1.0304,
+      "step": 3909
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9862693342143966e-06,
+      "loss": 1.1608,
+      "step": 3910
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9822965047121854e-06,
+      "loss": 1.0209,
+      "step": 3911
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9783272151408805e-06,
+      "loss": 1.0307,
+      "step": 3912
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9743614672529763e-06,
+      "loss": 0.962,
+      "step": 3913
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9703992627994097e-06,
+      "loss": 1.0932,
+      "step": 3914
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9664406035295493e-06,
+      "loss": 0.9133,
+      "step": 3915
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9624854911912006e-06,
+      "loss": 0.9313,
+      "step": 3916
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9585339275306025e-06,
+      "loss": 1.0204,
+      "step": 3917
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.954585914292426e-06,
+      "loss": 0.919,
+      "step": 3918
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.950641453219775e-06,
+      "loss": 0.8515,
+      "step": 3919
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9467005460541877e-06,
+      "loss": 1.0215,
+      "step": 3920
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.942763194535633e-06,
+      "loss": 1.1309,
+      "step": 3921
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9388294004025023e-06,
+      "loss": 1.105,
+      "step": 3922
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.934899165391623e-06,
+      "loss": 1.0186,
+      "step": 3923
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.930972491238253e-06,
+      "loss": 1.1368,
+      "step": 3924
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.927049379676078e-06,
+      "loss": 1.0389,
+      "step": 3925
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9231298324371982e-06,
+      "loss": 1.0731,
+      "step": 3926
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9192138512521585e-06,
+      "loss": 1.15,
+      "step": 3927
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.915301437849917e-06,
+      "loss": 0.8904,
+      "step": 3928
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.911392593957865e-06,
+      "loss": 1.1143,
+      "step": 3929
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.907487321301802e-06,
+      "loss": 0.9171,
+      "step": 3930
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9035856216059724e-06,
+      "loss": 1.0408,
+      "step": 3931
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.899687496593029e-06,
+      "loss": 1.0245,
+      "step": 3932
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8957929479840497e-06,
+      "loss": 1.017,
+      "step": 3933
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8919019774985336e-06,
+      "loss": 1.0128,
+      "step": 3934
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8880145868543997e-06,
+      "loss": 1.0079,
+      "step": 3935
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8841307777679874e-06,
+      "loss": 0.9668,
+      "step": 3936
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8802505519540525e-06,
+      "loss": 1.0722,
+      "step": 3937
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8763739111257718e-06,
+      "loss": 0.9253,
+      "step": 3938
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8725008569947366e-06,
+      "loss": 1.1113,
+      "step": 3939
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8686313912709564e-06,
+      "loss": 1.0581,
+      "step": 3940
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8647655156628552e-06,
+      "loss": 0.8143,
+      "step": 3941
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8609032318772735e-06,
+      "loss": 0.8388,
+      "step": 3942
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8570445416194639e-06,
+      "loss": 0.7683,
+      "step": 3943
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8531894465930911e-06,
+      "loss": 0.8695,
+      "step": 3944
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8493379485002427e-06,
+      "loss": 1.0809,
+      "step": 3945
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8454900490414019e-06,
+      "loss": 0.9693,
+      "step": 3946
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8416457499154727e-06,
+      "loss": 1.0126,
+      "step": 3947
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8378050528197722e-06,
+      "loss": 1.0907,
+      "step": 3948
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8339679594500225e-06,
+      "loss": 1.0443,
+      "step": 3949
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8301344715003522e-06,
+      "loss": 0.9426,
+      "step": 3950
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8263045906633004e-06,
+      "loss": 0.9938,
+      "step": 3951
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8224783186298177e-06,
+      "loss": 0.9556,
+      "step": 3952
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.818655657089261e-06,
+      "loss": 0.9697,
+      "step": 3953
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.81483660772938e-06,
+      "loss": 1.0634,
+      "step": 3954
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.811021172236348e-06,
+      "loss": 0.8655,
+      "step": 3955
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.807209352294732e-06,
+      "loss": 1.0743,
+      "step": 3956
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8034011495875082e-06,
+      "loss": 0.981,
+      "step": 3957
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7995965657960446e-06,
+      "loss": 0.8106,
+      "step": 3958
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7957956026001256e-06,
+      "loss": 0.8681,
+      "step": 3959
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7919982616779308e-06,
+      "loss": 1.0324,
+      "step": 3960
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7882045447060393e-06,
+      "loss": 1.0956,
+      "step": 3961
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7844144533594321e-06,
+      "loss": 0.8329,
+      "step": 3962
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7806279893114874e-06,
+      "loss": 1.0825,
+      "step": 3963
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7768451542339848e-06,
+      "loss": 0.9546,
+      "step": 3964
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7730659497970991e-06,
+      "loss": 0.8995,
+      "step": 3965
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7692903776694037e-06,
+      "loss": 1.0101,
+      "step": 3966
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7655184395178683e-06,
+      "loss": 0.87,
+      "step": 3967
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7617501370078538e-06,
+      "loss": 0.973,
+      "step": 3968
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7579854718031285e-06,
+      "loss": 0.9848,
+      "step": 3969
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7542244455658374e-06,
+      "loss": 0.9701,
+      "step": 3970
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.750467059956531e-06,
+      "loss": 1.0239,
+      "step": 3971
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7467133166341454e-06,
+      "loss": 0.9381,
+      "step": 3972
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7429632172560207e-06,
+      "loss": 0.9339,
+      "step": 3973
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.73921676347787e-06,
+      "loss": 1.0572,
+      "step": 3974
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7354739569538081e-06,
+      "loss": 0.842,
+      "step": 3975
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.731734799336342e-06,
+      "loss": 0.9589,
+      "step": 3976
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7279992922763644e-06,
+      "loss": 1.0066,
+      "step": 3977
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7242674374231483e-06,
+      "loss": 0.9704,
+      "step": 3978
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7205392364243623e-06,
+      "loss": 0.8309,
+      "step": 3979
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.716814690926064e-06,
+      "loss": 0.7911,
+      "step": 3980
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7130938025726963e-06,
+      "loss": 1.2051,
+      "step": 3981
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7093765730070743e-06,
+      "loss": 0.9036,
+      "step": 3982
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.705663003870418e-06,
+      "loss": 0.9521,
+      "step": 3983
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.701953096802318e-06,
+      "loss": 0.8833,
+      "step": 3984
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6982468534407527e-06,
+      "loss": 0.976,
+      "step": 3985
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6945442754220753e-06,
+      "loss": 0.9672,
+      "step": 3986
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6908453643810342e-06,
+      "loss": 0.8416,
+      "step": 3987
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.687150121950749e-06,
+      "loss": 1.0553,
+      "step": 3988
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6834585497627233e-06,
+      "loss": 0.9229,
+      "step": 3989
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.679770649446837e-06,
+      "loss": 1.0302,
+      "step": 3990
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6760864226313534e-06,
+      "loss": 0.9087,
+      "step": 3991
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6724058709429058e-06,
+      "loss": 1.0463,
+      "step": 3992
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6687289960065213e-06,
+      "loss": 0.8599,
+      "step": 3993
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.665055799445583e-06,
+      "loss": 0.928,
+      "step": 3994
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6613862828818628e-06,
+      "loss": 0.9336,
+      "step": 3995
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6577204479355026e-06,
+      "loss": 0.9341,
+      "step": 3996
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6540582962250284e-06,
+      "loss": 0.9627,
+      "step": 3997
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6503998293673252e-06,
+      "loss": 1.0856,
+      "step": 3998
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6467450489776581e-06,
+      "loss": 1.3381,
+      "step": 3999
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6430939566696713e-06,
+      "loss": 0.8575,
+      "step": 4000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6394465540553727e-06,
+      "loss": 0.956,
+      "step": 4001
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6358028427451379e-06,
+      "loss": 1.0251,
+      "step": 4002
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6321628243477194e-06,
+      "loss": 0.7714,
+      "step": 4003
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.628526500470241e-06,
+      "loss": 0.9875,
+      "step": 4004
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6248938727181917e-06,
+      "loss": 1.1749,
+      "step": 4005
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6212649426954208e-06,
+      "loss": 1.1088,
+      "step": 4006
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.617639712004162e-06,
+      "loss": 0.9488,
+      "step": 4007
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.614018182245003e-06,
+      "loss": 0.9216,
+      "step": 4008
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6104003550169034e-06,
+      "loss": 0.8451,
+      "step": 4009
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6067862319171778e-06,
+      "loss": 1.1064,
+      "step": 4010
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6031758145415222e-06,
+      "loss": 1.0394,
+      "step": 4011
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5995691044839845e-06,
+      "loss": 0.9396,
+      "step": 4012
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5959661033369777e-06,
+      "loss": 0.9878,
+      "step": 4013
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5923668126912806e-06,
+      "loss": 0.9714,
+      "step": 4014
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5887712341360294e-06,
+      "loss": 0.8497,
+      "step": 4015
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.585179369258726e-06,
+      "loss": 0.9064,
+      "step": 4016
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5815912196452265e-06,
+      "loss": 1.0544,
+      "step": 4017
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5780067868797545e-06,
+      "loss": 0.9208,
+      "step": 4018
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5744260725448845e-06,
+      "loss": 0.8396,
+      "step": 4019
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5708490782215535e-06,
+      "loss": 0.9172,
+      "step": 4020
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5672758054890624e-06,
+      "loss": 0.9473,
+      "step": 4021
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5637062559250537e-06,
+      "loss": 0.8456,
+      "step": 4022
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.560140431105539e-06,
+      "loss": 0.925,
+      "step": 4023
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5565783326048777e-06,
+      "loss": 1.0673,
+      "step": 4024
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5530199619957943e-06,
+      "loss": 1.2036,
+      "step": 4025
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5494653208493527e-06,
+      "loss": 1.1248,
+      "step": 4026
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5459144107349788e-06,
+      "loss": 0.999,
+      "step": 4027
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5423672332204552e-06,
+      "loss": 1.1109,
+      "step": 4028
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5388237898719105e-06,
+      "loss": 0.9572,
+      "step": 4029
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.535284082253823e-06,
+      "loss": 1.0613,
+      "step": 4030
