diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,29257 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.998615171564856,
+  "global_step": 4872,
+  "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": 2.1143,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.72108843537415e-07,
+      "loss": 2.2512,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0816326530612243e-07,
+      "loss": 1.9934,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.4421768707483e-07,
+      "loss": 2.0884,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.802721088435376e-07,
+      "loss": 1.9564,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.163265306122449e-07,
+      "loss": 2.1002,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.523809523809525e-07,
+      "loss": 2.0199,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.08843537414966e-06,
+      "loss": 2.0489,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2244897959183673e-06,
+      "loss": 1.9815,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3605442176870751e-06,
+      "loss": 1.9339,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4965986394557825e-06,
+      "loss": 2.0624,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6326530612244897e-06,
+      "loss": 2.0178,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7687074829931975e-06,
+      "loss": 1.9643,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.904761904761905e-06,
+      "loss": 1.9411,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0408163265306125e-06,
+      "loss": 1.9514,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.17687074829932e-06,
+      "loss": 1.9679,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3129251700680273e-06,
+      "loss": 1.9387,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4489795918367347e-06,
+      "loss": 1.9883,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5850340136054425e-06,
+      "loss": 1.9818,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7210884353741503e-06,
+      "loss": 1.9598,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 1.9382,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.993197278911565e-06,
+      "loss": 2.0274,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1292517006802725e-06,
+      "loss": 1.9687,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2653061224489794e-06,
+      "loss": 1.9527,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4013605442176872e-06,
+      "loss": 1.9152,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.537414965986395e-06,
+      "loss": 1.9232,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6734693877551024e-06,
+      "loss": 1.9105,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.80952380952381e-06,
+      "loss": 1.908,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.945578231292517e-06,
+      "loss": 2.0166,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.081632653061225e-06,
+      "loss": 1.9143,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.217687074829933e-06,
+      "loss": 1.9373,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.35374149659864e-06,
+      "loss": 1.9161,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.489795918367348e-06,
+      "loss": 1.9405,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6258503401360546e-06,
+      "loss": 1.9096,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.761904761904762e-06,
+      "loss": 1.8128,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.897959183673469e-06,
+      "loss": 1.9087,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.034013605442177e-06,
+      "loss": 1.915,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.170068027210885e-06,
+      "loss": 1.8744,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.306122448979593e-06,
+      "loss": 1.8804,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.442176870748301e-06,
+      "loss": 1.922,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.578231292517007e-06,
+      "loss": 1.8589,
+      "step": 41
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.7142857142857145e-06,
+      "loss": 1.915,
+      "step": 42
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.850340136054422e-06,
+      "loss": 1.8514,
+      "step": 43
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.98639455782313e-06,
+      "loss": 1.8794,
+      "step": 44
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.122448979591837e-06,
+      "loss": 1.866,
+      "step": 45
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.258503401360545e-06,
+      "loss": 1.9003,
+      "step": 46
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.394557823129253e-06,
+      "loss": 1.9631,
+      "step": 47
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.530612244897959e-06,
+      "loss": 1.8717,
+      "step": 48
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.921,
+      "step": 49
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.8027210884353745e-06,
+      "loss": 1.8965,
+      "step": 50
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.938775510204082e-06,
+      "loss": 1.8698,
+      "step": 51
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.07482993197279e-06,
+      "loss": 1.9618,
+      "step": 52
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.210884353741497e-06,
+      "loss": 1.9019,
+      "step": 53
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.346938775510205e-06,
+      "loss": 1.803,
+      "step": 54
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.482993197278913e-06,
+      "loss": 1.8846,
+      "step": 55
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.61904761904762e-06,
+      "loss": 1.8118,
+      "step": 56
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.755102040816327e-06,
+      "loss": 1.8313,
+      "step": 57
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.891156462585034e-06,
+      "loss": 1.8463,
+      "step": 58
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.027210884353741e-06,
+      "loss": 1.8718,
+      "step": 59
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.16326530612245e-06,
+      "loss": 1.9074,
+      "step": 60
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.299319727891157e-06,
+      "loss": 1.8211,
+      "step": 61
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.435374149659866e-06,
+      "loss": 1.8159,
+      "step": 62
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.571428571428571e-06,
+      "loss": 1.8069,
+      "step": 63
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.70748299319728e-06,
+      "loss": 1.9081,
+      "step": 64
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.843537414965987e-06,
+      "loss": 1.8763,
+      "step": 65
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.979591836734695e-06,
+      "loss": 1.8282,
+      "step": 66
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.115646258503402e-06,
+      "loss": 1.8345,
+      "step": 67
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.251700680272109e-06,
+      "loss": 1.8293,
+      "step": 68
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.387755102040818e-06,
+      "loss": 1.8644,
+      "step": 69
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.523809523809525e-06,
+      "loss": 1.7992,
+      "step": 70
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.659863945578232e-06,
+      "loss": 1.7735,
+      "step": 71
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.795918367346939e-06,
+      "loss": 1.8985,
+      "step": 72
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.931972789115647e-06,
+      "loss": 1.8368,
+      "step": 73
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0068027210884354e-05,
+      "loss": 1.8625,
+      "step": 74
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0204081632653063e-05,
+      "loss": 1.8241,
+      "step": 75
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.034013605442177e-05,
+      "loss": 1.8418,
+      "step": 76
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0476190476190477e-05,
+      "loss": 1.8482,
+      "step": 77
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0612244897959186e-05,
+      "loss": 1.8541,
+      "step": 78
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0748299319727893e-05,
+      "loss": 1.8734,
+      "step": 79
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0884353741496601e-05,
+      "loss": 1.8626,
+      "step": 80
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1020408163265306e-05,
+      "loss": 1.9158,
+      "step": 81
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1156462585034013e-05,
+      "loss": 1.8395,
+      "step": 82
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1292517006802722e-05,
+      "loss": 1.8508,
+      "step": 83
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1428571428571429e-05,
+      "loss": 1.8504,
+      "step": 84
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1564625850340136e-05,
+      "loss": 1.8661,
+      "step": 85
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1700680272108845e-05,
+      "loss": 1.7775,
+      "step": 86
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1836734693877552e-05,
+      "loss": 1.7705,
+      "step": 87
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.197278911564626e-05,
+      "loss": 1.8362,
+      "step": 88
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2108843537414967e-05,
+      "loss": 1.825,
+      "step": 89
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2244897959183674e-05,
+      "loss": 1.8525,
+      "step": 90
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2380952380952383e-05,
+      "loss": 1.8412,
+      "step": 91
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.251700680272109e-05,
+      "loss": 1.773,
+      "step": 92
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2653061224489798e-05,
+      "loss": 1.846,
+      "step": 93
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2789115646258505e-05,
+      "loss": 1.8183,
+      "step": 94
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2925170068027212e-05,
+      "loss": 1.7856,
+      "step": 95
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3061224489795918e-05,
+      "loss": 1.7527,
+      "step": 96
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3197278911564626e-05,
+      "loss": 1.7853,
+      "step": 97
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 1.7701,
+      "step": 98
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3469387755102042e-05,
+      "loss": 1.7964,
+      "step": 99
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3605442176870749e-05,
+      "loss": 1.7739,
+      "step": 100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3741496598639456e-05,
+      "loss": 1.7833,
+      "step": 101
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3877551020408165e-05,
+      "loss": 1.8123,
+      "step": 102
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4013605442176872e-05,
+      "loss": 1.8144,
+      "step": 103
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.414965986394558e-05,
+      "loss": 1.8539,
+      "step": 104
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4285714285714287e-05,
+      "loss": 1.8149,
+      "step": 105
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4421768707482994e-05,
+      "loss": 1.8043,
+      "step": 106
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4557823129251703e-05,
+      "loss": 1.7132,
+      "step": 107
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.469387755102041e-05,
+      "loss": 1.7786,
+      "step": 108
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4829931972789118e-05,
+      "loss": 1.7971,
+      "step": 109
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4965986394557825e-05,
+      "loss": 1.8154,
+      "step": 110
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.510204081632653e-05,
+      "loss": 1.8176,
+      "step": 111
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.523809523809524e-05,
+      "loss": 1.8765,
+      "step": 112
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5374149659863945e-05,
+      "loss": 1.7968,
+      "step": 113
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5510204081632655e-05,
+      "loss": 1.7189,
+      "step": 114
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5646258503401362e-05,
+      "loss": 1.8449,
+      "step": 115
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.578231292517007e-05,
+      "loss": 1.8743,
+      "step": 116
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5918367346938776e-05,
+      "loss": 1.6936,
+      "step": 117
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6054421768707483e-05,
+      "loss": 1.8581,
+      "step": 118
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6190476190476193e-05,
+      "loss": 1.8064,
+      "step": 119
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.63265306122449e-05,
+      "loss": 1.7922,
+      "step": 120
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6462585034013607e-05,
+      "loss": 1.7767,
+      "step": 121
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6598639455782314e-05,
+      "loss": 1.7612,
+      "step": 122
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.673469387755102e-05,
+      "loss": 1.7635,
+      "step": 123
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.687074829931973e-05,
+      "loss": 1.8011,
+      "step": 124
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7006802721088435e-05,
+      "loss": 1.8381,
+      "step": 125
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7142857142857142e-05,
+      "loss": 1.8368,
+      "step": 126
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7278911564625852e-05,
+      "loss": 1.7674,
+      "step": 127
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.741496598639456e-05,
+      "loss": 1.8235,
+      "step": 128
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7551020408163266e-05,
+      "loss": 1.8251,
+      "step": 129
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7687074829931973e-05,
+      "loss": 1.78,
+      "step": 130
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.782312925170068e-05,
+      "loss": 1.7815,
+      "step": 131
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.795918367346939e-05,
+      "loss": 1.8745,
+      "step": 132
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8095238095238097e-05,
+      "loss": 1.7771,
+      "step": 133
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8231292517006804e-05,
+      "loss": 1.79,
+      "step": 134
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.836734693877551e-05,
+      "loss": 1.7876,
+      "step": 135
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8503401360544218e-05,
+      "loss": 1.8083,
+      "step": 136
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.863945578231293e-05,
+      "loss": 1.828,
+      "step": 137
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8775510204081636e-05,
+      "loss": 1.6983,
+      "step": 138
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8911564625850343e-05,
+      "loss": 1.8437,
+      "step": 139
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.904761904761905e-05,
+      "loss": 1.797,
+      "step": 140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9183673469387756e-05,
+      "loss": 1.793,
+      "step": 141
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9319727891156463e-05,
+      "loss": 1.7969,
+      "step": 142
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.945578231292517e-05,
+      "loss": 1.8334,
+      "step": 143
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9591836734693877e-05,
+      "loss": 1.7748,
+      "step": 144
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9727891156462588e-05,
+      "loss": 1.7992,
+      "step": 145
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9863945578231295e-05,
+      "loss": 1.722,
+      "step": 146
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 1.8214,
+      "step": 147
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999999778962425e-05,
+      "loss": 1.7524,
+      "step": 148
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999991158497964e-05,
+      "loss": 1.8679,
+      "step": 149
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999998010662408e-05,
+      "loss": 1.8354,
+      "step": 150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999964634007485e-05,
+      "loss": 1.8372,
+      "step": 151
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999944740655016e-05,
+      "loss": 1.7316,
+      "step": 152
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999920426575468e-05,
+      "loss": 1.7528,
+      "step": 153
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999989169177959e-05,
+      "loss": 1.7926,
+      "step": 154
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999858536280083e-05,
+      "loss": 1.7758,
+      "step": 155
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999982096009161e-05,
+      "loss": 1.7616,
+      "step": 156
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999778963230775e-05,
+      "loss": 1.7439,
+      "step": 157
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999732545716147e-05,
+      "loss": 1.825,
+      "step": 158
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999968170756825e-05,
+      "loss": 1.7064,
+      "step": 159
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999626448809554e-05,
+      "loss": 1.7898,
+      "step": 160
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999566769464483e-05,
+      "loss": 1.6945,
+      "step": 161
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999502669559432e-05,
+      "loss": 1.8007,
+      "step": 162
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999943414912273e-05,
+      "loss": 1.761,
+      "step": 163
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999361208184666e-05,
+      "loss": 1.7337,
+      "step": 164
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999928384677749e-05,
+      "loss": 1.6781,
+      "step": 165
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999202064935398e-05,
+      "loss": 1.7285,
+      "step": 166
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999115862694547e-05,
+      "loss": 1.6908,
+      "step": 167
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999025240093045e-05,
+      "loss": 1.8349,
+      "step": 168
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999893019717095e-05,
+      "loss": 1.7074,
+      "step": 169
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998830733970282e-05,
+      "loss": 1.7208,
+      "step": 170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999872685053501e-05,
+      "loss": 1.8004,
+      "step": 171
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999861854691106e-05,
+      "loss": 1.8122,
+      "step": 172
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998505823146305e-05,
+      "loss": 1.811,
+      "step": 173
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998388679290583e-05,
+      "loss": 1.7624,
+      "step": 174
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999826711539568e-05,
+      "loss": 1.7029,
+      "step": 175
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998141131515333e-05,
+      "loss": 1.6969,
+      "step": 176
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998010727705237e-05,
+      "loss": 1.7704,
+      "step": 177
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999787590402304e-05,
+      "loss": 1.7334,
+      "step": 178
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999773666052835e-05,
+      "loss": 1.7564,
+      "step": 179
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999759299728272e-05,
+      "loss": 1.7488,
+      "step": 180
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999744491434965e-05,
+      "loss": 1.7691,
+      "step": 181
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999729241179462e-05,
+      "loss": 1.7707,
+      "step": 182
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997135489685035e-05,
+      "loss": 1.6795,
+      "step": 183
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996974148090274e-05,
+      "loss": 1.7452,
+      "step": 184
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999680838708166e-05,
+      "loss": 1.6999,
+      "step": 185
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996638206732467e-05,
+      "loss": 1.7271,
+      "step": 186
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999646360711794e-05,
+      "loss": 1.6804,
+      "step": 187
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996284588315248e-05,
+      "loss": 1.723,
+      "step": 188
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996101150403543e-05,
+      "loss": 1.7405,
+      "step": 189
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995913293463918e-05,
+      "loss": 1.7658,
+      "step": 190
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995721017579416e-05,
+      "loss": 1.7829,
+      "step": 191
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995524322835035e-05,
+      "loss": 1.7871,
+      "step": 192
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995323209317733e-05,
+      "loss": 1.7002,
+      "step": 193
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999511767711642e-05,
+      "loss": 1.8139,
+      "step": 194
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999490772632195e-05,
+      "loss": 1.7383,
+      "step": 195
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999469335702714e-05,
+      "loss": 1.7966,
+      "step": 196
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999447456932676e-05,
+      "loss": 1.71,
+      "step": 197
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994251363317522e-05,
+      "loss": 1.7346,
+      "step": 198
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994023739098112e-05,
+      "loss": 1.7595,
+      "step": 199
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999379169676915e-05,
+      "loss": 1.7667,
+      "step": 200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993555236433216e-05,
+      "loss": 1.8382,
+      "step": 201
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993314358194843e-05,
+      "loss": 1.7192,
+      "step": 202
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999306906216052e-05,
+      "loss": 1.7412,
+      "step": 203
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992819348438687e-05,
+      "loss": 1.7883,
+      "step": 204
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999256521713973e-05,
+      "loss": 1.7593,
+      "step": 205
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992306668376e-05,
+      "loss": 1.7224,
+      "step": 206
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992043702261795e-05,
+      "loss": 1.7882,
+      "step": 207
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991776318913365e-05,
+      "loss": 1.7685,
+      "step": 208
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999150451844891e-05,
+      "loss": 1.7526,
+      "step": 209
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991228300988586e-05,
+      "loss": 1.6788,
+      "step": 210
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990947666654507e-05,
+      "loss": 1.6839,
+      "step": 211
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999066261557073e-05,
+      "loss": 1.8029,
+      "step": 212
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990373147863276e-05,
+      "loss": 1.7238,
+      "step": 213
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990079263660106e-05,
+      "loss": 1.7412,
+      "step": 214
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989780963091133e-05,
+      "loss": 1.7246,
+      "step": 215
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989478246288242e-05,
+      "loss": 1.7952,
+      "step": 216
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998917111338525e-05,
+      "loss": 1.7567,
+      "step": 217
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988859564517927e-05,
+      "loss": 1.7631,
+      "step": 218
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988543599824008e-05,
+      "loss": 1.74,
+      "step": 219
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998822321944317e-05,
+      "loss": 1.7128,
+      "step": 220
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998789842351705e-05,
+      "loss": 1.7748,
+      "step": 221
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987569212189224e-05,
+      "loss": 1.7657,
+      "step": 222
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987235585605237e-05,
+      "loss": 1.7459,
+      "step": 223
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998689754391257e-05,
+      "loss": 1.7734,
+      "step": 224
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998655508726067e-05,
+      "loss": 1.7516,
+      "step": 225
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998620821580092e-05,
+      "loss": 1.7717,
+      "step": 226
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998585692968667e-05,
+      "loss": 1.7904,
+      "step": 227
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985501229073213e-05,
+      "loss": 1.7105,
+      "step": 228
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985141114117793e-05,
+      "loss": 1.7823,
+      "step": 229
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984776584979608e-05,
+      "loss": 1.7632,
+      "step": 230
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984407641819812e-05,
+      "loss": 1.7569,
+      "step": 231
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.99840342848015e-05,
+      "loss": 1.7193,
+      "step": 232
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998365651408973e-05,
+      "loss": 1.7666,
+      "step": 233
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.99832743298515e-05,
+      "loss": 1.7152,
+      "step": 234
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9982887732255764e-05,
+      "loss": 1.7483,
+      "step": 235
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998249672147343e-05,
+      "loss": 1.7416,
+      "step": 236
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998210129767735e-05,
+      "loss": 1.7306,
+      "step": 237
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998170146104234e-05,
+      "loss": 1.7681,
+      "step": 238
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981297211745148e-05,
+      "loss": 1.7682,
+      "step": 239
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998088854996449e-05,
+      "loss": 1.6474,
+      "step": 240
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980475475881017e-05,
+      "loss": 1.7064,
+      "step": 241
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980057989677345e-05,
+      "loss": 1.6828,
+      "step": 242
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979636091538033e-05,
+      "loss": 1.749,
+      "step": 243
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997920978164959e-05,
+      "loss": 1.6743,
+      "step": 244
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978779060200483e-05,
+      "loss": 1.6843,
+      "step": 245
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978343927381113e-05,
+      "loss": 1.7561,
+      "step": 246
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997790438338385e-05,
+      "loss": 1.8202,
+      "step": 247
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977460428403e-05,
+      "loss": 1.7899,
+      "step": 248
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977012062634832e-05,
+      "loss": 1.7177,
+      "step": 249
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997655928627755e-05,
+      "loss": 1.7885,
+      "step": 250
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976102099531317e-05,
+      "loss": 1.8025,
+      "step": 251
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9975640502598243e-05,
+      "loss": 1.7572,
+      "step": 252
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9975174495682394e-05,
+      "loss": 1.7426,
+      "step": 253
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974704078989774e-05,
+      "loss": 1.701,
+      "step": 254
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974229252728345e-05,
+      "loss": 1.8352,
+      "step": 255
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997375001710801e-05,
+      "loss": 1.7594,
+      "step": 256
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973266372340638e-05,
+      "loss": 1.7641,
+      "step": 257
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997277831864003e-05,
+      "loss": 1.7044,
+      "step": 258
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972285856221944e-05,
+      "loss": 1.7089,
+      "step": 259
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971788985304084e-05,
+      "loss": 1.7138,
+      "step": 260
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971287706106107e-05,
+      "loss": 1.7401,
+      "step": 261
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997078201884961e-05,
+      "loss": 1.8152,
+      "step": 262
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970271923758152e-05,
+      "loss": 1.7342,
+      "step": 263
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996975742105723e-05,
+      "loss": 1.6812,
+      "step": 264
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969238510974295e-05,
+      "loss": 1.7637,
+      "step": 265
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968715193738738e-05,
+      "loss": 1.7275,
+      "step": 266
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996818746958191e-05,
+      "loss": 1.7515,
+      "step": 267
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9967655338737107e-05,
+      "loss": 1.7346,
+      "step": 268
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9967118801439566e-05,
+      "loss": 1.6968,
+      "step": 269
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966577857926475e-05,
+      "loss": 1.7231,
+      "step": 270
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966032508436977e-05,
+      "loss": 1.7586,
+      "step": 271
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965482753212154e-05,
+      "loss": 1.7458,
+      "step": 272
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964928592495046e-05,
+      "loss": 1.7095,
+      "step": 273
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964370026530623e-05,
+      "loss": 1.7887,
+      "step": 274
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996380705556582e-05,
+      "loss": 1.7542,
+      "step": 275
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963239679849513e-05,
+      "loss": 1.7473,
+      "step": 276
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996266789963252e-05,
+      "loss": 1.7788,
+      "step": 277
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962091715167612e-05,
+      "loss": 1.7763,
+      "step": 278
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996151112670951e-05,
+      "loss": 1.7265,
+      "step": 279
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9960926134514875e-05,
+      "loss": 1.6965,
+      "step": 280
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9960336738842314e-05,
+      "loss": 1.7583,
+      "step": 281
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959742939952393e-05,
+      "loss": 1.6595,
+      "step": 282
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959144738107604e-05,
+      "loss": 1.6863,
+      "step": 283
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.995854213357241e-05,
+      "loss": 1.6622,
+      "step": 284
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957935126613197e-05,
+      "loss": 1.7448,
+      "step": 285
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957323717498315e-05,
+      "loss": 1.7035,
+      "step": 286
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956707906498046e-05,
+      "loss": 1.7201,
+      "step": 287
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956087693884628e-05,
+      "loss": 1.7156,
+      "step": 288
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955463079932244e-05,
+      "loss": 1.7048,
+      "step": 289
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995483406491702e-05,
+      "loss": 1.7447,
+      "step": 290
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995420064911702e-05,
+      "loss": 1.7462,
+      "step": 291
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995356283281227e-05,
+      "loss": 1.7298,
+      "step": 292
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9952920616284735e-05,
+      "loss": 1.712,
+      "step": 293
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9952273999818312e-05,
+      "loss": 1.756,
+      "step": 294
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951622983698865e-05,
+      "loss": 1.7757,
+      "step": 295
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9950967568214183e-05,
+      "loss": 1.7125,
+      "step": 296
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9950307753654016e-05,
+      "loss": 1.6921,
+      "step": 297
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949643540310046e-05,
+      "loss": 1.7548,
+      "step": 298
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994897492847591e-05,
+      "loss": 1.6795,
+      "step": 299
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9948301918447182e-05,
+      "loss": 1.6818,
+      "step": 300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9947624510521385e-05,
+      "loss": 1.6977,
+      "step": 301
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946942704997982e-05,
+      "loss": 1.7426,
+      "step": 302
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946256502178386e-05,
+      "loss": 1.6999,
+      "step": 303
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9945565902365942e-05,
+      "loss": 1.6233,
+      "step": 304
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944870905865956e-05,
+      "loss": 1.7676,
+      "step": 305
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944171512985662e-05,
+      "loss": 1.7711,
+      "step": 306
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943467724034252e-05,
+      "loss": 1.6484,
+      "step": 307
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942759539322845e-05,
+      "loss": 1.6843,
+      "step": 308
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942046959164516e-05,
+      "loss": 1.6819,
+      "step": 309
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994132998387428e-05,
+      "loss": 1.6893,
+      "step": 310
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994060861376909e-05,
+      "loss": 1.7052,
+      "step": 311
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939882849167853e-05,
+      "loss": 1.6654,
+      "step": 312
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939152690391404e-05,
+      "loss": 1.7304,
+      "step": 313
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938418137762534e-05,
+      "loss": 1.815,
+      "step": 314
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937679191605964e-05,
+      "loss": 1.6711,
+      "step": 315
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993693585224837e-05,
+      "loss": 1.6853,
+      "step": 316
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993618812001836e-05,
+      "loss": 1.7744,
+      "step": 317
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993543599524649e-05,
+      "loss": 1.7096,
+      "step": 318
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934679478265255e-05,
+      "loss": 1.712,
+      "step": 319
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933918569409088e-05,
+      "loss": 1.6599,
+      "step": 320
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933153269014373e-05,
+      "loss": 1.7549,
+      "step": 321
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932383577419432e-05,
+      "loss": 1.6571,
+      "step": 322
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931609494964525e-05,
+      "loss": 1.722,
+      "step": 323
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993083102199185e-05,
+      "loss": 1.7117,
+      "step": 324
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993004815884555e-05,
+      "loss": 1.7253,
+      "step": 325
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9929260905871718e-05,
+      "loss": 1.7297,
+      "step": 326
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9928469263418376e-05,
+      "loss": 1.7467,
+      "step": 327
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9927673231835484e-05,
+      "loss": 1.6924,
+      "step": 328
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9926872811474952e-05,
+      "loss": 1.595,
+      "step": 329
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9926068002690625e-05,
+      "loss": 1.664,
+      "step": 330
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992525880583829e-05,
+      "loss": 1.7315,
+      "step": 331
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9924445221275673e-05,
+      "loss": 1.6513,
+      "step": 332
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992362724936244e-05,
+      "loss": 1.6693,
+      "step": 333
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9922804890460194e-05,
+      "loss": 1.7067,
+      "step": 334
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992197814493248e-05,
+      "loss": 1.6577,
+      "step": 335
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9921147013144782e-05,
+      "loss": 1.6925,
+      "step": 336
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992031149546452e-05,
+      "loss": 1.6995,
+      "step": 337
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9919471592261058e-05,
+      "loss": 1.7008,
+      "step": 338
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.99186273039057e-05,
+      "loss": 1.707,
+      "step": 339
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991777863077168e-05,
+      "loss": 1.698,
+      "step": 340
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9916925573234176e-05,
+      "loss": 1.7414,
+      "step": 341
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9916068131670302e-05,
+      "loss": 1.6533,
+      "step": 342
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9915206306459117e-05,
+      "loss": 1.7535,
+      "step": 343
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991434009798161e-05,
+      "loss": 1.7002,
+      "step": 344
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991346950662071e-05,
+      "loss": 1.7082,
+      "step": 345
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991259453276128e-05,
+      "loss": 1.6885,
+      "step": 346
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991171517679013e-05,
+      "loss": 1.7001,
+      "step": 347
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9910831439095995e-05,
+      "loss": 1.7306,
+      "step": 348
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9909943320069564e-05,
+      "loss": 1.7108,
+      "step": 349
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990905082010344e-05,
+      "loss": 1.6716,
+      "step": 350
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9908153939592186e-05,
+      "loss": 1.6985,
+      "step": 351
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990725267893228e-05,
+      "loss": 1.6543,
+      "step": 352
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990634703852216e-05,
+      "loss": 1.7435,
+      "step": 353
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990543701876217e-05,
+      "loss": 1.733,
+      "step": 354
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9904522620054622e-05,
+      "loss": 1.6805,
+      "step": 355
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9903603842803744e-05,
+      "loss": 1.7267,
+      "step": 356
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9902680687415704e-05,
+      "loss": 1.6479,
+      "step": 357
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9901753154298606e-05,
+      "loss": 1.7358,
+      "step": 358
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990082124386249e-05,
+      "loss": 1.655,
+      "step": 359
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989988495651933e-05,
+      "loss": 1.6825,
+      "step": 360
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9898944292683033e-05,
+      "loss": 1.6828,
+      "step": 361
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989799925276945e-05,
+      "loss": 1.7322,
+      "step": 362
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989704983719635e-05,
+      "loss": 1.7264,
+      "step": 363
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9896096046383456e-05,
+      "loss": 1.7615,
+      "step": 364
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9895137880752407e-05,
+      "loss": 1.6998,
+      "step": 365
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989417534072679e-05,
+      "loss": 1.736,
+      "step": 366
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9893208426732115e-05,
+      "loss": 1.6715,
+      "step": 367
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9892237139195835e-05,
+      "loss": 1.7421,
+      "step": 368
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989126147854733e-05,
+      "loss": 1.6844,
+      "step": 369
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9890281445217915e-05,
+      "loss": 1.7221,
+      "step": 370
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988929703964084e-05,
+      "loss": 1.7238,
+      "step": 371
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9888308262251286e-05,
+      "loss": 1.7033,
+      "step": 372
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9887315113486368e-05,
+      "loss": 1.7812,
+      "step": 373
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988631759378513e-05,
+      "loss": 1.7407,
+      "step": 374
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9885315703588546e-05,
+      "loss": 1.7251,
+      "step": 375
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9884309443339535e-05,
+      "loss": 1.7278,
+      "step": 376
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988329881348294e-05,
+      "loss": 1.7249,
+      "step": 377
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988228381446553e-05,
+      "loss": 1.7167,
+      "step": 378
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9881264446736012e-05,
+      "loss": 1.7619,
+      "step": 379
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9880240710745027e-05,
+      "loss": 1.739,
+      "step": 380
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.987921260694514e-05,
+      "loss": 1.7288,
+      "step": 381
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9878180135790844e-05,
+      "loss": 1.7324,
+      "step": 382
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987714329773858e-05,
+      "loss": 1.7476,
+      "step": 383
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9876102093246706e-05,
+      "loss": 1.7313,
+      "step": 384
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9875056522775506e-05,
+      "loss": 1.6835,
+      "step": 385
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9874006586787207e-05,
+      "loss": 1.6515,
+      "step": 386
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9872952285745958e-05,
+      "loss": 1.7175,
+      "step": 387
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9871893620117838e-05,
+      "loss": 1.6585,
+      "step": 388
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987083059037086e-05,
+      "loss": 1.7134,
+      "step": 389
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9869763196974957e-05,
+      "loss": 1.7247,
+      "step": 390
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9868691440402005e-05,
+      "loss": 1.6871,
+      "step": 391
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9867615321125796e-05,
+      "loss": 1.6662,
+      "step": 392
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9866534839622053e-05,
+      "loss": 1.7018,
+      "step": 393
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9865449996368436e-05,
+      "loss": 1.6453,
+      "step": 394
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9864360791844522e-05,
+      "loss": 1.6685,
+      "step": 395
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9863267226531826e-05,
+      "loss": 1.7112,
+      "step": 396
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9862169300913784e-05,
+      "loss": 1.7229,
+      "step": 397
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986106701547576e-05,
+      "loss": 1.6983,
+      "step": 398
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985996037070505e-05,
+      "loss": 1.6205,
+      "step": 399
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9858849367090873e-05,
+      "loss": 1.657,
+      "step": 400
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9857734005124378e-05,
+      "loss": 1.6209,
+      "step": 401
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985661428529863e-05,
+      "loss": 1.7067,
+      "step": 402
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9855490208108642e-05,
+      "loss": 1.6212,
+      "step": 403
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9854361774051332e-05,
+      "loss": 1.6568,
+      "step": 404
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9853228983625556e-05,
+      "loss": 1.6661,
+      "step": 405
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985209183733209e-05,
+      "loss": 1.6444,
+      "step": 406
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985095033567364e-05,
+      "loss": 1.6117,
+      "step": 407
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9849804479154838e-05,
+      "loss": 1.6767,
+      "step": 408
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9848654268282236e-05,
+      "loss": 1.7028,
+      "step": 409
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984749970356431e-05,
+      "loss": 1.7758,
+      "step": 410
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984634078551147e-05,
+      "loss": 1.6757,
+      "step": 411
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984517751463604e-05,
+      "loss": 1.6891,
+      "step": 412
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984400989145228e-05,
+      "loss": 1.7453,
+      "step": 413
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9842837916476364e-05,
+      "loss": 1.7371,
+      "step": 414
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.984166159022639e-05,
+      "loss": 1.6569,
+      "step": 415
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9840480913222385e-05,
+      "loss": 1.7508,
+      "step": 416
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.98392958859863e-05,
+      "loss": 1.675,
+      "step": 417
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9838106509041996e-05,
+      "loss": 1.639,
+      "step": 418
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983691278291528e-05,
+      "loss": 1.6254,
+      "step": 419
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983571470813386e-05,
+      "loss": 1.636,
+      "step": 420
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983451228522738e-05,
+      "loss": 1.7175,
+      "step": 421
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9833305514727396e-05,
+      "loss": 1.6606,
+      "step": 422
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9832094397167396e-05,
+      "loss": 1.6712,
+      "step": 423
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983087893308278e-05,
+      "loss": 1.6928,
+      "step": 424
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9829659123010884e-05,
+      "loss": 1.7096,
+      "step": 425
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9828434967490944e-05,
+      "loss": 1.7369,
+      "step": 426
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9827206467064133e-05,