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.531748111929021e-06,
+      "loss": 1.0468,
+      "step": 4031
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5282158804586934e-06,
+      "loss": 1.0724,
+      "step": 4032
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5246873894023706e-06,
+      "loss": 0.9164,
+      "step": 4033
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5211626403179236e-06,
+      "loss": 0.7669,
+      "step": 4034
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5176416347615886e-06,
+      "loss": 0.9818,
+      "step": 4035
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5141243742879363e-06,
+      "loss": 0.9628,
+      "step": 4036
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.510610860449887e-06,
+      "loss": 0.9588,
+      "step": 4037
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5071010947987085e-06,
+      "loss": 0.9726,
+      "step": 4038
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5035950788840125e-06,
+      "loss": 1.0239,
+      "step": 4039
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5000928142537552e-06,
+      "loss": 1.0056,
+      "step": 4040
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4965943024542361e-06,
+      "loss": 0.9927,
+      "step": 4041
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4930995450300977e-06,
+      "loss": 0.9895,
+      "step": 4042
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4896085435243279e-06,
+      "loss": 1.0779,
+      "step": 4043
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4861212994782514e-06,
+      "loss": 1.0241,
+      "step": 4044
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4826378144315368e-06,
+      "loss": 1.1846,
+      "step": 4045
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4791580899221936e-06,
+      "loss": 1.0386,
+      "step": 4046
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4756821274865696e-06,
+      "loss": 0.8603,
+      "step": 4047
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4722099286593494e-06,
+      "loss": 0.8526,
+      "step": 4048
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4687414949735646e-06,
+      "loss": 0.8954,
+      "step": 4049
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4652768279605734e-06,
+      "loss": 0.8507,
+      "step": 4050
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4618159291500777e-06,
+      "loss": 1.1108,
+      "step": 4051
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.45835880007011e-06,
+      "loss": 0.9805,
+      "step": 4052
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.454905442247051e-06,
+      "loss": 0.9313,
+      "step": 4053
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4514558572055993e-06,
+      "loss": 0.9908,
+      "step": 4054
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4480100464687973e-06,
+      "loss": 0.8672,
+      "step": 4055
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.444568011558023e-06,
+      "loss": 1.1137,
+      "step": 4056
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.441129753992986e-06,
+      "loss": 1.1278,
+      "step": 4057
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4376952752917206e-06,
+      "loss": 0.9345,
+      "step": 4058
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4342645769705977e-06,
+      "loss": 0.8987,
+      "step": 4059
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4308376605443253e-06,
+      "loss": 1.0005,
+      "step": 4060
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4274145275259333e-06,
+      "loss": 0.8805,
+      "step": 4061
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4239951794267848e-06,
+      "loss": 0.8384,
+      "step": 4062
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4205796177565688e-06,
+      "loss": 0.9572,
+      "step": 4063
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4171678440233072e-06,
+      "loss": 1.2208,
+      "step": 4064
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4137598597333458e-06,
+      "loss": 1.2057,
+      "step": 4065
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4103556663913586e-06,
+      "loss": 1.0945,
+      "step": 4066
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.406955265500346e-06,
+      "loss": 0.7332,
+      "step": 4067
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4035586585616323e-06,
+      "loss": 1.0007,
+      "step": 4068
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4001658470748703e-06,
+      "loss": 1.1356,
+      "step": 4069
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3967768325380327e-06,
+      "loss": 0.988,
+      "step": 4070
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3933916164474193e-06,
+      "loss": 0.7883,
+      "step": 4071
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3900102002976512e-06,
+      "loss": 1.0605,
+      "step": 4072
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.386632585581672e-06,
+      "loss": 1.1061,
+      "step": 4073
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3832587737907476e-06,
+      "loss": 0.9444,
+      "step": 4074
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3798887664144634e-06,
+      "loss": 0.9243,
+      "step": 4075
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3765225649407232e-06,
+      "loss": 0.8786,
+      "step": 4076
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3731601708557608e-06,
+      "loss": 0.8658,
+      "step": 4077
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3698015856441138e-06,
+      "loss": 0.9305,
+      "step": 4078
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3664468107886496e-06,
+      "loss": 0.9826,
+      "step": 4079
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3630958477705448e-06,
+      "loss": 0.9906,
+      "step": 4080
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3597486980693065e-06,
+      "loss": 1.0306,
+      "step": 4081
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3564053631627406e-06,
+      "loss": 0.9887,
+      "step": 4082
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3530658445269784e-06,
+      "loss": 0.9647,
+      "step": 4083
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3497301436364696e-06,
+      "loss": 1.131,
+      "step": 4084
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3463982619639704e-06,
+      "loss": 0.8924,
+      "step": 4085
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3430702009805586e-06,
+      "loss": 0.9476,
+      "step": 4086
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.339745962155613e-06,
+      "loss": 0.8079,
+      "step": 4087
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.33642554695684e-06,
+      "loss": 1.0285,
+      "step": 4088
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3331089568502465e-06,
+      "loss": 1.0444,
+      "step": 4089
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.329796193300157e-06,
+      "loss": 1.1577,
+      "step": 4090
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3264872577692022e-06,
+      "loss": 1.0422,
+      "step": 4091
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.323182151718324e-06,
+      "loss": 1.0567,
+      "step": 4092
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.319880876606776e-06,
+      "loss": 0.8642,
+      "step": 4093
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.316583433892118e-06,
+      "loss": 0.849,
+      "step": 4094
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3132898250302173e-06,
+      "loss": 1.2195,
+      "step": 4095
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3100000514752497e-06,
+      "loss": 0.8758,
+      "step": 4096
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3067141146796968e-06,
+      "loss": 1.1606,
+      "step": 4097
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.303432016094348e-06,
+      "loss": 1.0156,
+      "step": 4098
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3001537571682965e-06,
+      "loss": 1.0641,
+      "step": 4099
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2968793393489398e-06,
+      "loss": 0.9509,
+      "step": 4100
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2936087640819805e-06,
+      "loss": 1.0846,
+      "step": 4101
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2903420328114247e-06,
+      "loss": 0.9022,
+      "step": 4102
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2870791469795818e-06,
+      "loss": 1.1757,
+      "step": 4103
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2838201080270585e-06,
+      "loss": 0.9203,
+      "step": 4104
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.280564917392776e-06,
+      "loss": 1.1176,
+      "step": 4105
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2773135765139389e-06,
+      "loss": 1.0243,
+      "step": 4106
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2740660868260634e-06,
+      "loss": 1.2112,
+      "step": 4107
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2708224497629617e-06,
+      "loss": 0.9424,
+      "step": 4108
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2675826667567504e-06,
+      "loss": 0.8504,
+      "step": 4109
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2643467392378383e-06,
+      "loss": 1.1366,
+      "step": 4110
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2611146686349284e-06,
+      "loss": 0.9066,
+      "step": 4111
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2578864563750326e-06,
+      "loss": 0.9596,
+      "step": 4112
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2546621038834506e-06,
+      "loss": 0.8735,
+      "step": 4113
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2514416125837825e-06,
+      "loss": 0.9195,
+      "step": 4114
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2482249838979143e-06,
+      "loss": 0.9991,
+      "step": 4115
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2450122192460412e-06,
+      "loss": 1.0172,
+      "step": 4116
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2418033200466418e-06,
+      "loss": 0.9033,
+      "step": 4117
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2385982877164905e-06,
+      "loss": 1.097,
+      "step": 4118
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2353971236706564e-06,
+      "loss": 0.9081,
+      "step": 4119
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.232199829322498e-06,
+      "loss": 0.9299,
+      "step": 4120
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2290064060836671e-06,
+      "loss": 1.043,
+      "step": 4121
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2258168553641058e-06,
+      "loss": 0.9456,
+      "step": 4122
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2226311785720468e-06,
+      "loss": 0.9675,
+      "step": 4123
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2194493771140115e-06,
+      "loss": 0.8962,
+      "step": 4124
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2162714523948104e-06,
+      "loss": 1.002,
+      "step": 4125
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2130974058175438e-06,
+      "loss": 1.0426,
+      "step": 4126
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.209927238783598e-06,
+      "loss": 0.8622,
+      "step": 4127
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.206760952692647e-06,
+      "loss": 1.002,
+      "step": 4128
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2035985489426526e-06,
+      "loss": 0.762,
+      "step": 4129
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2004400289298602e-06,
+      "loss": 0.9598,
+      "step": 4130
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1972853940488017e-06,
+      "loss": 1.0434,
+      "step": 4131
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.194134645692292e-06,
+      "loss": 1.0122,
+      "step": 4132
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1909877852514351e-06,
+      "loss": 0.8958,
+      "step": 4133
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1878448141156162e-06,
+      "loss": 1.0526,
+      "step": 4134
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1847057336724965e-06,
+      "loss": 0.9395,
+      "step": 4135
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.181570545308025e-06,
+      "loss": 0.96,
+      "step": 4136
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1784392504064378e-06,
+      "loss": 1.0734,
+      "step": 4137
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1753118503502447e-06,
+      "loss": 0.9987,
+      "step": 4138
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1721883465202333e-06,
+      "loss": 1.0294,
+      "step": 4139
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1690687402954792e-06,
+      "loss": 0.9591,
+      "step": 4140
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1659530330533331e-06,
+      "loss": 0.9806,
+      "step": 4141
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1628412261694256e-06,
+      "loss": 1.0023,
+      "step": 4142
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1597333210176587e-06,
+      "loss": 1.0094,
+      "step": 4143
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1566293189702215e-06,
+      "loss": 0.9417,
+      "step": 4144
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1535292213975747e-06,
+      "loss": 1.03,
+      "step": 4145
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1504330296684552e-06,
+      "loss": 0.8965,
+      "step": 4146
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1473407451498752e-06,
+      "loss": 0.9971,
+      "step": 4147
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.144252369207125e-06,
+      "loss": 0.8512,
+      "step": 4148
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1411679032037636e-06,
+      "loss": 1.0365,
+      "step": 4149
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1380873485016297e-06,
+      "loss": 1.083,
+      "step": 4150
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1350107064608317e-06,
+      "loss": 0.972,
+      "step": 4151