+      "loss": 1.7058,
+      "step": 427
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9825973622273547e-05,
+      "loss": 1.6241,
+      "step": 428
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9824736433664188e-05,
+      "loss": 1.6721,
+      "step": 429
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9823494901782986e-05,
+      "loss": 1.7028,
+      "step": 430
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9822249027178796e-05,
+      "loss": 1.7262,
+      "step": 431
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.982099881040239e-05,
+      "loss": 1.7435,
+      "step": 432
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9819744252006454e-05,
+      "loss": 1.684,
+      "step": 433
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9818485352545595e-05,
+      "loss": 1.7227,
+      "step": 434
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981722211257634e-05,
+      "loss": 1.6929,
+      "step": 435
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9815954532657145e-05,
+      "loss": 1.6342,
+      "step": 436
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981468261334837e-05,
+      "loss": 1.6458,
+      "step": 437
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9813406355212298e-05,
+      "loss": 1.6779,
+      "step": 438
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9812125758813132e-05,
+      "loss": 1.6106,
+      "step": 439
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9810840824716996e-05,
+      "loss": 1.7804,
+      "step": 440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9809551553491918e-05,
+      "loss": 1.6629,
+      "step": 441
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980825794570786e-05,
+      "loss": 1.7037,
+      "step": 442
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9806960001936693e-05,
+      "loss": 1.7032,
+      "step": 443
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9805657722752202e-05,
+      "loss": 1.7004,
+      "step": 444
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9804351108730095e-05,
+      "loss": 1.7101,
+      "step": 445
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9803040160447996e-05,
+      "loss": 1.6795,
+      "step": 446
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9801724878485438e-05,
+      "loss": 1.6522,
+      "step": 447
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.980040526342388e-05,
+      "loss": 1.6971,
+      "step": 448
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9799081315846686e-05,
+      "loss": 1.7681,
+      "step": 449
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9797753036339143e-05,
+      "loss": 1.5697,
+      "step": 450
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9796420425488448e-05,
+      "loss": 1.6081,
+      "step": 451
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9795083483883715e-05,
+      "loss": 1.6839,
+      "step": 452
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9793742212115978e-05,
+      "loss": 1.7478,
+      "step": 453
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979239661077817e-05,
+      "loss": 1.5961,
+      "step": 454
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979104668046516e-05,
+      "loss": 1.6752,
+      "step": 455
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9789692421773713e-05,
+      "loss": 1.7071,
+      "step": 456
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9788333835302512e-05,
+      "loss": 1.7451,
+      "step": 457
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9786970921652156e-05,
+      "loss": 1.6994,
+      "step": 458
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9785603681425152e-05,
+      "loss": 1.6657,
+      "step": 459
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9784232115225924e-05,
+      "loss": 1.6786,
+      "step": 460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978285622366081e-05,
+      "loss": 1.6658,
+      "step": 461
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9781476007338058e-05,
+      "loss": 1.7294,
+      "step": 462
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9780091466867826e-05,
+      "loss": 1.7454,
+      "step": 463
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977870260286218e-05,
+      "loss": 1.7094,
+      "step": 464
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977730941593511e-05,
+      "loss": 1.7095,
+      "step": 465
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9775911906702507e-05,
+      "loss": 1.7989,
+      "step": 466
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977451007578217e-05,
+      "loss": 1.6463,
+      "step": 467
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977310392379382e-05,
+      "loss": 1.6816,
+      "step": 468
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977169345135908e-05,
+      "loss": 1.717,
+      "step": 469
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9770278659101486e-05,
+      "loss": 1.744,
+      "step": 470
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9768859547646476e-05,
+      "loss": 1.6668,
+      "step": 471
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9767436117621416e-05,
+      "loss": 1.6877,
+      "step": 472
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9766008369655555e-05,
+      "loss": 1.6064,
+      "step": 473
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9764576304380078e-05,
+      "loss": 1.6774,
+      "step": 474
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9763139922428052e-05,
+      "loss": 1.6204,
+      "step": 475
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9761699224434476e-05,
+      "loss": 1.61,
+      "step": 476
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9760254211036245e-05,
+      "loss": 1.6397,
+      "step": 477
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9758804882872164e-05,
+      "loss": 1.6454,
+      "step": 478
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.975735124058294e-05,
+      "loss": 1.6529,
+      "step": 479
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9755893284811196e-05,
+      "loss": 1.6246,
+      "step": 480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9754431016201457e-05,
+      "loss": 1.7127,
+      "step": 481
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9752964435400156e-05,
+      "loss": 1.8029,
+      "step": 482
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9751493543055634e-05,
+      "loss": 1.6674,
+      "step": 483
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9750018339818134e-05,
+      "loss": 1.7457,
+      "step": 484
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9748538826339803e-05,
+      "loss": 1.7215,
+      "step": 485
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9747055003274704e-05,
+      "loss": 1.6431,
+      "step": 486
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9745566871278794e-05,
+      "loss": 1.7223,
+      "step": 487
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9744074431009942e-05,
+      "loss": 1.6063,
+      "step": 488
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974257768312791e-05,
+      "loss": 1.6228,
+      "step": 489
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9741076628294387e-05,
+      "loss": 1.6356,
+      "step": 490
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9739571267172945e-05,
+      "loss": 1.7065,
+      "step": 491
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9738061600429062e-05,
+      "loss": 1.6816,
+      "step": 492
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9736547628730136e-05,
+      "loss": 1.6305,
+      "step": 493
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9735029352745444e-05,
+      "loss": 1.7001,
+      "step": 494
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9733506773146182e-05,
+      "loss": 1.7131,
+      "step": 495
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.973197989060545e-05,
+      "loss": 1.6845,
+      "step": 496
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.973044870579824e-05,
+      "loss": 1.7027,
+      "step": 497
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972891321940145e-05,
+      "loss": 1.6292,
+      "step": 498
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9727373432093885e-05,
+      "loss": 1.6887,
+      "step": 499
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972582934455624e-05,
+      "loss": 1.717,
+      "step": 500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972428095747112e-05,
+      "loss": 1.7007,
+      "step": 501
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9722728271523035e-05,
+      "loss": 1.5967,
+      "step": 502
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9721171287398378e-05,
+      "loss": 1.6972,
+      "step": 503
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9719610005785466e-05,
+      "loss": 1.7268,
+      "step": 504
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971804442737449e-05,
+      "loss": 1.6832,
+      "step": 505
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971647455285756e-05,
+      "loss": 1.6756,
+      "step": 506
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9714900382928674e-05,
+      "loss": 1.6137,
+      "step": 507
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971332191828374e-05,
+      "loss": 1.6221,
+      "step": 508
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9711739159620557e-05,
+      "loss": 1.6297,
+      "step": 509
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9710152107638818e-05,
+      "loss": 1.6386,
+      "step": 510
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.970856076304012e-05,
+      "loss": 1.702,
+      "step": 511
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9706965126527963e-05,
+      "loss": 1.592,
+      "step": 512
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9705365198807734e-05,
+      "loss": 1.6107,
+      "step": 513
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970376098058672e-05,
+      "loss": 1.726,
+      "step": 514
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970215247257411e-05,
+      "loss": 1.6884,
+      "step": 515
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970053967548098e-05,
+      "loss": 1.667,
+      "step": 516
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9698922590020314e-05,
+      "loss": 1.732,
+      "step": 517
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969730121690698e-05,
+      "loss": 1.6389,
+      "step": 518
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9695675556857745e-05,
+      "loss": 1.6122,
+      "step": 519
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969404561059128e-05,
+      "loss": 1.7227,
+      "step": 520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969241137882814e-05,
+      "loss": 1.7423,
+      "step": 521
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969077286229078e-05,
+      "loss": 1.6153,
+      "step": 522
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9689130061703548e-05,
+      "loss": 1.6642,
+      "step": 523
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9687482977792678e-05,
+      "loss": 1.6044,
+      "step": 524
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9685831611286312e-05,
+      "loss": 1.6339,
+      "step": 525
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9684175962914476e-05,
+      "loss": 1.7043,
+      "step": 526
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9682516033409094e-05,
+      "loss": 1.6887,
+      "step": 527
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9680851823503974e-05,
+      "loss": 1.7085,
+      "step": 528
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9679183333934825e-05,
+      "loss": 1.6951,
+      "step": 529
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9677510565439246e-05,
+      "loss": 1.7263,
+      "step": 530
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9675833518756725e-05,
+      "loss": 1.7492,
+      "step": 531
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967415219462864e-05,
+      "loss": 1.667,
+      "step": 532
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9672466593798267e-05,
+      "loss": 1.6974,
+      "step": 533
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9670776717010767e-05,
+      "loss": 1.6958,
+      "step": 534
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966908256501319e-05,
+      "loss": 1.6793,
+      "step": 535
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9667384138554482e-05,
+      "loss": 1.6823,
+      "step": 536
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9665681438385475e-05,
+      "loss": 1.6405,
+      "step": 537
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9663974465258886e-05,
+      "loss": 1.65,
+      "step": 538
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966226321992933e-05,
+      "loss": 1.6869,
+      "step": 539
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9660547703153303e-05,
+      "loss": 1.659,
+      "step": 540
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9658827915689195e-05,
+      "loss": 1.6894,
+      "step": 541
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965710385829728e-05,
+      "loss": 1.6629,
+      "step": 542
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9655375531739718e-05,
+      "loss": 1.6417,
+      "step": 543
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965364293678057e-05,
+      "loss": 1.6532,
+      "step": 544
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9651906074185758e-05,
+      "loss": 1.7124,
+      "step": 545
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9650164944723116e-05,
+      "loss": 1.626,
+      "step": 546
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964841954916235e-05,
+      "loss": 1.6796,
+      "step": 547
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964666988827506e-05,
+      "loss": 1.7203,
+      "step": 548
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9644915962834725e-05,
+      "loss": 1.6504,
+      "step": 549
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9643157773616706e-05,
+      "loss": 1.5755,
+      "step": 550
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9641395321398267e-05,
+      "loss": 1.707,
+      "step": 551
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9639628606958535e-05,
+      "loss": 1.6568,
+      "step": 552
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9637857631078532e-05,
+      "loss": 1.7264,
+      "step": 553
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9636082394541168e-05,
+      "loss": 1.6615,
+      "step": 554
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9634302898131224e-05,
+      "loss": 1.6868,
+      "step": 555
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9632519142635372e-05,
+      "loss": 1.6973,
+      "step": 556
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963073112884217e-05,
+      "loss": 1.7097,
+      "step": 557
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9628938857542056e-05,
+      "loss": 1.6029,
+      "step": 558
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962714232952734e-05,
+      "loss": 1.71,
+      "step": 559
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9625341545592226e-05,
+      "loss": 1.6253,
+      "step": 560
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9623536506532802e-05,
+      "loss": 1.7284,
+      "step": 561
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9621727213147025e-05,
+      "loss": 1.7402,
+      "step": 562
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9619913666234738e-05,
+      "loss": 1.6596,
+      "step": 563
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9618095866597668e-05,
+      "loss": 1.6731,
+      "step": 564
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961627381503942e-05,
+      "loss": 1.7586,
+      "step": 565
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961444751236547e-05,
+      "loss": 1.6985,
+      "step": 566
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961261695938319e-05,
+      "loss": 1.7023,
+      "step": 567
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961078215690182e-05,
+      "loss": 1.6992,
+      "step": 568
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9608943105732476e-05,
+      "loss": 1.6906,
+      "step": 569
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9607099806688163e-05,
+      "loss": 1.5959,
+      "step": 570
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9605252260583752e-05,
+      "loss": 1.6427,
+      "step": 571
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9603400468236e-05,
+      "loss": 1.6997,
+      "step": 572
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960154443046354e-05,
+      "loss": 1.6281,
+      "step": 573
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9599684148086876e-05,
+      "loss": 1.7068,
+      "step": 574
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9597819621928398e-05,
+      "loss": 1.6465,
+      "step": 575
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.959595085281236e-05,
+      "loss": 1.6678,
+      "step": 576
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9594077841564905e-05,
+      "loss": 1.7708,
+      "step": 577
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9592200589014042e-05,
+      "loss": 1.687,
+      "step": 578
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.959031909598966e-05,
+      "loss": 1.6155,
+      "step": 579
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9588433363323514e-05,
+      "loss": 1.7511,
+      "step": 580
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9586543391849243e-05,
+      "loss": 1.6242,
+      "step": 581
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9584649182402358e-05,
+      "loss": 1.664,
+      "step": 582
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958275073582024e-05,
+      "loss": 1.6627,
+      "step": 583
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958084805294215e-05,
+      "loss": 1.6646,
+      "step": 584
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9578941134609207e-05,
+      "loss": 1.6656,
+      "step": 585
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957702998166442e-05,
+      "loss": 1.6758,
+      "step": 586
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9575114594952662e-05,
+      "loss": 1.6912,
+      "step": 587
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9573194975320672e-05,
+      "loss": 1.6513,
+      "step": 588
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9571271123617073e-05,
+      "loss": 1.6691,
+      "step": 589
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956934304069235e-05,
+      "loss": 1.694,
+      "step": 590
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956741072739886e-05,
+      "loss": 1.7242,
+      "step": 591
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9565474184590827e-05,
+      "loss": 1.6739,
+      "step": 592
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9563533413124355e-05,
+      "loss": 1.6416,
+      "step": 593
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9561588413857408e-05,
+      "loss": 1.6358,
+      "step": 594
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9559639187649817e-05,
+      "loss": 1.6876,
+      "step": 595
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9557685735363295e-05,
+      "loss": 1.7439,
+      "step": 596
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955572805786141e-05,
+      "loss": 1.7172,
+      "step": 597
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.95537661560096e-05,
+      "loss": 1.622,
+      "step": 598
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955180003067518e-05,
+      "loss": 1.6382,
+      "step": 599
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9549829682727322e-05,
+      "loss": 1.6467,
+      "step": 600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9547855113037065e-05,
+      "loss": 1.6722,
+      "step": 601
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954587632247732e-05,
+      "loss": 1.7927,
+      "step": 602
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9543893311922863e-05,
+      "loss": 1.6827,
+      "step": 603
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9541906082250328e-05,
+      "loss": 1.665,
+      "step": 604
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9539914634338225e-05,
+      "loss": 1.6693,
+      "step": 605
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9537918969066923e-05,
+      "loss": 1.6035,
+      "step": 606
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9535919087318654e-05,
+      "loss": 1.6251,
+      "step": 607
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9533914989977517e-05,
+      "loss": 1.6107,
+      "step": 608
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9531906677929472e-05,
+      "loss": 1.7052,
+      "step": 609
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9529894152062344e-05,
+      "loss": 1.6616,
+      "step": 610
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9527877413265827e-05,
+      "loss": 1.5613,
+      "step": 611
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9525856462431463e-05,
+      "loss": 1.6931,
+      "step": 612
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952383130045267e-05,
+      "loss": 1.623,
+      "step": 613
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9521801928224717e-05,
+      "loss": 1.7185,
+      "step": 614
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951976834664474e-05,
+      "loss": 1.6448,
+      "step": 615
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951773055661174e-05,
+      "loss": 1.6794,
+      "step": 616
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9515688559026564e-05,
+      "loss": 1.6336,
+      "step": 617
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9513642354791938e-05,
+      "loss": 1.7262,
+      "step": 618
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9511591944812432e-05,
+      "loss": 1.6919,
+      "step": 619
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950953732999448e-05,
+      "loss": 1.5808,
+      "step": 620
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9507478511246385e-05,
+      "loss": 1.6966,
+      "step": 621
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9505415489478293e-05,
+      "loss": 1.7469,
+      "step": 622
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9503348265602212e-05,
+      "loss": 1.6625,
+      "step": 623
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9501276840532017e-05,
+      "loss": 1.6897,
+      "step": 624
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.949920121518343e-05,
+      "loss": 1.674,
+      "step": 625
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949712139047403e-05,
+      "loss": 1.6575,
+      "step": 626
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9495037367323264e-05,
+      "loss": 1.6673,
+      "step": 627
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9492949146652423e-05,
+      "loss": 1.6563,
+      "step": 628
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949085672938465e-05,
+      "loss": 1.6318,
+      "step": 629
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9488760116444966e-05,
+      "loss": 1.6875,
+      "step": 630
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9486659308760222e-05,
+      "loss": 1.6869,
+      "step": 631
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948455430725913e-05,
+      "loss": 1.611,
+      "step": 632
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9482445112872265e-05,
+      "loss": 1.6855,
+      "step": 633
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9480331726532047e-05,
+      "loss": 1.6537,
+      "step": 634
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9478214149172752e-05,
+      "loss": 1.6747,
+      "step": 635
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9476092381730508e-05,