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1319379784397499e-06,
+      "loss": 1.0136,
+      "step": 4152
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.128869165795039e-06,
+      "loss": 0.8709,
+      "step": 4153
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1258042698816218e-06,
+      "loss": 1.1472,
+      "step": 4154
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.122743292052697e-06,
+      "loss": 1.0324,
+      "step": 4155
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.119686233659727e-06,
+      "loss": 0.9099,
+      "step": 4156
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1166330960524464e-06,
+      "loss": 0.9425,
+      "step": 4157
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1135838805788646e-06,
+      "loss": 1.0147,
+      "step": 4158
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1105385885852483e-06,
+      "loss": 0.9184,
+      "step": 4159
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1074972214161384e-06,
+      "loss": 0.8871,
+      "step": 4160
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1044597804143441e-06,
+      "loss": 1.018,
+      "step": 4161
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1014262669209397e-06,
+      "loss": 0.9817,
+      "step": 4162
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0983966822752624e-06,
+      "loss": 1.0285,
+      "step": 4163
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0953710278149155e-06,
+      "loss": 1.1891,
+      "step": 4164
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0923493048757728e-06,
+      "loss": 1.0003,
+      "step": 4165
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.089331514791969e-06,
+      "loss": 0.9435,
+      "step": 4166
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0863176588958957e-06,
+      "loss": 0.953,
+      "step": 4167
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.083307738518219e-06,
+      "loss": 1.0844,
+      "step": 4168
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0803017549878592e-06,
+      "loss": 0.9363,
+      "step": 4169
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0772997096320049e-06,
+      "loss": 1.1367,
+      "step": 4170
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0743016037760946e-06,
+      "loss": 1.0031,
+      "step": 4171
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0713074387438405e-06,
+      "loss": 0.8554,
+      "step": 4172
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0683172158572097e-06,
+      "loss": 0.8832,
+      "step": 4173
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.065330936436426e-06,
+      "loss": 0.9717,
+      "step": 4174
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0623486017999762e-06,
+      "loss": 1.1115,
+      "step": 4175
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0593702132646023e-06,
+      "loss": 0.9004,
+      "step": 4176
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0563957721453067e-06,
+      "loss": 0.8131,
+      "step": 4177
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0534252797553468e-06,
+      "loss": 1.038,
+      "step": 4178
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0504587374062392e-06,
+      "loss": 1.079,
+      "step": 4179
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0474961464077527e-06,
+      "loss": 0.9416,
+      "step": 4180
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0445375080679122e-06,
+      "loss": 0.9274,
+      "step": 4181
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0415828236930059e-06,
+      "loss": 0.9426,
+      "step": 4182
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0386320945875627e-06,
+      "loss": 0.9904,
+      "step": 4183
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0356853220543727e-06,
+      "loss": 1.0065,
+      "step": 4184
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0327425073944785e-06,
+      "loss": 0.9951,
+      "step": 4185
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0298036519071785e-06,
+      "loss": 0.9696,
+      "step": 4186
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0268687568900159e-06,
+      "loss": 1.0079,
+      "step": 4187
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0239378236387864e-06,
+      "loss": 0.8821,
+      "step": 4188
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0210108534475438e-06,
+      "loss": 0.8956,
+      "step": 4189
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0180878476085886e-06,
+      "loss": 0.9887,
+      "step": 4190
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0151688074124645e-06,
+      "loss": 1.0744,
+      "step": 4191
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.01225373414797e-06,
+      "loss": 0.9527,
+      "step": 4192
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0093426291021557e-06,
+      "loss": 1.2207,
+      "step": 4193
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.006435493560315e-06,
+      "loss": 0.9183,
+      "step": 4194
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0035323288059861e-06,
+      "loss": 0.8736,
+      "step": 4195
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.000633136120962e-06,
+      "loss": 0.9772,
+      "step": 4196
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.977379167852763e-07,
+      "loss": 1.0293,
+      "step": 4197
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.94846672077212e-07,
+      "loss": 0.8568,
+      "step": 4198
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.919594032732893e-07,
+      "loss": 0.9831,
+      "step": 4199
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.89076111648284e-07,
+      "loss": 1.1854,
+      "step": 4200
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.861967984752086e-07,
+      "loss": 1.0709,
+      "step": 4201
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.83321465025322e-07,
+      "loss": 0.8677,
+      "step": 4202
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.804501125681243e-07,
+      "loss": 0.9491,
+      "step": 4203
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.77582742371359e-07,
+      "loss": 0.9503,
+      "step": 4204
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.747193557010114e-07,
+      "loss": 0.9324,
+      "step": 4205
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.71859953821307e-07,
+      "loss": 1.028,
+      "step": 4206
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.690045379947134e-07,
+      "loss": 0.9903,
+      "step": 4207
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.661531094819366e-07,
+      "loss": 0.9346,
+      "step": 4208
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.633056695419229e-07,
+      "loss": 0.9327,
+      "step": 4209
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.604622194318625e-07,
+      "loss": 1.0455,
+      "step": 4210
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.576227604071731e-07,
+      "loss": 1.0115,
+      "step": 4211
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.547872937215187e-07,
+      "loss": 1.0878,
+      "step": 4212
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.519558206267964e-07,
+      "loss": 1.1005,
+      "step": 4213
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.491283423731501e-07,
+      "loss": 1.0428,
+      "step": 4214
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.463048602089431e-07,
+      "loss": 0.8291,
+      "step": 4215
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.434853753807838e-07,
+      "loss": 0.8994,
+      "step": 4216
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.406698891335209e-07,
+      "loss": 0.9755,
+      "step": 4217
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.3785840271023e-07,
+      "loss": 1.0114,
+      "step": 4218
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.350509173522193e-07,
+      "loss": 0.984,
+      "step": 4219
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.322474342990339e-07,
+      "loss": 0.9518,
+      "step": 4220
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.294479547884561e-07,
+      "loss": 0.8808,
+      "step": 4221
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.266524800564947e-07,
+      "loss": 0.8716,
+      "step": 4222
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.23861011337387e-07,
+      "loss": 1.115,
+      "step": 4223
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.210735498636125e-07,
+      "loss": 0.9221,
+      "step": 4224
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.182900968658736e-07,
+      "loss": 1.2857,
+      "step": 4225
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.155106535731029e-07,
+      "loss": 1.0759,
+      "step": 4226
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.127352212124663e-07,
+      "loss": 1.0406,
+      "step": 4227
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.09963801009357e-07,
+      "loss": 1.0245,
+      "step": 4228
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.071963941873952e-07,
+      "loss": 1.0904,
+      "step": 4229
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.044330019684311e-07,
+      "loss": 0.9719,
+      "step": 4230
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.016736255725434e-07,
+      "loss": 0.8701,
+      "step": 4231
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.98918266218034e-07,
+      "loss": 0.772,
+      "step": 4232
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.961669251214356e-07,
+      "loss": 0.8869,
+      "step": 4233
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.934196034975028e-07,
+      "loss": 1.0112,
+      "step": 4234
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.906763025592191e-07,
+      "loss": 1.0241,
+      "step": 4235
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.879370235177909e-07,
+      "loss": 1.0119,
+      "step": 4236
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.852017675826463e-07,
+      "loss": 0.7877,
+      "step": 4237
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.824705359614461e-07,
+      "loss": 1.1267,
+      "step": 4238
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.797433298600622e-07,
+      "loss": 0.9849,
+      "step": 4239
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.77020150482597e-07,
+      "loss": 1.0607,
+      "step": 4240
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.743009990313723e-07,
+      "loss": 1.1684,
+      "step": 4241
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.715858767069374e-07,
+      "loss": 0.9227,
+      "step": 4242
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.688747847080514e-07,
+      "loss": 1.1656,
+      "step": 4243
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.661677242317024e-07,
+      "loss": 0.8674,
+      "step": 4244
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.634646964730975e-07,
+      "loss": 0.961,
+      "step": 4245
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.607657026256633e-07,
+      "loss": 1.2805,
+      "step": 4246
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.580707438810398e-07,
+      "loss": 0.9894,
+      "step": 4247
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.55379821429092e-07,
+      "loss": 0.9625,
+      "step": 4248
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.526929364579018e-07,
+      "loss": 0.9648,
+      "step": 4249
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.500100901537678e-07,
+      "loss": 1.0866,
+      "step": 4250
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.473312837012027e-07,
+      "loss": 0.9007,
+      "step": 4251
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.446565182829403e-07,
+      "loss": 1.0164,
+      "step": 4252
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.419857950799259e-07,
+      "loss": 0.9791,
+      "step": 4253
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.393191152713231e-07,
+      "loss": 1.0644,
+      "step": 4254
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.366564800345089e-07,
+      "loss": 1.1425,
+      "step": 4255
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.339978905450752e-07,
+      "loss": 0.9186,
+      "step": 4256
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.313433479768262e-07,
+      "loss": 0.9178,
+      "step": 4257
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.286928535017824e-07,
+      "loss": 1.0002,
+      "step": 4258
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.260464082901732e-07,
+      "loss": 1.0155,
+      "step": 4259
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.234040135104437e-07,
+      "loss": 0.9955,
+      "step": 4260
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.207656703292477e-07,
+      "loss": 1.0477,
+      "step": 4261
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.181313799114509e-07,
+      "loss": 0.8988,
+      "step": 4262
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.155011434201332e-07,
+      "loss": 0.8883,
+      "step": 4263
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.128749620165788e-07,
+      "loss": 1.0553,
+      "step": 4264
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.102528368602835e-07,
+      "loss": 0.9184,
+      "step": 4265
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.076347691089592e-07,
+      "loss": 1.0527,
+      "step": 4266
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.050207599185134e-07,
+      "loss": 0.8743,
+      "step": 4267
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.024108104430717e-07,
+      "loss": 1.1729,
+      "step": 4268
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.998049218349624e-07,
+      "loss": 0.9165,
+      "step": 4269
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.972030952447274e-07,
+      "loss": 0.8856,
+      "step": 4270
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.946053318211045e-07,
+      "loss": 0.8258,
+      "step": 4271
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.920116327110449e-07,
+      "loss": 0.9168,
+      "step": 4272
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.894219990597063e-07,