+      "loss": 1.6963,
+      "step": 636
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9473966425143292e-05,
+      "loss": 1.5712,
+      "step": 637
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9471836280350945e-05,
+      "loss": 1.6548,
+      "step": 638
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9469701948295147e-05,
+      "loss": 1.7473,
+      "step": 639
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946756342991943e-05,
+      "loss": 1.6899,
+      "step": 640
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9465420726169186e-05,
+      "loss": 1.6341,
+      "step": 641
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9463273837991643e-05,
+      "loss": 1.6637,
+      "step": 642
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9461122766335894e-05,
+      "loss": 1.6793,
+      "step": 643
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9458967512152872e-05,
+      "loss": 1.6307,
+      "step": 644
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945680807639536e-05,
+      "loss": 1.7131,
+      "step": 645
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9454644460017992e-05,
+      "loss": 1.6292,
+      "step": 646
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945247666397725e-05,
+      "loss": 1.7633,
+      "step": 647
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945030468923146e-05,
+      "loss": 1.6494,
+      "step": 648
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9448128536740802e-05,
+      "loss": 1.6288,
+      "step": 649
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9445948207467294e-05,
+      "loss": 1.6081,
+      "step": 650
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944376370237481e-05,
+      "loss": 1.5821,
+      "step": 651
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9441575022429065e-05,
+      "loss": 1.7488,
+      "step": 652
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943938216859762e-05,
+      "loss": 1.6324,
+      "step": 653
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9437185141849877e-05,
+      "loss": 1.6745,
+      "step": 654
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943498394315709e-05,
+      "loss": 1.6417,
+      "step": 655
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9432778573492352e-05,
+      "loss": 1.6764,
+      "step": 656
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9430569033830606e-05,
+      "loss": 1.6562,
+      "step": 657
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9428355325148632e-05,
+      "loss": 1.6591,
+      "step": 658
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9426137448425058e-05,
+      "loss": 1.6506,
+      "step": 659
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942391540464035e-05,
+      "loss": 1.6744,
+      "step": 660
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942168919477682e-05,
+      "loss": 1.6798,
+      "step": 661
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9419458819818617e-05,
+      "loss": 1.6258,
+      "step": 662
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9417224280751733e-05,
+      "loss": 1.7131,
+      "step": 663
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941498557856401e-05,
+      "loss": 1.6143,
+      "step": 664
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941274271424512e-05,
+      "loss": 1.6786,
+      "step": 665
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941049568878657e-05,
+      "loss": 1.6702,
+      "step": 666
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9408244503181723e-05,
+      "loss": 1.6566,
+      "step": 667
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9405989158425766e-05,
+      "loss": 1.6639,
+      "step": 668
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9403729655515736e-05,
+      "loss": 1.6784,
+      "step": 669
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9401465995450502e-05,
+      "loss": 1.6853,
+      "step": 670
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939919817923077e-05,
+      "loss": 1.6081,
+      "step": 671
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9396926207859085e-05,
+      "loss": 1.7015,
+      "step": 672
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939465008233983e-05,
+      "loss": 1.6389,
+      "step": 673
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9392369803679225e-05,
+      "loss": 1.6895,
+      "step": 674
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9390085372885325e-05,
+      "loss": 1.7022,
+      "step": 675
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9387796790968013e-05,
+      "loss": 1.6373,
+      "step": 676
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9385504058939023e-05,
+      "loss": 1.6226,
+      "step": 677
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938320717781191e-05,
+      "loss": 1.6302,
+      "step": 678
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938090614860207e-05,
+      "loss": 1.6142,
+      "step": 679
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937860097232673e-05,
+      "loss": 1.5817,
+      "step": 680
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937629165000495e-05,
+      "loss": 1.6921,
+      "step": 681
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9373978182657628e-05,
+      "loss": 1.6263,
+      "step": 682
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9371660571307487e-05,
+      "loss": 1.6182,
+      "step": 683
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9369338816979085e-05,
+      "loss": 1.7171,
+      "step": 684
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9367012920698812e-05,
+      "loss": 1.6792,
+      "step": 685
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9364682883494892e-05,
+      "loss": 1.5908,
+      "step": 686
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9362348706397374e-05,
+      "loss": 1.6775,
+      "step": 687
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936001039043814e-05,
+      "loss": 1.5612,
+      "step": 688
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9357667936650902e-05,
+      "loss": 1.6577,
+      "step": 689
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.93553213460712e-05,
+      "loss": 1.6828,
+      "step": 690
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9352970619736403e-05,
+      "loss": 1.6218,
+      "step": 691
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.935061575868571e-05,
+      "loss": 1.7676,
+      "step": 692
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9348256763960146e-05,
+      "loss": 1.6459,
+      "step": 693
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9345893636602565e-05,
+      "loss": 1.7063,
+      "step": 694
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9343526377657646e-05,
+      "loss": 1.6706,
+      "step": 695
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934115498817189e-05,
+      "loss": 1.6789,
+      "step": 696
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9338779469193638e-05,
+      "loss": 1.6809,
+      "step": 697
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9336399821773043e-05,
+      "loss": 1.7282,
+      "step": 698
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933401604696209e-05,
+      "loss": 1.6884,
+      "step": 699
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9331628145814587e-05,
+      "loss": 1.6819,
+      "step": 700
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9329236119386166e-05,
+      "loss": 1.662,
+      "step": 701
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9326839968734278e-05,
+      "loss": 1.5986,
+      "step": 702
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9324439694918204e-05,
+      "loss": 1.6102,
+      "step": 703
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932203529899905e-05,
+      "loss": 1.6661,
+      "step": 704
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9319626782039735e-05,
+      "loss": 1.6419,
+      "step": 705
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9317214145105004e-05,
+      "loss": 1.6823,
+      "step": 706
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9314797389261426e-05,
+      "loss": 1.6365,
+      "step": 707
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9312376515577387e-05,
+      "loss": 1.6019,
+      "step": 708
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9309951525123096e-05,
+      "loss": 1.6158,
+      "step": 709
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930752241897058e-05,
+      "loss": 1.671,
+      "step": 710
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930508919819369e-05,
+      "loss": 1.7368,
+      "step": 711
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9302651863868093e-05,
+      "loss": 1.656,
+      "step": 712
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9300210417071264e-05,
+      "loss": 1.7277,
+      "step": 713
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9297764858882516e-05,
+      "loss": 1.727,
+      "step": 714
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9295315190382966e-05,
+      "loss": 1.634,
+      "step": 715
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9292861412655552e-05,
+      "loss": 1.6291,
+      "step": 716
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9290403526785025e-05,
+      "loss": 1.7319,
+      "step": 717
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9287941533857966e-05,
+      "loss": 1.6736,
+      "step": 718
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9285475434962747e-05,
+      "loss": 1.6222,
+      "step": 719
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9283005231189578e-05,
+      "loss": 1.6265,
+      "step": 720
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928053092363047e-05,
+      "loss": 1.6632,
+      "step": 721
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9278052513379256e-05,
+      "loss": 1.6147,
+      "step": 722
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9275570001531578e-05,
+      "loss": 1.593,
+      "step": 723
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9273083389184894e-05,
+      "loss": 1.5814,
+      "step": 724
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9270592677438472e-05,
+      "loss": 1.651,
+      "step": 725
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9268097867393395e-05,
+      "loss": 1.6834,
+      "step": 726
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9265598960152556e-05,
+      "loss": 1.6935,
+      "step": 727
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926309595682066e-05,
+      "loss": 1.6958,
+      "step": 728
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9260588858504223e-05,
+      "loss": 1.6499,
+      "step": 729
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925807766631157e-05,
+      "loss": 1.6378,
+      "step": 730
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9255562381352836e-05,
+      "loss": 1.7173,
+      "step": 731
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9253043004739967e-05,
+      "loss": 1.6157,
+      "step": 732
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925051953758672e-05,
+      "loss": 1.643,
+      "step": 733
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924799198100865e-05,
+      "loss": 1.6931,
+      "step": 734
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9245460336123136e-05,
+      "loss": 1.6202,
+      "step": 735
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9242924604049343e-05,
+      "loss": 1.6513,
+      "step": 736
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9240384785908267e-05,
+      "loss": 1.6004,
+      "step": 737
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.923784088282269e-05,
+      "loss": 1.6514,
+      "step": 738
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9235292895917218e-05,
+      "loss": 1.5918,
+      "step": 739
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.923274082631824e-05,
+      "loss": 1.6931,
+      "step": 740
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9230184675153974e-05,
+      "loss": 1.6935,
+      "step": 741
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9227624443554425e-05,
+      "loss": 1.6266,
+      "step": 742
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922506013265141e-05,
+      "loss": 1.5804,
+      "step": 743
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9222491743578545e-05,
+      "loss": 1.6296,
+      "step": 744
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9219919277471253e-05,
+      "loss": 1.6514,
+      "step": 745
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9217342735466756e-05,
+      "loss": 1.6732,
+      "step": 746
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921476211870408e-05,
+      "loss": 1.6384,
+      "step": 747
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921217742832405e-05,
+      "loss": 1.7107,
+      "step": 748
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9209588665469294e-05,
+      "loss": 1.6444,
+      "step": 749
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9206995831284244e-05,
+      "loss": 1.6838,
+      "step": 750
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920439892691512e-05,
+      "loss": 1.5651,
+      "step": 751
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9201797953509954e-05,
+      "loss": 1.7461,
+      "step": 752
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9199192912218572e-05,
+      "loss": 1.5917,
+      "step": 753
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9196583804192594e-05,
+      "loss": 1.6692,
+      "step": 754
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.919397063058545e-05,
+      "loss": 1.5721,
+      "step": 755
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9191353392552346e-05,
+      "loss": 1.6854,
+      "step": 756
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918873209125031e-05,
+      "loss": 1.597,
+      "step": 757
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9186106727838148e-05,
+      "loss": 1.5714,
+      "step": 758
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918347730347647e-05,
+      "loss": 1.7044,
+      "step": 759
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9180843819327677e-05,
+      "loss": 1.6283,
+      "step": 760
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9178206276555968e-05,
+      "loss": 1.6678,
+      "step": 761
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917556467632734e-05,
+      "loss": 1.6415,
+      "step": 762
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9172919019809572e-05,
+      "loss": 1.6904,
+      "step": 763
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9170269308172247e-05,
+      "loss": 1.5574,
+      "step": 764
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916761554258673e-05,
+      "loss": 1.7021,
+      "step": 765
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9164957724226197e-05,
+      "loss": 1.6012,
+      "step": 766
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9162295854265593e-05,
+      "loss": 1.5679,
+      "step": 767
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9159629933881666e-05,
+      "loss": 1.6299,
+      "step": 768
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9156959964252957e-05,
+      "loss": 1.6675,
+      "step": 769
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9154285946559792e-05,
+      "loss": 1.5442,
+      "step": 770
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9151607881984285e-05,
+      "loss": 1.6629,
+      "step": 771
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9148925771710347e-05,
+      "loss": 1.696,
+      "step": 772
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914623961692367e-05,
+      "loss": 1.7147,
+      "step": 773
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9143549418811732e-05,
+      "loss": 1.7003,
+      "step": 774
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9140855178563807e-05,
+      "loss": 1.6636,
+      "step": 775
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9138156897370955e-05,
+      "loss": 1.6636,
+      "step": 776
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913545457642601e-05,
+      "loss": 1.6502,
+      "step": 777
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913274821692361e-05,
+      "loss": 1.6105,
+      "step": 778
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913003782006016e-05,
+      "loss": 1.5923,
+      "step": 779
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912732338703387e-05,
+      "loss": 1.6099,
+      "step": 780
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9124604919044714e-05,
+      "loss": 1.6741,
+      "step": 781
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9121882417294462e-05,
+      "loss": 1.6754,
+      "step": 782
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911915588298667e-05,
+      "loss": 1.6611,
+      "step": 783
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911642531732666e-05,
+      "loss": 1.7149,
+      "step": 784
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911369072152156e-05,
+      "loss": 1.6554,
+      "step": 785
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911095209678026e-05,
+      "loss": 1.7663,
+      "step": 786
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9108209444313432e-05,
+      "loss": 1.7001,
+      "step": 787
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9105462765333544e-05,
+      "loss": 1.704,
+      "step": 788
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9102712061054835e-05,
+      "loss": 1.729,
+      "step": 789
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9099957332693314e-05,
+      "loss": 1.6469,
+      "step": 790
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9097198581466785e-05,
+      "loss": 1.6391,
+      "step": 791
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9094435808594823e-05,
+      "loss": 1.7281,
+      "step": 792
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9091669015298778e-05,
+      "loss": 1.6208,
+      "step": 793
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9088898202801786e-05,
+      "loss": 1.6393,
+      "step": 794
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9086123372328748e-05,
+      "loss": 1.6439,
+      "step": 795
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9083344525106347e-05,
+      "loss": 1.6517,
+      "step": 796
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908056166236305e-05,
+      "loss": 1.6376,
+      "step": 797
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907777478532909e-05,
+      "loss": 1.6341,
+      "step": 798
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9074983895236467e-05,
+      "loss": 1.6842,
+      "step": 799
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9072188993318973e-05,
+      "loss": 1.674,
+      "step": 800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9069390080812164e-05,
+      "loss": 1.6657,
+      "step": 801
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906658715895337e-05,
+      "loss": 1.6595,
+      "step": 802
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9063780228981687e-05,
+      "loss": 1.5641,
+      "step": 803
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9060969292137992e-05,
+      "loss": 1.6672,
+      "step": 804
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9058154349664932e-05,
+      "loss": 1.6645,
+      "step": 805
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905533540280692e-05,
+      "loss": 1.7174,
+      "step": 806
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905251245281015e-05,
+      "loss": 1.6904,
+      "step": 807
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904968550092257e-05,
+      "loss": 1.6409,
+      "step": 808
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904685454839391e-05,
+      "loss": 1.6472,
+      "step": 809
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904401959647566e-05,
+      "loss": 1.6432,
+      "step": 810
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9041180646421083e-05,
+      "loss": 1.6705,
+      "step": 811
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9038337699485207e-05,
+      "loss": 1.6438,
+      "step": 812
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9035490756924834e-05,
+      "loss": 1.6419,
+      "step": 813
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9032639819998517e-05,
+      "loss": 1.5527,
+      "step": 814
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9029784889966593e-05,
+      "loss": 1.6076,
+      "step": 815
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9026925968091154e-05,
+      "loss": 1.6468,
+      "step": 816
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9024063055636057e-05,
+      "loss": 1.6043,
+      "step": 817
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9021196153866922e-05,
+      "loss": 1.675,
+      "step": 818
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.901832526405114e-05,
+      "loss": 1.5967,
+      "step": 819
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9015450387457855e-05,
+      "loss": 1.6512,
+      "step": 820
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.901257152535798e-05,
+      "loss": 1.6257,
+      "step": 821
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900968867902419e-05,
+      "loss": 1.6259,
+      "step": 822
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9006801849730922e-05,
+      "loss": 1.5846,
+      "step": 823
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900391103875437e-05,
+      "loss": 1.6744,
+      "step": 824
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9001016247372483e-05,
+      "loss": 1.6256,
+      "step": 825
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8998117476864984e-05,
+      "loss": 1.6892,
+      "step": 826
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899521472851334e-05,
+      "loss": 1.6258,
+      "step": 827
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8992308003600794e-05,
+      "loss": 1.6354,
+      "step": 828
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898939730341233e-05,
+      "loss": 1.64,
+      "step": 829
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.89864826292347e-05,
+      "loss": 1.6476,
+      "step": 830
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8983563982356406e-05,
+      "loss": 1.646,
+      "step": 831
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8980641364067708e-05,
+      "loss": 1.6481,
+      "step": 832
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897771477566063e-05,
+      "loss": 1.6165,
+      "step": 833
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8974784218428936e-05,
+      "loss": 1.6576,
+      "step": 834
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8971849693668153e-05,
+      "loss": 1.6996,
+      "step": 835
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8968911202675568e-05,
+      "loss": 1.558,
+      "step": 836
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.896596874675021e-05,
+      "loss": 1.7839,
+      "step": 837
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8963022327192867e-05,
+      "loss": 1.6881,
+      "step": 838
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8960071945306078e-05,
+      "loss": 1.5758,
+      "step": 839
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895711760239413e-05,
+      "loss": 1.5513,
+      "step": 840
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895415929976307e-05,
+      "loss": 1.6511,
+      "step": 841
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895119703872069e-05,
+      "loss": 1.6684,
+      "step": 842
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8948230820576525e-05,
+      "loss": 1.6405,
+      "step": 843
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8945260646641873e-05,
+      "loss": 1.5933,
+      "step": 844
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894228651822977e-05,
+      "loss": 1.6319,
+      "step": 845
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8939308436655007e-05,
+      "loss": 1.6127,
+      "step": 846
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8936326403234125e-05,
+      "loss": 1.6796,
+      "step": 847
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8933340419285398e-05,
+      "loss": 1.6282,
+      "step": 848
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8930350486128855e-05,
+      "loss": 1.6527,
+      "step": 849
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8927356605086277e-05,
+      "loss": 1.6783,
+      "step": 850
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892435877748118e-05,
+      "loss": 1.6619,
+      "step": 851
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8921357004638837e-05,
+      "loss": 1.6674,
+      "step": 852
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.891835128788625e-05,
+      "loss": 1.6363,
+      "step": 853
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8915341628552166e-05,
+      "loss": 1.7026,
+      "step": 854
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8912328027967092e-05,
+      "loss": 1.7036,
+      "step": 855
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8909310487463263e-05,
+      "loss": 1.64,
+      "step": 856
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8906289008374654e-05,
+      "loss": 1.5867,
+      "step": 857
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890326359203699e-05,
+      "loss": 1.6074,
+      "step": 858
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8900234239787735e-05,
+      "loss": 1.6224,
+      "step": 859
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8897200952966082e-05,
+      "loss": 1.64,
+      "step": 860
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.889416373291298e-05,
+      "loss": 1.6033,
+      "step": 861
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.88911225809711e-05,
+      "loss": 1.6375,
+      "step": 862
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.888807749848487e-05,
+      "loss": 1.6341,
+      "step": 863
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8885028486800436e-05,
+      "loss": 1.6663,
+      "step": 864
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8881975547265695e-05,
+      "loss": 1.6728,
+      "step": 865
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8878918681230275e-05,
+      "loss": 1.6428,
+      "step": 866
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8875857890045544e-05,
+      "loss": 1.6178,
+      "step": 867
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8872793175064594e-05,
+      "loss": 1.6448,
+      "step": 868
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8869724537642263e-05,
+      "loss": 1.67,
+      "step": 869
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8866651979135124e-05,
+      "loss": 1.6652,
+      "step": 870
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8863575500901473e-05,
+      "loss": 1.6014,
+      "step": 871
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8860495104301346e-05,
+      "loss": 1.5407,
+      "step": 872
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885741079069651e-05,
+      "loss": 1.5991,
+      "step": 873
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8854322561450463e-05,
+      "loss": 1.5724,
+      "step": 874
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8851230417928433e-05,
+      "loss": 1.6598,
+      "step": 875
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8848134361497385e-05,
+      "loss": 1.6644,
+      "step": 876
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8845034393526005e-05,
+      "loss": 1.6983,
+      "step": 877
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8841930515384714e-05,
+      "loss": 1.6597,
+      "step": 878
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883882272844565e-05,
+      "loss": 1.5629,
+      "step": 879
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8835711034082705e-05,
+      "loss": 1.6353,
+      "step": 880
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8832595433671466e-05,
+      "loss": 1.6349,
+      "step": 881
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8829475928589272e-05,
+      "loss": 1.6229,
+      "step": 882
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8826352520215173e-05,
+      "loss": 1.6873,
+      "step": 883
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8823225209929955e-05,
+      "loss": 1.637,
+      "step": 884
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8820093999116124e-05,
+      "loss": 1.6049,
+      "step": 885
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8816958889157906e-05,
+      "loss": 1.5314,
+      "step": 886
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.881381988144126e-05,
+      "loss": 1.6414,
+      "step": 887
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.881067697735386e-05,
+      "loss": 1.6594,
+      "step": 888
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880753017828511e-05,
+      "loss": 1.6356,
+      "step": 889
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880437948562612e-05,
+      "loss": 1.6248,
+      "step": 890
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880122490076975e-05,
+      "loss": 1.7003,
+      "step": 891
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879806642511055e-05,
+      "loss": 1.6091,
+      "step": 892
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8794904060044813e-05,
+      "loss": 1.608,
+      "step": 893
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8791737806970538e-05,
+      "loss": 1.7353,
+      "step": 894
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8788567667287445e-05,
+      "loss": 1.6084,
+      "step": 895
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8785393642396976e-05,
+      "loss": 1.6647,
+      "step": 896
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8782215733702286e-05,
+      "loss": 1.5599,
+      "step": 897
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8779033942608253e-05,
+      "loss": 1.5954,
+      "step": 898
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8775848270521465e-05,
+      "loss": 1.6684,
+      "step": 899
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8772658718850234e-05,
+      "loss": 1.5627,
+      "step": 900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8769465289004577e-05,
+      "loss": 1.6843,
+      "step": 901
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8766267982396225e-05,
+      "loss": 1.6152,
+      "step": 902
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8763066800438638e-05,
+      "loss": 1.6161,
+      "step": 903
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875986174454697e-05,
+      "loss": 1.6097,
+      "step": 904
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8756652816138106e-05,
+      "loss": 1.7011,
+      "step": 905
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8753440016630623e-05,
+      "loss": 1.5868,
+      "step": 906
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875022334744483e-05,
+      "loss": 1.5795,
+      "step": 907
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8747002810002728e-05,
+      "loss": 1.6669,
+      "step": 908
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874377840572804e-05,
+      "loss": 1.6815,
+      "step": 909
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8740550136046195e-05,
+      "loss": 1.6063,
+      "step": 910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8737318002384334e-05,
+      "loss": 1.5567,
+      "step": 911
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.87340820061713e-05,
+      "loss": 1.6315,
+      "step": 912
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8730842148837643e-05,
+      "loss": 1.6491,
+      "step": 913
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872759843181563e-05,
+      "loss": 1.6464,
+      "step": 914
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872435085653922e-05,
+      "loss": 1.589,
+      "step": 915
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8721099424444097e-05,
+      "loss": 1.662,
+      "step": 916
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8717844136967626e-05,
+      "loss": 1.6536,
+      "step": 917
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8714584995548894e-05,
+      "loss": 1.7229,
+      "step": 918
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.871132200162869e-05,
+      "loss": 1.6121,
+      "step": 919
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8708055156649496e-05,
+      "loss": 1.5899,
+      "step": 920
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8704784462055504e-05,
+      "loss": 1.6521,
+      "step": 921
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870150991929261e-05,
+      "loss": 1.6553,
+      "step": 922
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8698231529808407e-05,
+      "loss": 1.7084,
+      "step": 923
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.869494929505219e-05,
+      "loss": 1.6894,
+      "step": 924
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8691663216474955e-05,
+      "loss": 1.6449,
+      "step": 925
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868837329552939e-05,
+      "loss": 1.6223,
+      "step": 926
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868507953366989e-05,
+      "loss": 1.59,
+      "step": 927
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868178193235255e-05,
+      "loss": 1.6151,
+      "step": 928
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8678480493035147e-05,
+      "loss": 1.6656,
+      "step": 929
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8675175217177176e-05,
+      "loss": 1.6386,
+      "step": 930
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8671866106239812e-05,
+      "loss": 1.691,
+      "step": 931
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8668553161685932e-05,
+      "loss": 1.7312,
+      "step": 932
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8665236384980107e-05,
+      "loss": 1.6348,
+      "step": 933
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.86619157775886e-05,
+      "loss": 1.6112,
+      "step": 934
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8658591340979374e-05,
+      "loss": 1.6816,
+      "step": 935
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.865526307662207e-05,
+      "loss": 1.5995,
+      "step": 936
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8651930985988037e-05,
+      "loss": 1.7044,
+      "step": 937
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8648595070550312e-05,
+      "loss": 1.6169,
+      "step": 938
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8645255331783615e-05,
+      "loss": 1.626,
+      "step": 939
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8641911771164364e-05,
+      "loss": 1.6243,
+      "step": 940
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8638564390170666e-05,
+      "loss": 1.6481,
+      "step": 941
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8635213190282312e-05,
+      "loss": 1.7065,
+      "step": 942
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8631858172980783e-05,
+      "loss": 1.6676,
+      "step": 943
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8628499339749252e-05,
+      "loss": 1.6216,
+      "step": 944
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8625136692072577e-05,
+      "loss": 1.6376,
+      "step": 945
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.86217702314373e-05,
+      "loss": 1.6189,
+      "step": 946
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8618399959331642e-05,
+      "loss": 1.6392,
+      "step": 947
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8615025877245524e-05,
+      "loss": 1.6171,
+      "step": 948
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8611647986670545e-05,
+      "loss": 1.6571,
+      "step": 949
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8608266289099985e-05,
+      "loss": 1.6309,
+      "step": 950
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8604880786028805e-05,
+      "loss": 1.673,
+      "step": 951
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.860149147895366e-05,
+      "loss": 1.6354,
+      "step": 952
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8598098369372862e-05,
+      "loss": 1.6269,
+      "step": 953
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.859470145878644e-05,
+      "loss": 1.6119,
+      "step": 954
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.859130074869607e-05,
+      "loss": 1.7157,
+      "step": 955
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.858789624060513e-05,
+      "loss": 1.5849,
+      "step": 956
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8584487936018663e-05,
+      "loss": 1.6456,
+      "step": 957
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8581075836443398e-05,
+      "loss": 1.7006,
+      "step": 958
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8577659943387737e-05,
+      "loss": 1.6456,
+      "step": 959
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8574240258361768e-05,
+      "loss": 1.5848,
+      "step": 960
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8570816782877238e-05,
+      "loss": 1.6324,
+      "step": 961
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.856738951844759e-05,
+      "loss": 1.6162,
+      "step": 962
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8563958466587932e-05,
+      "loss": 1.6108,
+      "step": 963
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.856052362881504e-05,
+      "loss": 1.6359,
+      "step": 964
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8557085006647377e-05,
+      "loss": 1.5745,
+      "step": 965
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.855364260160507e-05,
+      "loss": 1.6317,
+      "step": 966
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8550196415209916e-05,
+      "loss": 1.6215,
+      "step": 967
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8546746448985395e-05,
+      "loss": 1.5647,
+      "step": 968
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854329270445665e-05,
+      "loss": 1.657,
+      "step": 969
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8539835183150493e-05,
+      "loss": 1.6098,
+      "step": 970
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.853637388659541e-05,
+      "loss": 1.72,
+      "step": 971
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8532908816321557e-05,
+      "loss": 1.6384,
+      "step": 972
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.852943997386075e-05,
+      "loss": 1.6393,
+      "step": 973
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8525967360746485e-05,
+      "loss": 1.6444,
+      "step": 974
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.852249097851391e-05,
+      "loss": 1.6078,
+      "step": 975
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.851901082869985e-05,
+      "loss": 1.6184,
+      "step": 976
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8515526912842796e-05,
+      "loss": 1.6135,
+      "step": 977
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.85120392324829e-05,
+      "loss": 1.6915,
+      "step": 978
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8508547789161972e-05,
+      "loss": 1.6327,
+      "step": 979
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.85050525844235e-05,
+      "loss": 1.6137,
+      "step": 980
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8501553619812626e-05,
+      "loss": 1.5413,
+      "step": 981
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8498050896876152e-05,
+      "loss": 1.5734,
+      "step": 982
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8494544417162545e-05,
+      "loss": 1.5874,
+      "step": 983
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.849103418222194e-05,
+      "loss": 1.5841,
+      "step": 984
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.848752019360611e-05,
+      "loss": 1.6661,
+      "step": 985
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.848400245286852e-05,
+      "loss": 1.7319,
+      "step": 986
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.848048096156426e-05,
+      "loss": 1.6497,
+      "step": 987
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8476955721250106e-05,
+      "loss": 1.6444,
+      "step": 988
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847342673348447e-05,
+      "loss": 1.6446,
+      "step": 989
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8469893999827433e-05,
+      "loss": 1.6818,
+      "step": 990
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.846635752184073e-05,
+      "loss": 1.6375,
+      "step": 991
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.846281730108775e-05,
+      "loss": 1.6087,
+      "step": 992
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8459273339133538e-05,
+      "loss": 1.6499,
+      "step": 993
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8455725637544784e-05,
+      "loss": 1.6907,
+      "step": 994
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8452174197889846e-05,
+      "loss": 1.5533,
+      "step": 995
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8448619021738728e-05,
+      "loss": 1.6674,
+      "step": 996
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.844506011066308e-05,
+      "loss": 1.5221,
+      "step": 997
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8441497466236212e-05,
+      "loss": 1.6447,
+      "step": 998
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8437931090033077e-05,
+      "loss": 1.6311,
+      "step": 999
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8434360983630283e-05,
+      "loss": 1.65,
+      "step": 1000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8430787148606087e-05,
+      "loss": 1.603,
+      "step": 1001
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8427209586540392e-05,
+      "loss": 1.5696,
+      "step": 1002
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8423628299014747e-05,
+      "loss": 1.5241,
+      "step": 1003
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8420043287612356e-05,
+      "loss": 1.6648,
+      "step": 1004
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8416454553918056e-05,
+      "loss": 1.5861,
+      "step": 1005
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8412862099518342e-05,
+      "loss": 1.6698,
+      "step": 1006
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8409265926001342e-05,
+      "loss": 1.6007,
+      "step": 1007
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8405666034956842e-05,
+      "loss": 1.6961,
+      "step": 1008
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8402062427976265e-05,
+      "loss": 1.624,
+      "step": 1009
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8398455106652677e-05,
+      "loss": 1.6918,
+      "step": 1010
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8394844072580772e-05,
+      "loss": 1.6346,
+      "step": 1011
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8391229327356916e-05,
+      "loss": 1.6172,
+      "step": 1012
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8387610872579085e-05,
+      "loss": 1.6911,
+      "step": 1013
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8383988709846918e-05,
+      "loss": 1.6641,
+      "step": 1014
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8380362840761675e-05,
+      "loss": 1.6758,
+      "step": 1015
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8376733266926265e-05,
+      "loss": 1.6212,
+      "step": 1016
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8373099989945236e-05,
+      "loss": 1.521,
+      "step": 1017
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8369463011424767e-05,
+      "loss": 1.6611,
+      "step": 1018
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8365822332972673e-05,
+      "loss": 1.6442,
+      "step": 1019
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8362177956198407e-05,
+      "loss": 1.5803,
+      "step": 1020
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8358529882713068e-05,
+      "loss": 1.6416,
+      "step": 1021
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8354878114129368e-05,
+      "loss": 1.5744,
+      "step": 1022
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.835122265206166e-05,
+      "loss": 1.6442,
+      "step": 1023
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8347563498125947e-05,
+      "loss": 1.6485,
+      "step": 1024
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8343900653939843e-05,
+      "loss": 1.7591,
+      "step": 1025
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.834023412112259e-05,
+      "loss": 1.6118,
+      "step": 1026
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833656390129509e-05,
+      "loss": 1.567,
+      "step": 1027
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8332889996079845e-05,
+      "loss": 1.5811,
+      "step": 1028
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8329212407100996e-05,
+      "loss": 1.6058,
+      "step": 1029
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8325531135984317e-05,
+      "loss": 1.6636,
+      "step": 1030
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832184618435721e-05,
+      "loss": 1.6372,
+      "step": 1031
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8318157553848694e-05,
+      "loss": 1.6864,
+      "step": 1032
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.831446524608942e-05,
+      "loss": 1.6134,
+      "step": 1033
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8310769262711673e-05,
+      "loss": 1.6917,
+      "step": 1034
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.830706960534935e-05,
+      "loss": 1.6631,
+      "step": 1035
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8303366275637977e-05,
+      "loss": 1.5771,
+      "step": 1036
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8299659275214708e-05,
+      "loss": 1.662,
+      "step": 1037
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8295948605718316e-05,
+      "loss": 1.6118,
+      "step": 1038
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.829223426878919e-05,
+      "loss": 1.6555,
+      "step": 1039
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828851626606935e-05,
+      "loss": 1.6158,
+      "step": 1040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8284794599202432e-05,
+      "loss": 1.6046,
+      "step": 1041
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8281069269833694e-05,
+      "loss": 1.6753,
+      "step": 1042
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.827734027961001e-05,
+      "loss": 1.6211,
+      "step": 1043
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.827360763017987e-05,
+      "loss": 1.5716,
+      "step": 1044
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8269871323193397e-05,
+      "loss": 1.6586,
+      "step": 1045
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8266131360302306e-05,
+      "loss": 1.6368,
+      "step": 1046
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.826238774315995e-05,
+      "loss": 1.608,
+      "step": 1047
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8258640473421288e-05,
+      "loss": 1.6658,
+      "step": 1048
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.825488955274289e-05,
+      "loss": 1.6118,
+      "step": 1049
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8251134982782952e-05,
+      "loss": 1.6327,
+      "step": 1050
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8247376765201276e-05,
+      "loss": 1.629,
+      "step": 1051
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8243614901659265e-05,
+      "loss": 1.6137,
+      "step": 1052
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8239849393819958e-05,
+      "loss": 1.6563,
+      "step": 1053