+      "loss": 0.9257,
+      "step": 4273
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.868364320104482e-07,
+      "loss": 1.0306,
+      "step": 4274
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.842549327048366e-07,
+      "loss": 0.9286,
+      "step": 4275
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.816775022826339e-07,
+      "loss": 0.8951,
+      "step": 4276
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.791041418818213e-07,
+      "loss": 0.8989,
+      "step": 4277
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.765348526385685e-07,
+      "loss": 0.9689,
+      "step": 4278
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.739696356872562e-07,
+      "loss": 0.9444,
+      "step": 4279
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.714084921604636e-07,
+      "loss": 0.8693,
+      "step": 4280
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.688514231889721e-07,
+      "loss": 1.0165,
+      "step": 4281
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.66298429901765e-07,
+      "loss": 0.9924,
+      "step": 4282
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.637495134260242e-07,
+      "loss": 1.1435,
+      "step": 4283
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.612046748871327e-07,
+      "loss": 1.0221,
+      "step": 4284
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.58663915408675e-07,
+      "loss": 1.0458,
+      "step": 4285
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.561272361124306e-07,
+      "loss": 0.9074,
+      "step": 4286
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.535946381183812e-07,
+      "loss": 1.0361,
+      "step": 4287
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.510661225447057e-07,
+      "loss": 1.1379,
+      "step": 4288
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.48541690507778e-07,
+      "loss": 0.9967,
+      "step": 4289
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.460213431221719e-07,
+      "loss": 0.9756,
+      "step": 4290
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.435050815006562e-07,
+      "loss": 0.8576,
+      "step": 4291
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.409929067541977e-07,
+      "loss": 0.865,
+      "step": 4292
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.384848199919558e-07,
+      "loss": 0.9805,
+      "step": 4293
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.359808223212905e-07,
+      "loss": 1.0502,
+      "step": 4294
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.334809148477484e-07,
+      "loss": 0.93,
+      "step": 4295
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.309850986750766e-07,
+      "loss": 1.0795,
+      "step": 4296
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.284933749052103e-07,
+      "loss": 0.9807,
+      "step": 4297
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.260057446382862e-07,
+      "loss": 0.9985,
+      "step": 4298
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.23522208972628e-07,
+      "loss": 1.0071,
+      "step": 4299
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.210427690047484e-07,
+      "loss": 0.9931,
+      "step": 4300
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.185674258293584e-07,
+      "loss": 1.0245,
+      "step": 4301
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.160961805393574e-07,
+      "loss": 1.0174,
+      "step": 4302
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.136290342258378e-07,
+      "loss": 1.0206,
+      "step": 4303
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.11165987978073e-07,
+      "loss": 1.0214,
+      "step": 4304
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.087070428835397e-07,
+      "loss": 0.8917,
+      "step": 4305
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.062522000278959e-07,
+      "loss": 1.0987,
+      "step": 4306
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.038014604949883e-07,
+      "loss": 1.0097,
+      "step": 4307
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.013548253668556e-07,
+      "loss": 1.0326,
+      "step": 4308
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.989122957237204e-07,
+      "loss": 1.0474,
+      "step": 4309
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.964738726439957e-07,
+      "loss": 0.9374,
+      "step": 4310
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.940395572042791e-07,
+      "loss": 0.9552,
+      "step": 4311
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.916093504793564e-07,
+      "loss": 0.8741,
+      "step": 4312
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.891832535421994e-07,
+      "loss": 1.0187,
+      "step": 4313
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.867612674639634e-07,
+      "loss": 1.0327,
+      "step": 4314
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.843433933139909e-07,
+      "loss": 1.0374,
+      "step": 4315
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.819296321598067e-07,
+      "loss": 0.9907,
+      "step": 4316
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.79519985067123e-07,
+      "loss": 0.9165,
+      "step": 4317
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.771144530998308e-07,
+      "loss": 1.0188,
+      "step": 4318
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.747130373200095e-07,
+      "loss": 0.9826,
+      "step": 4319
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.723157387879164e-07,
+      "loss": 0.883,
+      "step": 4320
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.699225585619928e-07,
+      "loss": 1.0537,
+      "step": 4321
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.675334976988656e-07,
+      "loss": 0.8011,
+      "step": 4322
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.651485572533379e-07,
+      "loss": 1.0735,
+      "step": 4323
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.627677382783926e-07,
+      "loss": 0.9626,
+      "step": 4324
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.603910418251958e-07,
+      "loss": 0.837,
+      "step": 4325
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.580184689430958e-07,
+      "loss": 1.1296,
+      "step": 4326
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.556500206796179e-07,
+      "loss": 0.8955,
+      "step": 4327
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.532856980804602e-07,
+      "loss": 0.9545,
+      "step": 4328
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.509255021895111e-07,
+      "loss": 0.9801,
+      "step": 4329
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.48569434048828e-07,
+      "loss": 0.9999,
+      "step": 4330
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.462174946986511e-07,
+      "loss": 0.992,
+      "step": 4331
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.438696851773906e-07,
+      "loss": 1.004,
+      "step": 4332
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.415260065216422e-07,
+      "loss": 1.0383,
+      "step": 4333
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.391864597661723e-07,
+      "loss": 0.9372,
+      "step": 4334
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.368510459439248e-07,
+      "loss": 0.9306,
+      "step": 4335
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.34519766086017e-07,
+      "loss": 0.9375,
+      "step": 4336
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.321926212217432e-07,
+      "loss": 0.9463,
+      "step": 4337
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.298696123785719e-07,
+      "loss": 1.0763,
+      "step": 4338
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.275507405821435e-07,
+      "loss": 0.8196,
+      "step": 4339
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.252360068562735e-07,
+      "loss": 0.9615,
+      "step": 4340
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.2292541222295e-07,
+      "loss": 1.0619,
+      "step": 4341
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.206189577023347e-07,
+      "loss": 0.9002,
+      "step": 4342
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.183166443127587e-07,
+      "loss": 0.9264,
+      "step": 4343
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.160184730707264e-07,
+      "loss": 1.0547,
+      "step": 4344
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.137244449909153e-07,
+      "loss": 1.0491,
+      "step": 4345
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.114345610861672e-07,
+      "loss": 1.0457,
+      "step": 4346
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.091488223675058e-07,
+      "loss": 0.9774,
+      "step": 4347
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.068672298441114e-07,
+      "loss": 1.016,
+      "step": 4348
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.045897845233395e-07,
+      "loss": 1.0659,
+      "step": 4349
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.023164874107202e-07,
+      "loss": 0.8713,
+      "step": 4350
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.000473395099438e-07,
+      "loss": 1.0862,
+      "step": 4351
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.977823418228701e-07,
+      "loss": 0.8451,
+      "step": 4352
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.95521495349527e-07,
+      "loss": 1.1505,
+      "step": 4353
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.932648010881159e-07,
+      "loss": 0.9376,
+      "step": 4354
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.910122600349966e-07,
+      "loss": 0.9907,
+      "step": 4355
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.887638731846967e-07,
+      "loss": 0.9468,
+      "step": 4356
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.865196415299135e-07,
+      "loss": 1.0921,
+      "step": 4357
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.842795660615064e-07,
+      "loss": 0.8396,
+      "step": 4358
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.820436477685021e-07,
+      "loss": 1.0526,
+      "step": 4359
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.798118876380876e-07,
+      "loss": 1.2487,
+      "step": 4360
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.775842866556191e-07,
+      "loss": 0.868,
+      "step": 4361
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.753608458046145e-07,
+      "loss": 0.9867,
+      "step": 4362
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.73141566066755e-07,
+      "loss": 1.1652,
+      "step": 4363
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.709264484218835e-07,
+      "loss": 1.1823,
+      "step": 4364
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.687154938480055e-07,
+      "loss": 1.0467,
+      "step": 4365
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.665087033212902e-07,
+      "loss": 0.8832,
+      "step": 4366
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.64306077816068e-07,
+      "loss": 0.8616,
+      "step": 4367
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.62107618304828e-07,
+      "loss": 0.9081,
+      "step": 4368
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.599133257582212e-07,
+      "loss": 1.0579,
+      "step": 4369
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.577232011450595e-07,
+      "loss": 0.87,
+      "step": 4370
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.555372454323182e-07,
+      "loss": 0.9812,
+      "step": 4371
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.533554595851231e-07,
+      "loss": 1.0055,
+      "step": 4372
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.511778445667648e-07,
+      "loss": 0.9253,
+      "step": 4373
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.490044013386919e-07,
+      "loss": 1.0893,
+      "step": 4374
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.468351308605135e-07,
+      "loss": 0.9505,
+      "step": 4375
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.446700340899891e-07,
+      "loss": 0.8757,
+      "step": 4376
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.425091119830417e-07,
+      "loss": 1.0613,
+      "step": 4377
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.403523654937504e-07,
+      "loss": 0.9849,
+      "step": 4378
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.3819979557435e-07,
+      "loss": 1.0139,
+      "step": 4379
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.360514031752284e-07,
+      "loss": 0.9303,
+      "step": 4380
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.339071892449321e-07,
+      "loss": 0.9052,
+      "step": 4381
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.317671547301651e-07,
+      "loss": 0.9595,
+      "step": 4382
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.296313005757814e-07,
+      "loss": 0.8892,
+      "step": 4383
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.27499627724789e-07,
+      "loss": 0.979,
+      "step": 4384
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.253721371183563e-07,
+      "loss": 0.7617,
+      "step": 4385
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.232488296957983e-07,
+      "loss": 1.0207,
+      "step": 4386
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.211297063945875e-07,
+      "loss": 0.9647,
+      "step": 4387
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.190147681503421e-07,
+      "loss": 0.8682,
+      "step": 4388
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.169040158968431e-07,
+      "loss": 1.0245,
+      "step": 4389
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.14797450566017e-07,
+      "loss": 0.9565,
+      "step": 4390
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.126950730879399e-07,
+      "loss": 0.9432,
+      "step": 4391
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.105968843908449e-07,
+      "loss": 0.9912,
+      "step": 4392
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.085028854011098e-07,
+      "loss": 1.1338,
+      "step": 4393
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.064130770432651e-07,
+      "loss": 0.9807,
+      "step": 4394
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.043274602399939e-07,
+      "loss": 0.9086,
+      "step": 4395
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.022460359121228e-07,
+      "loss": 1.121,
+      "step": 4396
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.001688049786324e-07,
+      "loss": 0.8934,
+      "step": 4397
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.980957683566468e-07,
+      "loss": 0.9823,
+      "step": 4398
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.96026926961447e-07,
+      "loss": 1.0694,
+      "step": 4399
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.939622817064516e-07,
+      "loss": 1.1431,
+      "step": 4400
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.919018335032322e-07,
+      "loss": 1.0009,
+      "step": 4401
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.898455832615057e-07,
+      "loss": 1.0161,
+      "step": 4402
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.877935318891381e-07,
+      "loss": 0.9971,
+      "step": 4403
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.857456802921368e-07,
+      "loss": 0.9359,
+      "step": 4404
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.837020293746575e-07,
+      "loss": 0.8115,
+      "step": 4405
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.81662580039004e-07,
+      "loss": 0.8385,
+      "step": 4406
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.796273331856227e-07,
+      "loss": 0.9354,
+      "step": 4407
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.775962897131004e-07,
+      "loss": 0.9401,
+      "step": 4408
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.755694505181729e-07,
+      "loss": 0.9086,
+      "step": 4409
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.735468164957213e-07,
+      "loss": 1.0027,
+      "step": 4410
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.715283885387678e-07,
+      "loss": 0.9715,
+      "step": 4411
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.695141675384729e-07,
+      "loss": 0.8281,
+      "step": 4412
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6750415438414745e-07,
+      "loss": 1.0271,
+      "step": 4413
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.654983499632393e-07,
+      "loss": 0.9369,
+      "step": 4414
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.634967551613423e-07,
+      "loss": 1.0177,
+      "step": 4415
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.614993708621862e-07,
+      "loss": 1.0324,
+      "step": 4416
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5950619794764674e-07,
+      "loss": 0.9145,
+      "step": 4417
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5751723729773766e-07,
+      "loss": 0.8909,
+      "step": 4418
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.555324897906133e-07,
+      "loss": 1.0732,
+      "step": 4419
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.535519563025692e-07,
+      "loss": 1.066,
+      "step": 4420
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5157563770803826e-07,
+      "loss": 0.9527,
+      "step": 4421
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4960353487959354e-07,
+      "loss": 0.9623,
+      "step": 4422
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.476356486879474e-07,
+      "loss": 1.0576,
+      "step": 4423
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.456719800019482e-07,
+      "loss": 1.0141,
+      "step": 4424
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.437125296885847e-07,
+      "loss": 1.1028,
+      "step": 4425
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4175729861298144e-07,
+      "loss": 0.9912,
+      "step": 4426
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.398062876384046e-07,
+      "loss": 1.0066,
+      "step": 4427
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.378594976262496e-07,
+      "loss": 0.9553,
+      "step": 4428
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3591692943605323e-07,
+      "loss": 1.0118,
+      "step": 4429
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.339785839254851e-07,
+      "loss": 1.052,
+      "step": 4430
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.320444619503583e-07,
+      "loss": 0.9726,
+      "step": 4431
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3011456436460985e-07,
+      "loss": 1.0846,
+      "step": 4432
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.28188892020317e-07,
+      "loss": 1.0459,
+      "step": 4433
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.262674457676952e-07,
+      "loss": 1.0531,
+      "step": 4434
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.243502264550903e-07,
+      "loss": 1.0078,
+      "step": 4435
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.224372349289796e-07,
+      "loss": 1.004,
+      "step": 4436
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.205284720339764e-07,
+      "loss": 0.8984,
+      "step": 4437
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.186239386128288e-07,
+      "loss": 1.0344,
+      "step": 4438
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1672363550641415e-07,
+      "loss": 1.0757,
+      "step": 4439
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1482756355374463e-07,
+      "loss": 0.9278,
+      "step": 4440
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.129357235919618e-07,
+      "loss": 1.1423,
+      "step": 4441
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.110481164563407e-07,
+      "loss": 0.9456,
+      "step": 4442
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0916474298028694e-07,
+      "loss": 0.9344,
+      "step": 4443
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.072856039953366e-07,
+      "loss": 1.1171,
+      "step": 4444
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0541070033115804e-07,
+      "loss": 0.9176,
+      "step": 4445
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0354003281554586e-07,
+      "loss": 1.0748,
+      "step": 4446
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.016736022744272e-07,
+      "loss": 1.0148,
+      "step": 4447
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.998114095318584e-07,
+      "loss": 0.9414,
+      "step": 4448
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9795345541002395e-07,
+      "loss": 0.9776,
+      "step": 4449
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.960997407292366e-07,
+      "loss": 1.0787,
+      "step": 4450
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.942502663079395e-07,
+      "loss": 0.943,
+      "step": 4451
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.924050329627005e-07,
+      "loss": 1.1135,
+      "step": 4452
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9056404150821793e-07,
+      "loss": 0.9853,
+      "step": 4453
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.887272927573127e-07,
+      "loss": 1.0605,
+      "step": 4454
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8689478752094167e-07,
+      "loss": 1.044,
+      "step": 4455
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.850665266081766e-07,
+      "loss": 1.0506,
+      "step": 4456
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8324251082622276e-07,
+      "loss": 1.0642,
+      "step": 4457
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.814227409804094e-07,
+      "loss": 0.9416,
+      "step": 4458
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7960721787419164e-07,
+      "loss": 0.9329,
+      "step": 4459
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.777959423091482e-07,
+      "loss": 0.8797,
+      "step": 4460
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.759889150849805e-07,
+      "loss": 1.0677,
+      "step": 4461
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7418613699952255e-07,
+      "loss": 1.0593,
+      "step": 4462
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7238760884872216e-07,
+      "loss": 0.8809,
+      "step": 4463
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7059333142665854e-07,
+      "loss": 0.9282,
+      "step": 4464
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.688033055255269e-07,
+      "loss": 0.9979,
+      "step": 4465
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.670175319356528e-07,
+      "loss": 1.0038,
+      "step": 4466
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6523601144548003e-07,
+      "loss": 0.9582,
+      "step": 4467
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.634587448415738e-07,
+      "loss": 0.9538,
+      "step": 4468
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.616857329086254e-07,
+      "loss": 1.1528,
+      "step": 4469
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.599169764294419e-07,
+      "loss": 0.9887,
+      "step": 4470
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5815247618495753e-07,
+      "loss": 1.0715,
+      "step": 4471
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5639223295422245e-07,
+      "loss": 0.961,
+      "step": 4472
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.546362475144105e-07,
+      "loss": 0.95,
+      "step": 4473
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5288452064081267e-07,
+      "loss": 0.8849,
+      "step": 4474
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5113705310684363e-07,
+      "loss": 0.885,
+      "step": 4475
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.49393845684034e-07,
+      "loss": 1.1941,
+      "step": 4476
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.476548991420359e-07,
+      "loss": 1.0391,
+      "step": 4477
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4592021424861735e-07,
+      "loss": 1.1294,
+      "step": 4478
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.441897917696679e-07,
+      "loss": 0.9808,
+      "step": 4479
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.424636324691932e-07,
+      "loss": 1.0542,
+      "step": 4480
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4074173710931804e-07,
+      "loss": 1.1125,
+      "step": 4481
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3902410645028197e-07,
+      "loss": 1.1166,
+      "step": 4482
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.373107412504473e-07,
+      "loss": 0.9994,
+      "step": 4483
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3560164226628666e-07,
+      "loss": 0.9469,
+      "step": 4484
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3389681025239095e-07,
+      "loss": 1.1468,
+      "step": 4485
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3219624596146806e-07,
+      "loss": 1.0531,
+      "step": 4486
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.30499950144344e-07,
+      "loss": 0.7478,
+      "step": 4487
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.288079235499575e-07,
+      "loss": 1.0422,
+      "step": 4488
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2712016692535877e-07,
+      "loss": 0.8277,
+      "step": 4489
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2543668101571946e-07,
+      "loss": 1.0162,
+      "step": 4490
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2375746656432284e-07,
+      "loss": 0.9383,
+      "step": 4491
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2208252431256694e-07,
+      "loss": 1.032,
+      "step": 4492
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2041185499995797e-07,
+      "loss": 0.924,
+      "step": 4493
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.187454593641248e-07,
+      "loss": 0.9154,
+      "step": 4494
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.170833381408045e-07,
+      "loss": 1.0007,
+      "step": 4495
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1542549206384666e-07,
+      "loss": 1.0676,
+      "step": 4496
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1377192186521357e-07,
+      "loss": 0.935,
+      "step": 4497
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1212262827498117e-07,
+      "loss": 0.9367,
+      "step": 4498
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.10477612021336e-07,
+      "loss": 1.0986,
+      "step": 4499
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.088368738305747e-07,
+      "loss": 1.1367,
+      "step": 4500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.07200414427109e-07,
+      "loss": 1.0579,
+      "step": 4501
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0556823453345765e-07,
+      "loss": 1.0745,
+      "step": 4502
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.03940334870253e-07,
+      "loss": 1.0583,
+      "step": 4503
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.023167161562335e-07,
+      "loss": 0.8935,
+      "step": 4504
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0069737910825235e-07,
+      "loss": 1.2312,
+      "step": 4505
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9908232444126993e-07,
+      "loss": 0.974,
+      "step": 4506
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.974715528683547e-07,
+      "loss": 0.9067,
+      "step": 4507
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.958650651006856e-07,
+      "loss": 1.098,
+      "step": 4508
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.942628618475507e-07,
+      "loss": 1.3981,
+      "step": 4509
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9266494381634427e-07,
+      "loss": 1.0267,
+      "step": 4510
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.910713117125719e-07,
+      "loss": 1.0208,
+      "step": 4511
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.894819662398463e-07,
+      "loss": 0.8876,
+      "step": 4512
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.878969080998817e-07,
+      "loss": 1.0157,
+      "step": 4513
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8631613799250615e-07,
+      "loss": 0.8772,
+      "step": 4514
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8473965661565353e-07,
+      "loss": 1.0865,