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.823608024334799e-05,
+      "loss": 1.6271,
+      "step": 1054
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8232307451909605e-05,
+      "loss": 1.6766,
+      "step": 1055
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8228531021172657e-05,
+      "loss": 1.6809,
+      "step": 1056
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8224750952806626e-05,
+      "loss": 1.5813,
+      "step": 1057
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8220967248482574e-05,
+      "loss": 1.6644,
+      "step": 1058
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8217179909873185e-05,
+      "loss": 1.5532,
+      "step": 1059
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.821338893865275e-05,
+      "loss": 1.634,
+      "step": 1060
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.820959433649716e-05,
+      "loss": 1.5808,
+      "step": 1061
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8205796105083917e-05,
+      "loss": 1.6189,
+      "step": 1062
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8201994246092124e-05,
+      "loss": 1.7472,
+      "step": 1063
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8198188761202487e-05,
+      "loss": 1.6294,
+      "step": 1064
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.819437965209732e-05,
+      "loss": 1.6214,
+      "step": 1065
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.819056692046053e-05,
+      "loss": 1.6458,
+      "step": 1066
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8186750567977638e-05,
+      "loss": 1.6046,
+      "step": 1067
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8182930596335752e-05,
+      "loss": 1.6624,
+      "step": 1068
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8179107007223588e-05,
+      "loss": 1.5712,
+      "step": 1069
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8175279802331466e-05,
+      "loss": 1.5947,
+      "step": 1070
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8171448983351284e-05,
+      "loss": 1.7121,
+      "step": 1071
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.816761455197657e-05,
+      "loss": 1.6978,
+      "step": 1072
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8163776509902414e-05,
+      "loss": 1.6802,
+      "step": 1073
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.815993485882553e-05,
+      "loss": 1.681,
+      "step": 1074
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8156089600444215e-05,
+      "loss": 1.5377,
+      "step": 1075
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.815224073645836e-05,
+      "loss": 1.5789,
+      "step": 1076
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8148388268569453e-05,
+      "loss": 1.5775,
+      "step": 1077
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.814453219848057e-05,
+      "loss": 1.6539,
+      "step": 1078
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.814067252789639e-05,
+      "loss": 1.6127,
+      "step": 1079
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8136809258523173e-05,
+      "loss": 1.6795,
+      "step": 1080
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.813294239206878e-05,
+      "loss": 1.625,
+      "step": 1081
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8129071930242648e-05,
+      "loss": 1.6659,
+      "step": 1082
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8125197874755823e-05,
+      "loss": 1.6172,
+      "step": 1083
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.812132022732092e-05,
+      "loss": 1.6419,
+      "step": 1084
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.811743898965215e-05,
+      "loss": 1.5627,
+      "step": 1085
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.811355416346532e-05,
+      "loss": 1.6621,
+      "step": 1086
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8109665750477806e-05,
+      "loss": 1.6314,
+      "step": 1087
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8105773752408584e-05,
+      "loss": 1.58,
+      "step": 1088
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.810187817097821e-05,
+      "loss": 1.5391,
+      "step": 1089
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8097979007908822e-05,
+      "loss": 1.6476,
+      "step": 1090
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.809407626492414e-05,
+      "loss": 1.6338,
+      "step": 1091
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8090169943749477e-05,
+      "loss": 1.6212,
+      "step": 1092
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8086260046111715e-05,
+      "loss": 1.5695,
+      "step": 1093
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8082346573739322e-05,
+      "loss": 1.6113,
+      "step": 1094
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.807842952836235e-05,
+      "loss": 1.6431,
+      "step": 1095
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.807450891171243e-05,
+      "loss": 1.5781,
+      "step": 1096
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8070584725522763e-05,
+      "loss": 1.6335,
+      "step": 1097
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8066656971528136e-05,
+      "loss": 1.6717,
+      "step": 1098
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8062725651464913e-05,
+      "loss": 1.6491,
+      "step": 1099
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.805879076707103e-05,
+      "loss": 1.6505,
+      "step": 1100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.805485232008601e-05,
+      "loss": 1.6515,
+      "step": 1101
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.805091031225093e-05,
+      "loss": 1.6128,
+      "step": 1102
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.804696474530846e-05,
+      "loss": 1.6556,
+      "step": 1103
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8043015621002843e-05,
+      "loss": 1.536,
+      "step": 1104
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.803906294107988e-05,
+      "loss": 1.6555,
+      "step": 1105
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8035106707286957e-05,
+      "loss": 1.5557,
+      "step": 1106
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.803114692137302e-05,
+      "loss": 1.6096,
+      "step": 1107
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.80271835850886e-05,
+      "loss": 1.6428,
+      "step": 1108
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8023216700185787e-05,
+      "loss": 1.6239,
+      "step": 1109
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.801924626841824e-05,
+      "loss": 1.5114,
+      "step": 1110
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8015272291541192e-05,
+      "loss": 1.6484,
+      "step": 1111
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8011294771311436e-05,
+      "loss": 1.6862,
+      "step": 1112
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8007313709487334e-05,
+      "loss": 1.5444,
+      "step": 1113
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8003329107828822e-05,
+      "loss": 1.593,
+      "step": 1114
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7999340968097386e-05,
+      "loss": 1.6724,
+      "step": 1115
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7995349292056087e-05,
+      "loss": 1.6072,
+      "step": 1116
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.799135408146954e-05,
+      "loss": 1.6413,
+      "step": 1117
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7987355338103935e-05,
+      "loss": 1.6719,
+      "step": 1118
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7983353063727014e-05,
+      "loss": 1.6071,
+      "step": 1119
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7979347260108088e-05,
+      "loss": 1.6617,
+      "step": 1120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7975337929018017e-05,
+      "loss": 1.6249,
+      "step": 1121
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7971325072229227e-05,
+      "loss": 1.6559,
+      "step": 1122
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7967308691515704e-05,
+      "loss": 1.6457,
+      "step": 1123
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7963288788652992e-05,
+      "loss": 1.5726,
+      "step": 1124
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7959265365418186e-05,
+      "loss": 1.5699,
+      "step": 1125
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7955238423589943e-05,
+      "loss": 1.612,
+      "step": 1126
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.795120796494848e-05,
+      "loss": 1.6354,
+      "step": 1127
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7947173991275556e-05,
+      "loss": 1.5475,
+      "step": 1128
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7943136504354485e-05,
+      "loss": 1.6364,
+      "step": 1129
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7939095505970156e-05,
+      "loss": 1.6182,
+      "step": 1130
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7935050997908977e-05,
+      "loss": 1.559,
+      "step": 1131
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7931002981958933e-05,
+      "loss": 1.676,
+      "step": 1132
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.792695145990955e-05,
+      "loss": 1.5872,
+      "step": 1133
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.792289643355191e-05,
+      "loss": 1.6032,
+      "step": 1134
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.791883790467863e-05,
+      "loss": 1.5966,
+      "step": 1135
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.791477587508389e-05,
+      "loss": 1.6939,
+      "step": 1136
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7910710346563417e-05,
+      "loss": 1.6108,
+      "step": 1137
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.790664132091447e-05,
+      "loss": 1.6978,
+      "step": 1138
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.790256879993587e-05,
+      "loss": 1.6908,
+      "step": 1139
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7898492785427973e-05,
+      "loss": 1.6178,
+      "step": 1140
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7894413279192693e-05,
+      "loss": 1.5668,
+      "step": 1141
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7890330283033467e-05,
+      "loss": 1.6238,
+      "step": 1142
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7886243798755297e-05,
+      "loss": 1.6,
+      "step": 1143
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7882153828164707e-05,
+      "loss": 1.605,
+      "step": 1144
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7878060373069778e-05,
+      "loss": 1.6204,
+      "step": 1145
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7873963435280122e-05,
+      "loss": 1.5986,
+      "step": 1146
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7869863016606893e-05,
+      "loss": 1.6243,
+      "step": 1147
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7865759118862784e-05,
+      "loss": 1.5164,
+      "step": 1148
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7861651743862025e-05,
+      "loss": 1.5872,
+      "step": 1149
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7857540893420387e-05,
+      "loss": 1.7671,
+      "step": 1150
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7853426569355173e-05,
+      "loss": 1.6584,
+      "step": 1151
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7849308773485226e-05,
+      "loss": 1.7002,
+      "step": 1152
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7845187507630922e-05,
+      "loss": 1.6708,
+      "step": 1153
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7841062773614165e-05,
+      "loss": 1.6346,
+      "step": 1154
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.78369345732584e-05,
+      "loss": 1.5545,
+      "step": 1155
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.78328029083886e-05,
+      "loss": 1.5605,
+      "step": 1156
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.782866778083128e-05,
+      "loss": 1.633,
+      "step": 1157
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7824529192414465e-05,
+      "loss": 1.6169,
+      "step": 1158
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.782038714496773e-05,
+      "loss": 1.6513,
+      "step": 1159
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.781624164032217e-05,
+      "loss": 1.5405,
+      "step": 1160
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7812092680310405e-05,
+      "loss": 1.6894,
+      "step": 1161
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7807940266766595e-05,
+      "loss": 1.66,
+      "step": 1162
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.780378440152641e-05,
+      "loss": 1.6922,
+      "step": 1163
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7799625086427064e-05,
+      "loss": 1.5744,
+      "step": 1164
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.779546232330728e-05,
+      "loss": 1.6982,
+      "step": 1165
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7791296114007314e-05,
+      "loss": 1.6854,
+      "step": 1166
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.778712646036894e-05,
+      "loss": 1.6768,
+      "step": 1167
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.778295336423547e-05,
+      "loss": 1.6185,
+      "step": 1168
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7778776827451715e-05,
+      "loss": 1.6526,
+      "step": 1169
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.777459685186402e-05,
+      "loss": 1.5037,
+      "step": 1170
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.777041343932025e-05,
+      "loss": 1.6046,
+      "step": 1171
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7766226591669787e-05,
+      "loss": 1.675,
+      "step": 1172
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7762036310763533e-05,
+      "loss": 1.6137,
+      "step": 1173
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7757842598453907e-05,
+      "loss": 1.6488,
+      "step": 1174
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7753645456594844e-05,
+      "loss": 1.6081,
+      "step": 1175
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7749444887041797e-05,
+      "loss": 1.6041,
+      "step": 1176
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7745240891651736e-05,
+      "loss": 1.5841,
+      "step": 1177
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7741033472283137e-05,
+      "loss": 1.5665,
+      "step": 1178
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7736822630796e-05,
+      "loss": 1.6403,
+      "step": 1179
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.773260836905183e-05,
+      "loss": 1.697,
+      "step": 1180
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7728390688913655e-05,
+      "loss": 1.5903,
+      "step": 1181
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7724169592245996e-05,
+      "loss": 1.6871,
+      "step": 1182
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7719945080914902e-05,
+      "loss": 1.654,
+      "step": 1183
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7715717156787923e-05,
+      "loss": 1.6733,
+      "step": 1184
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771148582173412e-05,
+      "loss": 1.566,
+      "step": 1185
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7707251077624056e-05,
+      "loss": 1.6809,
+      "step": 1186
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7703012926329813e-05,
+      "loss": 1.6299,
+      "step": 1187
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7698771369724967e-05,
+      "loss": 1.6069,
+      "step": 1188
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7694526409684608e-05,
+      "loss": 1.5711,
+      "step": 1189
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7690278048085327e-05,
+      "loss": 1.6625,
+      "step": 1190
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.768602628680522e-05,
+      "loss": 1.5967,
+      "step": 1191
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7681771127723883e-05,
+      "loss": 1.6667,
+      "step": 1192
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7677512572722416e-05,
+      "loss": 1.538,
+      "step": 1193
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7673250623683423e-05,
+      "loss": 1.5223,
+      "step": 1194
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7668985282491e-05,
+      "loss": 1.5767,
+      "step": 1195
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7664716551030752e-05,
+      "loss": 1.6491,
+      "step": 1196
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.766044443118978e-05,
+      "loss": 1.6591,
+      "step": 1197
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7656168924856686e-05,
+      "loss": 1.629,
+      "step": 1198
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7651890033921554e-05,
+      "loss": 1.6718,
+      "step": 1199
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7647607760275987e-05,
+      "loss": 1.6218,
+      "step": 1200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7643322105813065e-05,
+      "loss": 1.5588,
+      "step": 1201
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7639033072427367e-05,
+      "loss": 1.5505,
+      "step": 1202
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7634740662014977e-05,
+      "loss": 1.6463,
+      "step": 1203
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.763044487647345e-05,
+      "loss": 1.6214,
+      "step": 1204
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7626145717701862e-05,
+      "loss": 1.5721,
+      "step": 1205
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7621843187600754e-05,
+      "loss": 1.5465,
+      "step": 1206
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.761753728807217e-05,
+      "loss": 1.5972,
+      "step": 1207
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.761322802101964e-05,
+      "loss": 1.6394,
+      "step": 1208
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7608915388348185e-05,
+      "loss": 1.6038,
+      "step": 1209
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7604599391964315e-05,
+      "loss": 1.578,
+      "step": 1210
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7600280033776018e-05,
+      "loss": 1.4491,
+      "step": 1211
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7595957315692782e-05,
+      "loss": 1.6803,
+      "step": 1212
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.759163123962557e-05,
+      "loss": 1.6816,
+      "step": 1213
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7587301807486835e-05,
+      "loss": 1.606,
+      "step": 1214
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7582969021190507e-05,
+      "loss": 1.6734,
+      "step": 1215
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7578632882652006e-05,
+      "loss": 1.6662,
+      "step": 1216
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7574293393788236e-05,
+      "loss": 1.607,
+      "step": 1217
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7569950556517566e-05,
+      "loss": 1.6543,
+      "step": 1218
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7565604372759865e-05,
+      "loss": 1.6227,
+      "step": 1219
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.756125484443647e-05,
+      "loss": 1.665,
+      "step": 1220
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.75569019734702e-05,
+      "loss": 1.595,
+      "step": 1221
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.755254576178535e-05,
+      "loss": 1.5895,
+      "step": 1222
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7548186211307694e-05,
+      "loss": 1.6965,
+      "step": 1223
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.754382332396448e-05,
+      "loss": 1.6074,
+      "step": 1224
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7539457101684434e-05,
+      "loss": 1.6515,
+      "step": 1225
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7535087546397754e-05,
+      "loss": 1.5966,
+      "step": 1226
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7530714660036112e-05,
+      "loss": 1.5986,
+      "step": 1227
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7526338444532648e-05,
+      "loss": 1.6805,
+      "step": 1228
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7521958901821984e-05,
+      "loss": 1.609,
+      "step": 1229
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7517576033840204e-05,
+      "loss": 1.5896,
+      "step": 1230
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7513189842524865e-05,
+      "loss": 1.6939,
+      "step": 1231
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7508800329814993e-05,
+      "loss": 1.7009,
+      "step": 1232
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7504407497651085e-05,
+      "loss": 1.6441,
+      "step": 1233
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.75000113479751e-05,
+      "loss": 1.6293,
+      "step": 1234
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.749561188273047e-05,
+      "loss": 1.6775,
+      "step": 1235
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7491209103862083e-05,
+      "loss": 1.6425,
+      "step": 1236
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.74868030133163e-05,
+      "loss": 1.5778,
+      "step": 1237
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7482393613040954e-05,
+      "loss": 1.5782,
+      "step": 1238
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.747798090498532e-05,
+      "loss": 1.6202,
+      "step": 1239
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.747356489110015e-05,
+      "loss": 1.6323,
+      "step": 1240
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7469145573337654e-05,
+      "loss": 1.5881,
+      "step": 1241
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7464722953651504e-05,
+      "loss": 1.553,
+      "step": 1242
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7460297033996827e-05,
+      "loss": 1.5757,
+      "step": 1243
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7455867816330215e-05,
+      "loss": 1.7018,
+      "step": 1244
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7451435302609713e-05,
+      "loss": 1.6107,
+      "step": 1245
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.744699949479483e-05,
+      "loss": 1.6951,
+      "step": 1246
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7442560394846518e-05,
+      "loss": 1.6662,
+      "step": 1247
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7438118004727197e-05,
+      "loss": 1.6415,
+      "step": 1248
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.743367232640074e-05,
+      "loss": 1.5901,
+      "step": 1249
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7429223361832474e-05,
+      "loss": 1.6375,
+      "step": 1250
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.742477111298916e-05,
+      "loss": 1.6077,
+      "step": 1251
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7420315581839045e-05,
+      "loss": 1.6497,
+      "step": 1252
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7415856770351797e-05,
+      "loss": 1.54,
+      "step": 1253
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.741139468049855e-05,
+      "loss": 1.592,
+      "step": 1254
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7406929314251883e-05,
+      "loss": 1.6156,
+      "step": 1255
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7402460673585823e-05,
+      "loss": 1.6022,
+      "step": 1256
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7397988760475842e-05,
+      "loss": 1.6628,
+      "step": 1257
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7393513576898868e-05,
+      "loss": 1.6627,
+      "step": 1258
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.738903512483326e-05,
+      "loss": 1.5699,
+      "step": 1259
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7384553406258842e-05,
+      "loss": 1.625,
+      "step": 1260