+      "step": 4515
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8316746466536373e-07,
+      "loss": 1.0191,
+      "step": 4516
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.815995628357771e-07,
+      "loss": 1.0808,
+      "step": 4517
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8003595181914866e-07,
+      "loss": 1.0229,
+      "step": 4518
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.784766323058352e-07,
+      "loss": 0.9344,
+      "step": 4519
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.769216049842982e-07,
+      "loss": 0.8801,
+      "step": 4520
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.753708705411018e-07,
+      "loss": 1.0844,
+      "step": 4521
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.738244296609216e-07,
+      "loss": 1.1017,
+      "step": 4522
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7228228302653037e-07,
+      "loss": 0.9803,
+      "step": 4523
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.707444313188101e-07,
+      "loss": 1.0123,
+      "step": 4524
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6921087521674214e-07,
+      "loss": 0.8673,
+      "step": 4525
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.676816153974149e-07,
+      "loss": 0.8301,
+      "step": 4526
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.661566525360193e-07,
+      "loss": 0.7205,
+      "step": 4527
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.646359873058457e-07,
+      "loss": 0.86,
+      "step": 4528
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.631196203782915e-07,
+      "loss": 0.9789,
+      "step": 4529
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.616075524228545e-07,
+      "loss": 0.965,
+      "step": 4530
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.600997841071329e-07,
+      "loss": 0.8438,
+      "step": 4531
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5859631609682966e-07,
+      "loss": 1.0744,
+      "step": 4532
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.570971490557461e-07,
+      "loss": 0.9449,
+      "step": 4533
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.556022836457861e-07,
+      "loss": 0.9945,
+      "step": 4534
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.54111720526955e-07,
+      "loss": 1.0904,
+      "step": 4535
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.526254603573586e-07,
+      "loss": 1.0079,
+      "step": 4536
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.511435037931986e-07,
+      "loss": 1.0889,
+      "step": 4537
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4966585148878287e-07,
+      "loss": 0.9906,
+      "step": 4538
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4819250409651605e-07,
+      "loss": 1.035,
+      "step": 4539
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4672346226690123e-07,
+      "loss": 0.9133,
+      "step": 4540
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.452587266485418e-07,
+      "loss": 0.7733,
+      "step": 4541
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4379829788813815e-07,
+      "loss": 1.0812,
+      "step": 4542
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.423421766304934e-07,
+      "loss": 0.9935,
+      "step": 4543
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4089036351850447e-07,
+      "loss": 1.1486,
+      "step": 4544
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3944285919316525e-07,
+      "loss": 0.9782,
+      "step": 4545
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3799966429357335e-07,
+      "loss": 0.9817,
+      "step": 4546
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3656077945691802e-07,
+      "loss": 0.9383,
+      "step": 4547
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3512620531848885e-07,
+      "loss": 0.8279,
+      "step": 4548
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3369594251166804e-07,
+      "loss": 0.9633,
+      "step": 4549
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.322699916679394e-07,
+      "loss": 1.1094,
+      "step": 4550
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.308483534168815e-07,
+      "loss": 0.8756,
+      "step": 4551
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.294310283861656e-07,
+      "loss": 0.7927,
+      "step": 4552
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2801801720156224e-07,
+      "loss": 0.8989,
+      "step": 4553
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2660932048693686e-07,
+      "loss": 0.9581,
+      "step": 4554
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2520493886424743e-07,
+      "loss": 1.1256,
+      "step": 4555
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2380487295355025e-07,
+      "loss": 0.9555,
+      "step": 4556
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2240912337299304e-07,
+      "loss": 1.0176,
+      "step": 4557
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.210176907388195e-07,
+      "loss": 1.125,
+      "step": 4558
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1963057566536715e-07,
+      "loss": 0.9785,
+      "step": 4559
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.182477787650694e-07,
+      "loss": 0.8675,
+      "step": 4560
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.168693006484479e-07,
+      "loss": 0.8877,
+      "step": 4561
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1549514192412136e-07,
+      "loss": 0.9926,
+      "step": 4562
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1412530319879888e-07,
+      "loss": 0.9846,
+      "step": 4563
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.127597850772889e-07,
+      "loss": 1.0909,
+      "step": 4564
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1139858816248248e-07,
+      "loss": 1.1596,
+      "step": 4565
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.100417130553678e-07,
+      "loss": 0.9429,
+      "step": 4566
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0868916035502785e-07,
+      "loss": 0.9585,
+      "step": 4567
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0734093065863158e-07,
+      "loss": 0.969,
+      "step": 4568
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0599702456144178e-07,
+      "loss": 0.9667,
+      "step": 4569
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0465744265681263e-07,
+      "loss": 1.078,
+      "step": 4570
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0332218553618888e-07,
+      "loss": 1.1468,
+      "step": 4571
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0199125378910666e-07,
+      "loss": 1.0258,
+      "step": 4572
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0066464800318818e-07,
+      "loss": 1.0162,
+      "step": 4573
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.993423687641516e-07,
+      "loss": 1.0075,
+      "step": 4574
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9802441665580208e-07,
+      "loss": 1.0894,
+      "step": 4575
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.967107922600342e-07,
+      "loss": 0.9292,
+      "step": 4576
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9540149615683068e-07,
+      "loss": 0.9502,
+      "step": 4577
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9409652892426578e-07,
+      "loss": 0.945,
+      "step": 4578
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9279589113850082e-07,
+      "loss": 1.0608,
+      "step": 4579
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9149958337378538e-07,
+      "loss": 1.1677,
+      "step": 4580
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9020760620245938e-07,
+      "loss": 1.0343,
+      "step": 4581
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8891996019494764e-07,
+      "loss": 0.8725,
+      "step": 4582
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8763664591976426e-07,
+      "loss": 1.1508,
+      "step": 4583
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8635766394351384e-07,
+      "loss": 0.8854,
+      "step": 4584
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8508301483088131e-07,
+      "loss": 1.0429,
+      "step": 4585
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8381269914464538e-07,
+      "loss": 1.1417,
+      "step": 4586
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.825467174456652e-07,
+      "loss": 0.9593,
+      "step": 4587
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8128507029289477e-07,
+      "loss": 1.0293,
+      "step": 4588
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8002775824336628e-07,
+      "loss": 0.9551,
+      "step": 4589
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.787747818522001e-07,
+      "loss": 0.983,
+      "step": 4590
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7752614167260484e-07,
+      "loss": 1.0107,
+      "step": 4591
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7628183825587397e-07,
+      "loss": 1.0366,
+      "step": 4592
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7504187215138358e-07,
+      "loss": 1.0384,
+      "step": 4593
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7380624390659572e-07,
+      "loss": 1.0403,
+      "step": 4594
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7257495406705959e-07,
+      "loss": 0.9716,
+      "step": 4595
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.713480031764081e-07,
+      "loss": 0.8286,
+      "step": 4596
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7012539177635457e-07,
+      "loss": 0.94,
+      "step": 4597
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6890712040670277e-07,
+      "loss": 0.9366,
+      "step": 4598
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6769318960533465e-07,
+      "loss": 0.9319,
+      "step": 4599
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6648359990822038e-07,
+      "loss": 0.8141,
+      "step": 4600
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6527835184940722e-07,
+      "loss": 1.0262,
+      "step": 4601
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.640774459610328e-07,
+      "loss": 1.046,
+      "step": 4602
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6288088277331303e-07,
+      "loss": 0.9719,
+      "step": 4603
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6168866281454866e-07,
+      "loss": 0.9027,
+      "step": 4604
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6050078661112078e-07,
+      "loss": 0.828,
+      "step": 4605
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5931725468749436e-07,
+      "loss": 0.9452,
+      "step": 4606
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5813806756621475e-07,
+      "loss": 1.1436,
+      "step": 4607
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5696322576791101e-07,
+      "loss": 0.9607,
+      "step": 4608
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5579272981129268e-07,
+      "loss": 0.9571,
+      "step": 4609
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5462658021314968e-07,
+      "loss": 0.8773,
+      "step": 4610
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5346477748835353e-07,
+      "loss": 1.0528,
+      "step": 4611
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5230732214985833e-07,
+      "loss": 0.9922,
+      "step": 4612
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.511542147086964e-07,
+      "loss": 1.0249,
+      "step": 4613
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.500054556739805e-07,
+      "loss": 0.9343,
+      "step": 4614
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.488610455529038e-07,
+      "loss": 1.0248,
+      "step": 4615
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4772098485074216e-07,
+      "loss": 1.1393,
+      "step": 4616
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4658527407084732e-07,
+      "loss": 0.8525,
+      "step": 4617
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4545391371465046e-07,
+      "loss": 0.966,
+      "step": 4618
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4432690428166528e-07,
+      "loss": 1.1263,
+      "step": 4619
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.432042462694827e-07,
+      "loss": 0.9395,
+      "step": 4620
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4208594017377065e-07,
+      "loss": 1.1035,
+      "step": 4621
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.409719864882775e-07,
+      "loss": 1.1078,
+      "step": 4622
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.39862385704832e-07,
+      "loss": 1.0862,
+      "step": 4623
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.387571383133368e-07,
+      "loss": 0.9725,
+      "step": 4624
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3765624480177488e-07,
+      "loss": 0.9275,
+      "step": 4625
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3655970565620623e-07,
+      "loss": 1.103,
+      "step": 4626
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3546752136076924e-07,
+      "loss": 1.0737,
+      "step": 4627
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3437969239767811e-07,
+      "loss": 0.9734,
+      "step": 4628
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3329621924722536e-07,
+      "loss": 1.0384,
+      "step": 4629
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3221710238778052e-07,
+      "loss": 0.9398,
+      "step": 4630
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3114234229578803e-07,
+      "loss": 0.8297,
+      "step": 4631
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3007193944577056e-07,
+      "loss": 0.9001,
+      "step": 4632
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2900589431032563e-07,
+      "loss": 1.0897,
+      "step": 4633
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.279442073601278e-07,
+      "loss": 0.8555,
+      "step": 4634
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2688687906392772e-07,
+      "loss": 1.0319,
+      "step": 4635
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.258339098885497e-07,
+      "loss": 0.9444,
+      "step": 4636
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2478530029889635e-07,
+      "loss": 1.0383,
+      "step": 4637
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2374105075794175e-07,
+      "loss": 0.844,
+      "step": 4638
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2270116172673818e-07,
+      "loss": 0.819,
+      "step": 4639
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2166563366441287e-07,
+      "loss": 0.9454,
+      "step": 4640
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.206344670281645e-07,
+      "loss": 0.9893,
+      "step": 4641
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1960766227326782e-07,
+      "loss": 0.8984,
+      "step": 4642
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1858521985307126e-07,
+      "loss": 1.0702,
+      "step": 4643
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.175671402190004e-07,
+      "loss": 0.7939,
+      "step": 4644
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1655342382055013e-07,
+      "loss": 0.9141,
+      "step": 4645
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.155440711052902e-07,
+      "loss": 1.1089,
+      "step": 4646
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1453908251886636e-07,
+      "loss": 0.9433,
+      "step": 4647
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1353845850499478e-07,
+      "loss": 1.0607,
+      "step": 4648
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1254219950546319e-07,
+      "loss": 0.9416,
+      "step": 4649
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1155030596013638e-07,
+      "loss": 0.9939,
+      "step": 4650
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.105627783069485e-07,
+      "loss": 1.1134,
+      "step": 4651
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0957961698190856e-07,
+      "loss": 0.9836,
+      "step": 4652
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0860082241909487e-07,
+      "loss": 1.089,
+      "step": 4653
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0762639505065952e-07,
+      "loss": 1.0105,
+      "step": 4654
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0665633530682618e-07,
+      "loss": 0.8486,
+      "step": 4655
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.056906436158911e-07,
+      "loss": 0.7145,
+      "step": 4656
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.047293204042199e-07,
+      "loss": 0.9762,
+      "step": 4657
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0377236609625086e-07,
+      "loss": 1.0721,
+      "step": 4658
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0281978111449375e-07,
+      "loss": 1.1299,
+      "step": 4659
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0187156587952662e-07,
+      "loss": 1.0519,
+      "step": 4660
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0092772081000235e-07,
+      "loss": 1.1049,
+      "step": 4661
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.998824632263981e-08,
+      "loss": 0.7968,
+      "step": 4662
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.905314283223166e-08,
+      "loss": 0.8866,
+      "step": 4663
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.812241075163986e-08,
+      "loss": 0.9941,
+      "step": 4664
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.71960504917957e-08,
+      "loss": 0.9792,
+      "step": 4665
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.627406246169978e-08,
+      "loss": 0.9207,
+      "step": 4666
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.535644706842318e-08,
+      "loss": 1.0863,
+      "step": 4667
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.444320471710732e-08,
+      "loss": 0.9796,
+      "step": 4668
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.353433581096082e-08,
+      "loss": 1.1005,
+      "step": 4669
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.26298407512627e-08,
+      "loss": 1.0476,
+      "step": 4670
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.172971993735902e-08,
+      "loss": 0.9958,
+      "step": 4671
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.083397376666858e-08,
+      "loss": 0.956,
+      "step": 4672
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.994260263467502e-08,
+      "loss": 0.9154,
+      "step": 4673
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.90556069349302e-08,
+      "loss": 0.8514,
+      "step": 4674
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.817298705905642e-08,
+      "loss": 1.0665,
+      "step": 4675
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.729474339674415e-08,
+      "loss": 1.0867,
+      "step": 4676
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.64208763357488e-08,
+      "loss": 0.8971,
+      "step": 4677
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.555138626189619e-08,
+      "loss": 0.9764,
+      "step": 4678
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.468627355907811e-08,
+      "loss": 1.0102,
+      "step": 4679
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.382553860925458e-08,
+      "loss": 1.0877,
+      "step": 4680
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.296918179245161e-08,
+      "loss": 1.0391,
+      "step": 4681
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.211720348676456e-08,
+      "loss": 1.0199,
+      "step": 4682
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.12696040683525e-08,
+      "loss": 1.1327,
+      "step": 4683
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.042638391144276e-08,
+      "loss": 0.9846,
+      "step": 4684
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.958754338833085e-08,
+      "loss": 0.927,
+      "step": 4685
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.8753082869375e-08,
+      "loss": 1.0512,
+      "step": 4686
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.792300272300268e-08,
+      "loss": 1.041,
+      "step": 4687
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.70973033157052e-08,
+      "loss": 0.9629,
+      "step": 4688
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.627598501204092e-08,
+      "loss": 0.8201,
+      "step": 4689
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.545904817463423e-08,
+      "loss": 0.9963,
+      "step": 4690
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.464649316417438e-08,
+      "loss": 0.9851,
+      "step": 4691
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.383832033941551e-08,
+      "loss": 0.9389,
+      "step": 4692
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.303453005717665e-08,
+      "loss": 0.9573,
+      "step": 4693
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.2235122672345e-08,
+      "loss": 0.8138,
+      "step": 4694
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.144009853786826e-08,
+      "loss": 1.0321,
+      "step": 4695
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.064945800476342e-08,
+      "loss": 0.9166,
+      "step": 4696
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.986320142210678e-08,
+      "loss": 0.9356,
+      "step": 4697
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.908132913704291e-08,
+      "loss": 0.9691,
+      "step": 4698
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.830384149478009e-08,
+      "loss": 0.9734,
+      "step": 4699
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.75307388385893e-08,
+      "loss": 1.029,
+      "step": 4700
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.67620215098086e-08,
+      "loss": 0.9111,
+      "step": 4701
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.59976898478354e-08,
+      "loss": 0.9566,
+      "step": 4702
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.523774419013418e-08,
+      "loss": 0.9245,
+      "step": 4703
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.448218487223101e-08,
+      "loss": 0.9601,
+      "step": 4704
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.373101222771793e-08,
+      "loss": 0.9387,
+      "step": 4705
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.298422658824521e-08,
+      "loss": 0.7524,
+      "step": 4706
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.224182828353243e-08,
+      "loss": 1.0387,
+      "step": 4707
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.150381764135849e-08,
+      "loss": 1.1361,
+      "step": 4708
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.077019498756497e-08,
+      "loss": 1.0068,
+      "step": 4709
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.004096064605613e-08,
+      "loss": 0.8382,
+      "step": 4710
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.9316114938801076e-08,
+      "loss": 0.9117,
+      "step": 4711
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.859565818582824e-08,
+      "loss": 0.9724,
+      "step": 4712
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.787959070523097e-08,
+      "loss": 1.1487,
+      "step": 4713
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.7167912813160806e-08,
+      "loss": 1.0419,
+      "step": 4714
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.6460624823836405e-08,
+      "loss": 1.045,
+      "step": 4715
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.575772704953464e-08,
+      "loss": 0.868,
+      "step": 4716
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.505921980059503e-08,
+      "loss": 1.0737,
+      "step": 4717
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.436510338541756e-08,
+      "loss": 0.904,
+      "step": 4718
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.367537811046486e-08,
+      "loss": 0.8965,
+      "step": 4719
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.299004428026222e-08,
+      "loss": 0.973,
+      "step": 4720
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.230910219739205e-08,
+      "loss": 1.1272,
+      "step": 4721
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.163255216250162e-08,
+      "loss": 1.0089,
+      "step": 4722
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.096039447429535e-08,
+      "loss": 0.9338,
+      "step": 4723
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.0292629429543606e-08,
+      "loss": 1.0158,
+      "step": 4724
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.962925732307278e-08,
+      "loss": 0.9353,
+      "step": 4725
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.89702784477708e-08,
+      "loss": 0.9408,
+      "step": 4726
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.8315693094584945e-08,
+      "loss": 1.041,
+      "step": 4727
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.766550155252736e-08,
+      "loss": 1.0371,
+      "step": 4728
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.7019704108665074e-08,
+      "loss": 1.0595,
+      "step": 4729
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.637830104812557e-08,
+      "loss": 1.0532,
+      "step": 4730
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.57412926541001e-08,
+      "loss": 0.9641,
+      "step": 4731
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.510867920783479e-08,
+      "loss": 0.9299,
+      "step": 4732
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.4480460988639564e-08,
+      "loss": 0.9705,
+      "step": 4733
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.385663827388031e-08,
+      "loss": 0.9533,
+      "step": 4734
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.3237211338983396e-08,
+      "loss": 0.9441,
+      "step": 4735
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.2622180457435604e-08,
+      "loss": 1.0563,
+      "step": 4736
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.2011545900781936e-08,
+      "loss": 0.9385,
+      "step": 4737
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.140530793862674e-08,
+      "loss": 0.9011,
+      "step": 4738
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.0803466838631454e-08,
+      "loss": 1.0601,
+      "step": 4739
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.020602286651687e-08,
+      "loss": 1.0378,
+      "step": 4740
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.9612976286065305e-08,
+      "loss": 0.8963,
+      "step": 4741
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.90243273591151e-08,
+      "loss": 1.0219,
+      "step": 4742
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.8440076345561685e-08,
+      "loss": 0.9768,
+      "step": 4743
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.786022350335983e-08,
+      "loss": 0.8747,
+      "step": 4744
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.728476908852585e-08,
+      "loss": 1.057,
+      "step": 4745
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.6713713355127634e-08,
+      "loss": 0.8301,
+      "step": 4746
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.6147056555296825e-08,
+      "loss": 0.8379,
+      "step": 4747
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5584798939219956e-08,
+      "loss": 0.9817,
+      "step": 4748
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.502694075514179e-08,
+      "loss": 0.9591,
+      "step": 4749
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.44734822493642e-08,
+      "loss": 1.1137,
+      "step": 4750
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.392442366624615e-08,
+      "loss": 0.9957,
+      "step": 4751
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.3379765248207077e-08,
+      "loss": 1.0948,
+      "step": 4752
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.283950723572127e-08,
+      "loss": 0.9653,
+      "step": 4753
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.230364986731904e-08,
+      "loss": 1.0826,
+      "step": 4754
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.177219337958892e-08,
+      "loss": 0.9921,
+      "step": 4755
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.124513800717877e-08,
+      "loss": 0.9062,
+      "step": 4756
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.0722483982789095e-08,
+      "loss": 1.1278,
+      "step": 4757