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7380068423156856e-05,
+      "loss": 1.6655,
+      "step": 1261
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7375580177510017e-05,
+      "loss": 1.6118,
+      "step": 1262
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.737108867130245e-05,
+      "loss": 1.6572,
+      "step": 1263
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7366593906519758e-05,
+      "loss": 1.6109,
+      "step": 1264
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.736209588514895e-05,
+      "loss": 1.6529,
+      "step": 1265
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.735759460917849e-05,
+      "loss": 1.668,
+      "step": 1266
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.735309008059829e-05,
+      "loss": 1.6035,
+      "step": 1267
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7348582301399682e-05,
+      "loss": 1.6204,
+      "step": 1268
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7344071273575443e-05,
+      "loss": 1.6254,
+      "step": 1269
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7339556999119785e-05,
+      "loss": 1.6646,
+      "step": 1270
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.733503948002837e-05,
+      "loss": 1.6553,
+      "step": 1271
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7330518718298263e-05,
+      "loss": 1.5871,
+      "step": 1272
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7325994715927995e-05,
+      "loss": 1.6091,
+      "step": 1273
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7321467474917502e-05,
+      "loss": 1.5411,
+      "step": 1274
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7316936997268178e-05,
+      "loss": 1.6134,
+      "step": 1275
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.731240328498282e-05,
+      "loss": 1.6087,
+      "step": 1276
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7307866340065684e-05,
+      "loss": 1.6328,
+      "step": 1277
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7303326164522433e-05,
+      "loss": 1.6249,
+      "step": 1278
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7298782760360167e-05,
+      "loss": 1.7081,
+      "step": 1279
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.729423612958741e-05,
+      "loss": 1.6301,
+      "step": 1280
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7289686274214116e-05,
+      "loss": 1.6122,
+      "step": 1281
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7285133196251664e-05,
+      "loss": 1.5972,
+      "step": 1282
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7280576897712856e-05,
+      "loss": 1.6365,
+      "step": 1283
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7276017380611922e-05,
+      "loss": 1.6197,
+      "step": 1284
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7271454646964503e-05,
+      "loss": 1.6468,
+      "step": 1285
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7266888698787676e-05,
+      "loss": 1.5744,
+      "step": 1286
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.726231953809993e-05,
+      "loss": 1.6719,
+      "step": 1287
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7257747166921186e-05,
+      "loss": 1.5971,
+      "step": 1288
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7253171587272764e-05,
+      "loss": 1.6854,
+      "step": 1289
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.724859280117742e-05,
+      "loss": 1.6109,
+      "step": 1290
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.724401081065932e-05,
+      "loss": 1.5927,
+      "step": 1291
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.723942561774405e-05,
+      "loss": 1.6426,
+      "step": 1292
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7234837224458606e-05,
+      "loss": 1.6637,
+      "step": 1293
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.723024563283141e-05,
+      "loss": 1.6573,
+      "step": 1294
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.722565084489228e-05,
+      "loss": 1.5728,
+      "step": 1295
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.722105286267247e-05,
+      "loss": 1.6432,
+      "step": 1296
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7216451688204623e-05,
+      "loss": 1.642,
+      "step": 1297
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7211847323522806e-05,
+      "loss": 1.6975,
+      "step": 1298
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7207239770662502e-05,
+      "loss": 1.6655,
+      "step": 1299
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7202629031660587e-05,
+      "loss": 1.5738,
+      "step": 1300
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7198015108555358e-05,
+      "loss": 1.5974,
+      "step": 1301
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7193398003386514e-05,
+      "loss": 1.6097,
+      "step": 1302
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.718877771819516e-05,
+      "loss": 1.6097,
+      "step": 1303
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7184154255023817e-05,
+      "loss": 1.683,
+      "step": 1304
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7179527615916397e-05,
+      "loss": 1.6166,
+      "step": 1305
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7174897802918223e-05,
+      "loss": 1.7077,
+      "step": 1306
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7170264818076027e-05,
+      "loss": 1.6371,
+      "step": 1307
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7165628663437923e-05,
+      "loss": 1.5902,
+      "step": 1308
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.716098934105345e-05,
+      "loss": 1.5921,
+      "step": 1309
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7156346852973538e-05,
+      "loss": 1.6116,
+      "step": 1310
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7151701201250505e-05,
+      "loss": 1.5618,
+      "step": 1311
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7147052387938094e-05,
+      "loss": 1.5217,
+      "step": 1312
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7142400415091413e-05,
+      "loss": 1.6416,
+      "step": 1313
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7137745284766995e-05,
+      "loss": 1.6308,
+      "step": 1314
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7133086999022756e-05,
+      "loss": 1.657,
+      "step": 1315
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7128425559918006e-05,
+      "loss": 1.5611,
+      "step": 1316
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.712376096951345e-05,
+      "loss": 1.6321,
+      "step": 1317
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.711909322987119e-05,
+      "loss": 1.6054,
+      "step": 1318
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.711442234305472e-05,
+      "loss": 1.5925,
+      "step": 1319
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7109748311128918e-05,
+      "loss": 1.6142,
+      "step": 1320
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.710507113616006e-05,
+      "loss": 1.5341,
+      "step": 1321
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7100390820215805e-05,
+      "loss": 1.6193,
+      "step": 1322
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.709570736536521e-05,
+      "loss": 1.6663,
+      "step": 1323
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7091020773678712e-05,
+      "loss": 1.5551,
+      "step": 1324
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.708633104722814e-05,
+      "loss": 1.6913,
+      "step": 1325
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7081638188086698e-05,
+      "loss": 1.6394,
+      "step": 1326
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7076942198328987e-05,
+      "loss": 1.489,
+      "step": 1327
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7072243080030986e-05,
+      "loss": 1.6811,
+      "step": 1328
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.706754083527006e-05,
+      "loss": 1.583,
+      "step": 1329
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7062835466124953e-05,
+      "loss": 1.7238,
+      "step": 1330
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7058126974675798e-05,
+      "loss": 1.5522,
+      "step": 1331
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.705341536300409e-05,
+      "loss": 1.6503,
+      "step": 1332
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7048700633192725e-05,
+      "loss": 1.6113,
+      "step": 1333
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7043982787325966e-05,
+      "loss": 1.5888,
+      "step": 1334
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7039261827489452e-05,
+      "loss": 1.6081,
+      "step": 1335
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7034537755770204e-05,
+      "loss": 1.6025,
+      "step": 1336
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.702981057425662e-05,
+      "loss": 1.673,
+      "step": 1337
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7025080285038465e-05,
+      "loss": 1.6537,
+      "step": 1338
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7020346890206884e-05,
+      "loss": 1.6586,
+      "step": 1339
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7015610391854393e-05,
+      "loss": 1.6209,
+      "step": 1340
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.701087079207488e-05,
+      "loss": 1.6648,
+      "step": 1341
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7006128092963604e-05,
+      "loss": 1.5542,
+      "step": 1342
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7001382296617196e-05,
+      "loss": 1.6399,
+      "step": 1343
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6996633405133656e-05,
+      "loss": 1.5523,
+      "step": 1344
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6991881420612347e-05,
+      "loss": 1.657,
+      "step": 1345
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6987126345154003e-05,
+      "loss": 1.594,
+      "step": 1346
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.698236818086073e-05,
+      "loss": 1.5724,
+      "step": 1347
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.697760692983599e-05,
+      "loss": 1.6246,
+      "step": 1348
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6972842594184613e-05,
+      "loss": 1.6689,
+      "step": 1349
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6968075176012795e-05,
+      "loss": 1.6128,
+      "step": 1350
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6963304677428096e-05,
+      "loss": 1.588,
+      "step": 1351
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6958531100539428e-05,
+      "loss": 1.5959,
+      "step": 1352
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6953754447457078e-05,
+      "loss": 1.6306,
+      "step": 1353
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.694897472029268e-05,
+      "loss": 1.6231,
+      "step": 1354
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6944191921159238e-05,
+      "loss": 1.648,
+      "step": 1355
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6939406052171103e-05,
+      "loss": 1.613,
+      "step": 1356
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6934617115443993e-05,
+      "loss": 1.6258,
+      "step": 1357
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6929825113094972e-05,
+      "loss": 1.6094,
+      "step": 1358
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6925030047242473e-05,
+      "loss": 1.6308,
+      "step": 1359
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6920231920006275e-05,
+      "loss": 1.6001,
+      "step": 1360
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6915430733507503e-05,
+      "loss": 1.5782,
+      "step": 1361
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.691062648986865e-05,
+      "loss": 1.6361,
+      "step": 1362
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6905819191213546e-05,
+      "loss": 1.7259,
+      "step": 1363
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6901008839667384e-05,
+      "loss": 1.614,
+      "step": 1364
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.68961954373567e-05,
+      "loss": 1.6507,
+      "step": 1365
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6891378986409378e-05,
+      "loss": 1.5769,
+      "step": 1366
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6886559488954647e-05,
+      "loss": 1.6692,
+      "step": 1367
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.68817369471231e-05,
+      "loss": 1.647,
+      "step": 1368
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6876911363046647e-05,
+      "loss": 1.5502,
+      "step": 1369
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6872082738858574e-05,
+      "loss": 1.6484,
+      "step": 1370
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6867251076693483e-05,
+      "loss": 1.627,
+      "step": 1371
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.686241637868734e-05,
+      "loss": 1.6028,
+      "step": 1372
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.685757864697744e-05,
+      "loss": 1.6192,
+      "step": 1373
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6852737883702423e-05,
+      "loss": 1.618,
+      "step": 1374
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6847894091002274e-05,
+      "loss": 1.6066,
+      "step": 1375
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.684304727101832e-05,
+      "loss": 1.6175,
+      "step": 1376
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.68381974258932e-05,
+      "loss": 1.6289,
+      "step": 1377
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.683334455777093e-05,
+      "loss": 1.5672,
+      "step": 1378
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6828488668796836e-05,
+      "loss": 1.6154,
+      "step": 1379
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.682362976111758e-05,
+      "loss": 1.5721,
+      "step": 1380
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.681876783688117e-05,
+      "loss": 1.558,
+      "step": 1381
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.681390289823694e-05,
+      "loss": 1.5731,
+      "step": 1382
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6809034947335557e-05,
+      "loss": 1.5257,
+      "step": 1383
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.680416398632903e-05,
+      "loss": 1.6544,
+      "step": 1384
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6799290017370677e-05,
+      "loss": 1.6097,
+      "step": 1385
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6794413042615168e-05,
+      "loss": 1.5713,
+      "step": 1386
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6789533064218487e-05,
+      "loss": 1.6742,
+      "step": 1387
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.678465008433795e-05,
+      "loss": 1.6436,
+      "step": 1388
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.677976410513221e-05,
+      "loss": 1.57,
+      "step": 1389
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6774875128761228e-05,
+      "loss": 1.5958,
+      "step": 1390
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.67699831573863e-05,
+      "loss": 1.55,
+      "step": 1391
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6765088193170055e-05,
+      "loss": 1.5531,
+      "step": 1392
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6760190238276418e-05,
+      "loss": 1.6674,
+      "step": 1393
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.675528929487067e-05,
+      "loss": 1.7039,
+      "step": 1394
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6750385365119383e-05,
+      "loss": 1.6193,
+      "step": 1395
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.674547845119047e-05,
+      "loss": 1.5164,
+      "step": 1396
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6740568555253153e-05,
+      "loss": 1.6592,
+      "step": 1397
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.673565567947798e-05,
+      "loss": 1.5272,
+      "step": 1398
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6730739826036805e-05,
+      "loss": 1.7193,
+      "step": 1399
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6725820997102804e-05,
+      "loss": 1.5812,
+      "step": 1400
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.672089919485047e-05,
+      "loss": 1.6482,
+      "step": 1401
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6715974421455615e-05,
+      "loss": 1.5839,
+      "step": 1402
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6711046679095356e-05,
+      "loss": 1.5763,
+      "step": 1403
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6706115969948124e-05,
+      "loss": 1.5988,
+      "step": 1404
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6701182296193662e-05,
+      "loss": 1.6682,
+      "step": 1405
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6696245660013027e-05,
+      "loss": 1.5906,
+      "step": 1406
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6691306063588583e-05,
+      "loss": 1.5996,
+      "step": 1407
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6686363509104002e-05,
+      "loss": 1.516,
+      "step": 1408
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6681417998744265e-05,
+      "loss": 1.5747,
+      "step": 1409
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6676469534695657e-05,
+      "loss": 1.6665,
+      "step": 1410
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6671518119145776e-05,
+      "loss": 1.6606,
+      "step": 1411
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6666563754283517e-05,
+      "loss": 1.5325,
+      "step": 1412
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.666160644229908e-05,
+      "loss": 1.5945,
+      "step": 1413
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.665664618538397e-05,
+      "loss": 1.6298,
+      "step": 1414
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6651682985730993e-05,
+      "loss": 1.5505,
+      "step": 1415
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.664671684553426e-05,
+      "loss": 1.5773,
+      "step": 1416
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6641747766989173e-05,
+      "loss": 1.6359,
+      "step": 1417
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.663677575229244e-05,
+      "loss": 1.5474,
+      "step": 1418
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6631800803642067e-05,
+      "loss": 1.5948,
+      "step": 1419
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6626822923237356e-05,
+      "loss": 1.662,
+      "step": 1420
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6621842113278902e-05,
+      "loss": 1.7078,
+      "step": 1421
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6616858375968596e-05,
+      "loss": 1.6396,
+      "step": 1422
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6611871713509628e-05,
+      "loss": 1.5765,
+      "step": 1423
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6606882128106472e-05,
+      "loss": 1.6261,
+      "step": 1424
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6601889621964903e-05,
+      "loss": 1.5622,
+      "step": 1425
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.659689419729199e-05,
+      "loss": 1.5144,
+      "step": 1426
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6591895856296075e-05,
+      "loss": 1.631,
+      "step": 1427
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6586894601186804e-05,
+      "loss": 1.5589,
+      "step": 1428
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6581890434175117e-05,
+      "loss": 1.5875,
+      "step": 1429
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.657688335747322e-05,
+      "loss": 1.6465,
+      "step": 1430
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.657187337329462e-05,
+      "loss": 1.5456,
+      "step": 1431
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6566860483854106e-05,
+      "loss": 1.6714,
+      "step": 1432
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6561844691367755e-05,
+      "loss": 1.5623,
+      "step": 1433
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6556825998052924e-05,
+      "loss": 1.5611,
+      "step": 1434
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.655180440612825e-05,
+      "loss": 1.631,
+      "step": 1435
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6546779917813658e-05,
+      "loss": 1.552,
+      "step": 1436
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6541752535330345e-05,
+      "loss": 1.5702,
+      "step": 1437
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6536722260900795e-05,
+      "loss": 1.5813,
+      "step": 1438
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6531689096748767e-05,
+      "loss": 1.6917,
+      "step": 1439
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6526653045099294e-05,
+      "loss": 1.6356,
+      "step": 1440
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6521614108178694e-05,
+      "loss": 1.6405,
+      "step": 1441
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6516572288214555e-05,
+      "loss": 1.6738,
+      "step": 1442
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6511527587435736e-05,
+      "loss": 1.595,
+      "step": 1443
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6506480008072377e-05,
+      "loss": 1.5872,
+      "step": 1444
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.650142955235589e-05,
+      "loss": 1.6381,
+      "step": 1445
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6496376222518958e-05,
+      "loss": 1.5406,
+      "step": 1446
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.649132002079552e-05,
+      "loss": 1.5423,
+      "step": 1447
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6486260949420808e-05,
+      "loss": 1.6342,
+      "step": 1448
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6481199010631312e-05,
+      "loss": 1.5541,
+      "step": 1449
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6476134206664783e-05,
+      "loss": 1.6325,
+      "step": 1450
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.647106653976025e-05,
+      "loss": 1.6762,
+      "step": 1451
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6465996012157996e-05,
+      "loss": 1.6021,
+      "step": 1452
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6460922626099586e-05,
+      "loss": 1.6339,
+      "step": 1453
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6455846383827832e-05,
+      "loss": 1.6605,
+      "step": 1454
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.645076728758681e-05,
+      "loss": 1.6333,
+      "step": 1455
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.644568533962187e-05,
+      "loss": 1.5489,
+      "step": 1456
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6440600542179613e-05,
+      "loss": 1.6033,
+      "step": 1457
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6435512897507896e-05,
+      "loss": 1.6387,
+      "step": 1458
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6430422407855852e-05,
+      "loss": 1.5126,
+      "step": 1459
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6425329075473847e-05,
+      "loss": 1.589,
+      "step": 1460
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6420232902613525e-05,
+      "loss": 1.6063,
+      "step": 1461
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.641513389152777e-05,
+      "loss": 1.6669,
+      "step": 1462
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6410032044470735e-05,
+      "loss": 1.6306,
+      "step": 1463
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.640492736369782e-05,
+      "loss": 1.5557,
+      "step": 1464
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6399819851465674e-05,
+      "loss": 1.6292,
+      "step": 1465
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6394709510032204e-05,
+      "loss": 1.6566,
+      "step": 1466
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.638959634165656e-05,
+      "loss": 1.6412,