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.020423153717866e-08,
+      "loss": 0.937,
+      "step": 4758
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.969038089916443e-08,
+      "loss": 0.9776,
+      "step": 4759
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.918093229561936e-08,
+      "loss": 1.0925,
+      "step": 4760
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8675885951472415e-08,
+      "loss": 1.0439,
+      "step": 4761
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8175242089706345e-08,
+      "loss": 1.0543,
+      "step": 4762
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.767900093136544e-08,
+      "loss": 1.0822,
+      "step": 4763
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7187162695545554e-08,
+      "loss": 1.0738,
+      "step": 4764
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.669972759940076e-08,
+      "loss": 1.0169,
+      "step": 4765
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6216695858141127e-08,
+      "loss": 1.0288,
+      "step": 4766
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.573806768503162e-08,
+      "loss": 1.1036,
+      "step": 4767
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5263843291393197e-08,
+      "loss": 1.0483,
+      "step": 4768
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.479402288660282e-08,
+      "loss": 0.9493,
+      "step": 4769
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.4328606678093446e-08,
+      "loss": 0.9708,
+      "step": 4770
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.386759487135293e-08,
+      "loss": 0.9185,
+      "step": 4771
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3410987669924002e-08,
+      "loss": 1.0465,
+      "step": 4772
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2958785275406514e-08,
+      "loss": 0.936,
+      "step": 4773
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2510987887451875e-08,
+      "loss": 0.9937,
+      "step": 4774
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.206759570377193e-08,
+      "loss": 0.9146,
+      "step": 4775
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.162860892012786e-08,
+      "loss": 0.8509,
+      "step": 4776
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1194027730341294e-08,
+      "loss": 0.9369,
+      "step": 4777
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0763852326284307e-08,
+      "loss": 1.0245,
+      "step": 4778
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.033808289788608e-08,
+      "loss": 1.1259,
+      "step": 4779
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.991671963312958e-08,
+      "loss": 1.0178,
+      "step": 4780
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.949976271805376e-08,
+      "loss": 0.9668,
+      "step": 4781
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.908721233675026e-08,
+      "loss": 1.0623,
+      "step": 4782
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8679068671364488e-08,
+      "loss": 0.9642,
+      "step": 4783
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.827533190210118e-08,
+      "loss": 1.0327,
+      "step": 4784
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.7876002207213304e-08,
+      "loss": 0.9514,
+      "step": 4785
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7481079763010943e-08,
+      "loss": 1.1541,
+      "step": 4786
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7090564743857952e-08,
+      "loss": 1.0591,
+      "step": 4787
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6704457322173073e-08,
+      "loss": 1.0465,
+      "step": 4788
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6322757668428836e-08,
+      "loss": 0.9731,
+      "step": 4789
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5945465951149318e-08,
+      "loss": 0.9496,
+      "step": 4790
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.557258233691572e-08,
+      "loss": 0.9682,
+      "step": 4791
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.520410699035968e-08,
+      "loss": 1.1103,
+      "step": 4792
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4840040074169947e-08,
+      "loss": 1.076,
+      "step": 4793
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4480381749087945e-08,
+      "loss": 1.0515,
+      "step": 4794
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4125132173905542e-08,
+      "loss": 1.0661,
+      "step": 4795
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3774291505472826e-08,
+      "loss": 0.9392,
+      "step": 4796
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3427859898689221e-08,
+      "loss": 0.8626,
+      "step": 4797
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3085837506510158e-08,
+      "loss": 1.009,
+      "step": 4798
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2748224479943727e-08,
+      "loss": 0.8704,
+      "step": 4799
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2415020968049585e-08,
+      "loss": 0.8965,
+      "step": 4800
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2086227117943383e-08,
+      "loss": 1.1428,
+      "step": 4801
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1761843074792334e-08,
+      "loss": 1.0424,
+      "step": 4802
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1441868981815207e-08,
+      "loss": 0.9363,
+      "step": 4803
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1126304980285663e-08,
+      "loss": 1.0513,
+      "step": 4804
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0815151209530029e-08,
+      "loss": 0.891,
+      "step": 4805
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0508407806927302e-08,
+      "loss": 1.025,
+      "step": 4806
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0206074907909147e-08,
+      "loss": 0.9666,
+      "step": 4807
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.908152645958791e-09,
+      "loss": 0.8376,
+      "step": 4808
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.614641152615457e-09,
+      "loss": 0.9652,
+      "step": 4809
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.325540557465484e-09,
+      "loss": 0.9855,
+      "step": 4810
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.040850988153437e-09,
+      "loss": 0.9556,
+      "step": 4811
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.760572570373215e-09,
+      "loss": 1.0004,
+      "step": 4812
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.484705427871387e-09,
+      "loss": 1.075,
+      "step": 4813
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.213249682447188e-09,
+      "loss": 0.9004,
+      "step": 4814
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.946205453953638e-09,
+      "loss": 1.1016,
+      "step": 4815
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.683572860291976e-09,
+      "loss": 1.1398,
+      "step": 4816
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.425352017420562e-09,
+      "loss": 1.0688,
+      "step": 4817
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.171543039345974e-09,
+      "loss": 0.8768,
+      "step": 4818
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.9221460381296845e-09,
+      "loss": 1.2175,
+      "step": 4819
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.6771611238836135e-09,
+      "loss": 0.983,
+      "step": 4820
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.4365884047723486e-09,
+      "loss": 0.8511,
+      "step": 4821
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.200427987012037e-09,
+      "loss": 0.9236,
+      "step": 4822
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.968679974870384e-09,
+      "loss": 0.9244,
+      "step": 4823
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.741344470667765e-09,
+      "loss": 0.9828,
+      "step": 4824
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.518421574777222e-09,
+      "loss": 0.96,
+      "step": 4825
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.299911385620027e-09,
+      "loss": 0.9706,
+      "step": 4826
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.08581399967345e-09,
+      "loss": 1.1539,
+      "step": 4827
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.8761295114641005e-09,
+      "loss": 0.9563,
+      "step": 4828
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.670858013570146e-09,
+      "loss": 0.9208,
+      "step": 4829
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.469999596622421e-09,
+      "loss": 0.979,
+      "step": 4830
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.273554349303321e-09,
+      "loss": 0.9115,
+      "step": 4831
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.081522358345691e-09,
+      "loss": 1.2024,
+      "step": 4832
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.8939037085339305e-09,
+      "loss": 0.8696,
+      "step": 4833
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.7106984827040003e-09,
+      "loss": 1.0674,
+      "step": 4834
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5319067617445302e-09,
+      "loss": 1.027,
+      "step": 4835
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.3575286245957073e-09,
+      "loss": 1.0577,
+      "step": 4836
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.1875641482459476e-09,
+      "loss": 1.1837,
+      "step": 4837
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.0220134077374452e-09,
+      "loss": 0.9928,
+      "step": 4838
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.8608764761639542e-09,
+      "loss": 1.0035,
+      "step": 4839
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.7041534246696756e-09,
+      "loss": 1.0547,
+      "step": 4840
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.5518443224492594e-09,
+      "loss": 0.942,
+      "step": 4841
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.403949236750025e-09,
+      "loss": 1.0756,
+      "step": 4842
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.2604682328697393e-09,
+      "loss": 1.0222,
+      "step": 4843
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.1214013741577276e-09,
+      "loss": 0.9585,
+      "step": 4844
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9867487220126546e-09,
+      "loss": 1.0252,
+      "step": 4845
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.856510335886963e-09,
+      "loss": 0.9541,
+      "step": 4846
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.7306862732813234e-09,
+      "loss": 0.8279,
+      "step": 4847
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.609276589750186e-09,
+      "loss": 0.9543,
+      "step": 4848
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.492281338897339e-09,
+      "loss": 0.9464,
+      "step": 4849
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3797005723781287e-09,
+      "loss": 1.0101,
+      "step": 4850
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2715343398972401e-09,
+      "loss": 1.0659,
+      "step": 4851
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1677826892131373e-09,
+      "loss": 0.9567,
+      "step": 4852
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0684456661325116e-09,
+      "loss": 1.0921,
+      "step": 4853
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.73523314515834e-10,
+      "loss": 1.1943,
+      "step": 4854
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.830156762706932e-10,
+      "loss": 0.8882,
+      "step": 4855
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.969227913584565e-10,
+      "loss": 0.8796,
+      "step": 4856
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.152446977898297e-10,
+      "loss": 0.9055,
+      "step": 4857
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.379814316281873e-10,
+      "loss": 0.9697,
+      "step": 4858
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.651330269840216e-10,
+      "loss": 0.9389,
+      "step": 4859
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.966995160238242e-10,
+      "loss": 0.9196,
+      "step": 4860
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.326809289600942e-10,
+      "loss": 0.9326,
+      "step": 4861
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.7307729405799965e-10,
+      "loss": 1.0629,
+      "step": 4862
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.178886376331569e-10,
+      "loss": 0.9185,
+      "step": 4863
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.6711498405385117e-10,
+      "loss": 1.0158,
+      "step": 4864
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.2075635573659549e-10,
+      "loss": 0.9392,
+      "step": 4865
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.7881277314835132e-10,
+      "loss": 0.9415,
+      "step": 4866
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4128425480763874e-10,
+      "loss": 1.026,
+      "step": 4867
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.081708172856466e-10,
+      "loss": 1.1676,
+      "step": 4868
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.947247520179169e-11,
+      "loss": 1.0403,
+      "step": 4869
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.518924122704938e-11,
+      "loss": 0.9475,
+      "step": 4870
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.532112608062299e-11,
+      "loss": 0.9126,
+      "step": 4871
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9868138537715297e-11,
+      "loss": 1.1289,
+      "step": 4872
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.830285418426343e-12,
+      "loss": 0.914,
+      "step": 4873
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.207571598855651e-12,
+      "loss": 0.9988,
+      "step": 4874
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 0.8948,
+      "step": 4875
+    },
+    {
+      "epoch": 3.0,
+      "step": 4875,
+      "total_flos": 1.955461129394258e+17,
+      "train_loss": 1.2763952842003259,
+      "train_runtime": 7409.0952,
+      "train_samples_per_second": 21.056,
+      "train_steps_per_second": 0.658
+    }
+  ],
+  "max_steps": 4875,
+  "num_train_epochs": 3,
+  "total_flos": 1.955461129394258e+17,
+  "trial_name": null,
+  "trial_params": null
+}