+      "step": 1467
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.638448034859915e-05,
+      "loss": 1.5684,
+      "step": 1468
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.637936153312163e-05,
+      "loss": 1.5731,
+      "step": 1469
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.63742398974869e-05,
+      "loss": 1.6205,
+      "step": 1470
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6369115443959102e-05,
+      "loss": 1.5534,
+      "step": 1471
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6363988174803638e-05,
+      "loss": 1.6416,
+      "step": 1472
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.635885809228714e-05,
+      "loss": 1.5729,
+      "step": 1473
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6353725198677494e-05,
+      "loss": 1.5291,
+      "step": 1474
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6348589496243818e-05,
+      "loss": 1.6424,
+      "step": 1475
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6343450987256487e-05,
+      "loss": 1.645,
+      "step": 1476
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.63383096739871e-05,
+      "loss": 1.5867,
+      "step": 1477
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6333165558708515e-05,
+      "loss": 1.5816,
+      "step": 1478
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.632801864369481e-05,
+      "loss": 1.6573,
+      "step": 1479
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.632286893122131e-05,
+      "loss": 1.6079,
+      "step": 1480
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.631771642356457e-05,
+      "loss": 1.5683,
+      "step": 1481
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.631256112300239e-05,
+      "loss": 1.5929,
+      "step": 1482
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6307403031813803e-05,
+      "loss": 1.5283,
+      "step": 1483
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6302242152279068e-05,
+      "loss": 1.5312,
+      "step": 1484
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6297078486679683e-05,
+      "loss": 1.6878,
+      "step": 1485
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6291912037298376e-05,
+      "loss": 1.6077,
+      "step": 1486
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.628674280641911e-05,
+      "loss": 1.6566,
+      "step": 1487
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.628157079632707e-05,
+      "loss": 1.5891,
+      "step": 1488
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.627639600930867e-05,
+      "loss": 1.5935,
+      "step": 1489
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6271218447651556e-05,
+      "loss": 1.6182,
+      "step": 1490
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6266038113644605e-05,
+      "loss": 1.6202,
+      "step": 1491
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6260855009577912e-05,
+      "loss": 1.6356,
+      "step": 1492
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6255669137742793e-05,
+      "loss": 1.583,
+      "step": 1493
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.62504805004318e-05,
+      "loss": 1.6325,
+      "step": 1494
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6245289099938695e-05,
+      "loss": 1.6182,
+      "step": 1495
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6240094938558473e-05,
+      "loss": 1.5951,
+      "step": 1496
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6234898018587336e-05,
+      "loss": 1.5996,
+      "step": 1497
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.622969834232272e-05,
+      "loss": 1.5939,
+      "step": 1498
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.622449591206327e-05,
+      "loss": 1.652,
+      "step": 1499
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6219290730108853e-05,
+      "loss": 1.5731,
+      "step": 1500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6214082798760545e-05,
+      "loss": 1.5402,
+      "step": 1501
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6208872120320647e-05,
+      "loss": 1.6117,
+      "step": 1502
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6203658697092673e-05,
+      "loss": 1.5986,
+      "step": 1503
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6198442531381348e-05,
+      "loss": 1.593,
+      "step": 1504
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6193223625492604e-05,
+      "loss": 1.647,
+      "step": 1505
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.618800198173359e-05,
+      "loss": 1.5389,
+      "step": 1506
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.618277760241267e-05,
+      "loss": 1.5798,
+      "step": 1507
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6177550489839403e-05,
+      "loss": 1.5868,
+      "step": 1508
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6172320646324576e-05,
+      "loss": 1.6955,
+      "step": 1509
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.616708807418017e-05,
+      "loss": 1.5839,
+      "step": 1510
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6161852775719377e-05,
+      "loss": 1.5918,
+      "step": 1511
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6156614753256583e-05,
+      "loss": 1.5981,
+      "step": 1512
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6151374009107398e-05,
+      "loss": 1.6086,
+      "step": 1513
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.614613054558862e-05,
+      "loss": 1.6224,
+      "step": 1514
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6140884365018253e-05,
+      "loss": 1.5001,
+      "step": 1515
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6135635469715505e-05,
+      "loss": 1.6405,
+      "step": 1516
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6130383862000783e-05,
+      "loss": 1.6181,
+      "step": 1517
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.612512954419569e-05,
+      "loss": 1.5986,
+      "step": 1518
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.611987251862303e-05,
+      "loss": 1.6463,
+      "step": 1519
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6114612787606802e-05,
+      "loss": 1.6159,
+      "step": 1520
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6109350353472207e-05,
+      "loss": 1.6235,
+      "step": 1521
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6104085218545633e-05,
+      "loss": 1.6317,
+      "step": 1522
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6098817385154665e-05,
+      "loss": 1.6092,
+      "step": 1523
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6093546855628085e-05,
+      "loss": 1.6903,
+      "step": 1524
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6088273632295857e-05,
+      "loss": 1.6049,
+      "step": 1525
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6082997717489145e-05,
+      "loss": 1.6369,
+      "step": 1526
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6077719113540303e-05,
+      "loss": 1.609,
+      "step": 1527
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.607243782278286e-05,
+      "loss": 1.589,
+      "step": 1528
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.606715384755156e-05,
+      "loss": 1.6208,
+      "step": 1529
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6061867190182304e-05,
+      "loss": 1.6622,
+      "step": 1530
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.60565778530122e-05,
+      "loss": 1.6021,
+      "step": 1531
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6051285838379525e-05,
+      "loss": 1.5738,
+      "step": 1532
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6045991148623752e-05,
+      "loss": 1.5351,
+      "step": 1533
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.604069378608553e-05,
+      "loss": 1.5596,
+      "step": 1534
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.603539375310669e-05,
+      "loss": 1.6999,
+      "step": 1535
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.603009105203025e-05,
+      "loss": 1.6335,
+      "step": 1536
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6024785685200396e-05,
+      "loss": 1.567,
+      "step": 1537
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6019477654962506e-05,
+      "loss": 1.6242,
+      "step": 1538
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.601416696366312e-05,
+      "loss": 1.5913,
+      "step": 1539
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.600885361364997e-05,
+      "loss": 1.6096,
+      "step": 1540
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6003537607271948e-05,
+      "loss": 1.4961,
+      "step": 1541
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.599821894687914e-05,
+      "loss": 1.6368,
+      "step": 1542
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.599289763482278e-05,
+      "loss": 1.6124,
+      "step": 1543
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5987573673455302e-05,
+      "loss": 1.5794,
+      "step": 1544
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5982247065130284e-05,
+      "loss": 1.5431,
+      "step": 1545
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5976917812202498e-05,
+      "loss": 1.7059,
+      "step": 1546
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5971585917027864e-05,
+      "loss": 1.6137,
+      "step": 1547
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5966251381963486e-05,
+      "loss": 1.602,
+      "step": 1548
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5960914209367632e-05,
+      "loss": 1.5095,
+      "step": 1549
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5955574401599726e-05,
+      "loss": 1.5998,
+      "step": 1550
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5950231961020373e-05,
+      "loss": 1.5731,
+      "step": 1551
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5944886889991326e-05,
+      "loss": 1.5864,
+      "step": 1552
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5939539190875512e-05,
+      "loss": 1.5296,
+      "step": 1553
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5934188866037017e-05,
+      "loss": 1.5423,
+      "step": 1554
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.592883591784108e-05,
+      "loss": 1.6748,
+      "step": 1555
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5923480348654116e-05,
+      "loss": 1.6377,
+      "step": 1556
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.591812216084368e-05,
+      "loss": 1.5631,
+      "step": 1557
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5912761356778498e-05,
+      "loss": 1.6125,
+      "step": 1558
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5907397938828447e-05,
+      "loss": 1.5525,
+      "step": 1559
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5902031909364563e-05,
+      "loss": 1.5361,
+      "step": 1560
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5896663270759034e-05,
+      "loss": 1.5665,
+      "step": 1561
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.58912920253852e-05,
+      "loss": 1.5607,
+      "step": 1562
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5885918175617552e-05,
+      "loss": 1.5345,
+      "step": 1563
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5880541723831745e-05,
+      "loss": 1.5776,
+      "step": 1564
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.587516267240457e-05,
+      "loss": 1.6265,
+      "step": 1565
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5869781023713964e-05,
+      "loss": 1.5754,
+      "step": 1566
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.586439678013903e-05,
+      "loss": 1.7225,
+      "step": 1567
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5859009944060005e-05,
+      "loss": 1.6285,
+      "step": 1568
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5853620517858276e-05,
+      "loss": 1.6524,
+      "step": 1569
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5848228503916375e-05,
+      "loss": 1.5673,
+      "step": 1570
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5842833904617978e-05,
+      "loss": 1.5817,
+      "step": 1571
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5837436722347902e-05,
+      "loss": 1.5405,
+      "step": 1572
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5832036959492104e-05,
+      "loss": 1.5688,
+      "step": 1573
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.582663461843769e-05,
+      "loss": 1.5428,
+      "step": 1574
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5821229701572897e-05,
+      "loss": 1.6447,
+      "step": 1575
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5815822211287106e-05,
+      "loss": 1.6013,
+      "step": 1576
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5810412149970832e-05,
+      "loss": 1.5625,
+      "step": 1577
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5804999520015735e-05,
+      "loss": 1.6252,
+      "step": 1578
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.57995843238146e-05,
+      "loss": 1.6011,
+      "step": 1579
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.579416656376135e-05,
+      "loss": 1.5555,
+      "step": 1580
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.578874624225104e-05,
+      "loss": 1.5477,
+      "step": 1581
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5783323361679865e-05,
+      "loss": 1.5932,
+      "step": 1582
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5777897924445144e-05,
+      "loss": 1.59,
+      "step": 1583
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5772469932945328e-05,
+      "loss": 1.5825,
+      "step": 1584
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5767039389579994e-05,
+      "loss": 1.5367,
+      "step": 1585
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.576160629674985e-05,
+      "loss": 1.5695,
+      "step": 1586
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.575617065685674e-05,
+      "loss": 1.5355,
+      "step": 1587
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5750732472303612e-05,
+      "loss": 1.5462,
+      "step": 1588
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5745291745494563e-05,
+      "loss": 1.646,
+      "step": 1589
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.57398484788348e-05,
+      "loss": 1.5324,
+      "step": 1590
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5734402674730652e-05,
+      "loss": 1.5964,
+      "step": 1591
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.572895433558958e-05,
+      "loss": 1.5874,
+      "step": 1592
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.572350346382016e-05,
+      "loss": 1.5611,
+      "step": 1593
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5718050061832078e-05,
+      "loss": 1.6467,
+      "step": 1594
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5712594132036153e-05,
+      "loss": 1.6635,
+      "step": 1595
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.570713567684432e-05,
+      "loss": 1.6809,
+      "step": 1596
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.570167469866962e-05,
+      "loss": 1.526,
+      "step": 1597
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5696211199926217e-05,
+      "loss": 1.5504,
+      "step": 1598
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5690745183029388e-05,
+      "loss": 1.636,
+      "step": 1599
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5685276650395527e-05,
+      "loss": 1.5375,
+      "step": 1600
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5679805604442134e-05,
+      "loss": 1.5789,
+      "step": 1601
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.567433204758782e-05,
+      "loss": 1.5838,
+      "step": 1602
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.566885598225231e-05,
+      "loss": 1.6191,
+      "step": 1603
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5663377410856438e-05,
+      "loss": 1.616,
+      "step": 1604
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5657896335822147e-05,
+      "loss": 1.6481,
+      "step": 1605
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5652412759572478e-05,
+      "loss": 1.5012,
+      "step": 1606
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5646926684531586e-05,
+      "loss": 1.5525,
+      "step": 1607
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5641438113124726e-05,
+      "loss": 1.5833,
+      "step": 1608
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5635947047778265e-05,
+      "loss": 1.5614,
+      "step": 1609
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5630453490919663e-05,
+      "loss": 1.604,
+      "step": 1610
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5624957444977482e-05,
+      "loss": 1.6404,
+      "step": 1611
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5619458912381397e-05,
+      "loss": 1.6088,
+      "step": 1612
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.561395789556216e-05,
+      "loss": 1.6054,
+      "step": 1613
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5608454396951646e-05,
+      "loss": 1.5761,
+      "step": 1614
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.56029484189828e-05,
+      "loss": 1.4894,
+      "step": 1615
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5597439964089693e-05,
+      "loss": 1.5249,
+      "step": 1616
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5591929034707468e-05,
+      "loss": 1.6223,
+      "step": 1617
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5586415633272377e-05,
+      "loss": 1.5245,
+      "step": 1618
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5580899762221746e-05,
+      "loss": 1.7127,
+      "step": 1619
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5575381423994015e-05,
+      "loss": 1.6112,
+      "step": 1620
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5569860621028703e-05,
+      "loss": 1.6009,
+      "step": 1621
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5564337355766412e-05,
+      "loss": 1.5646,
+      "step": 1622
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5558811630648848e-05,
+      "loss": 1.6383,
+      "step": 1623
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5553283448118795e-05,
+      "loss": 1.5735,
+      "step": 1624
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.554775281062012e-05,
+      "loss": 1.4877,
+      "step": 1625
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.554221972059779e-05,
+      "loss": 1.4599,
+      "step": 1626
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5536684180497838e-05,
+      "loss": 1.4145,
+      "step": 1627
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5531146192767396e-05,
+      "loss": 1.3499,
+      "step": 1628
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5525605759854666e-05,
+      "loss": 1.3401,
+      "step": 1629
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5520062884208935e-05,
+      "loss": 1.4198,
+      "step": 1630
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5514517568280573e-05,
+      "loss": 1.3861,
+      "step": 1631
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5508969814521026e-05,
+      "loss": 1.4597,
+      "step": 1632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.550341962538282e-05,
+      "loss": 1.3893,
+      "step": 1633
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.549786700331955e-05,
+      "loss": 1.4233,
+      "step": 1634
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5492311950785895e-05,
+      "loss": 1.3617,
+      "step": 1635
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.548675447023761e-05,
+      "loss": 1.3964,
+      "step": 1636
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5481194564131512e-05,
+      "loss": 1.4094,
+      "step": 1637
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5475632234925505e-05,
+      "loss": 1.3891,
+      "step": 1638
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.547006748507855e-05,
+      "loss": 1.3647,
+      "step": 1639
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5464500317050687e-05,
+      "loss": 1.4175,
+      "step": 1640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.545893073330302e-05,
+      "loss": 1.3859,
+      "step": 1641
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5453358736297727e-05,
+      "loss": 1.3643,
+      "step": 1642
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5447784328498053e-05,
+      "loss": 1.4284,
+      "step": 1643
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5442207512368294e-05,
+      "loss": 1.2712,
+      "step": 1644
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5436628290373835e-05,
+      "loss": 1.36,
+      "step": 1645
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5431046664981102e-05,
+      "loss": 1.3722,
+      "step": 1646
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5425462638657597e-05,
+      "loss": 1.3734,
+      "step": 1647
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5419876213871878e-05,
+      "loss": 1.4223,
+      "step": 1648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5414287393093564e-05,
+      "loss": 1.4618,
+      "step": 1649
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5408696178793333e-05,
+      "loss": 1.3498,
+      "step": 1650
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5403102573442922e-05,
+      "loss": 1.4227,
+      "step": 1651
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.539750657951513e-05,
+      "loss": 1.365,
+      "step": 1652
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5391908199483804e-05,
+      "loss": 1.3476,
+      "step": 1653
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5386307435823847e-05,
+      "loss": 1.3873,
+      "step": 1654
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5380704291011214e-05,
+      "loss": 1.3632,
+      "step": 1655
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5375098767522926e-05,
+      "loss": 1.283,
+      "step": 1656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5369490867837037e-05,
+      "loss": 1.3846,
+      "step": 1657
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5363880594432665e-05,
+      "loss": 1.3625,
+      "step": 1658
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5358267949789968e-05,
+      "loss": 1.3362,
+      "step": 1659
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.535265293639016e-05,
+      "loss": 1.3332,
+      "step": 1660
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5347035556715498e-05,
+      "loss": 1.3296,
+      "step": 1661
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.534141581324929e-05,
+      "loss": 1.3637,
+      "step": 1662
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5335793708475878e-05,
+      "loss": 1.4352,
+      "step": 1663
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5330169244880657e-05,
+      "loss": 1.3511,
+      "step": 1664
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5324542424950067e-05,
+      "loss": 1.36,
+      "step": 1665
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.531891325117158e-05,
+      "loss": 1.3169,
+      "step": 1666
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5313281726033714e-05,
+      "loss": 1.3626,
+      "step": 1667
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.530764785202603e-05,
+      "loss": 1.3044,
+      "step": 1668
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.530201163163912e-05,
+      "loss": 1.3733,
+      "step": 1669
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.529637306736462e-05,
+      "loss": 1.319,
+      "step": 1670
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5290732161695198e-05,
+      "loss": 1.2932,
+      "step": 1671
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5285088917124555e-05,
+      "loss": 1.3487,
+      "step": 1672
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5279443336147437e-05,
+      "loss": 1.3599,
+      "step": 1673
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5273795421259608e-05,
+      "loss": 1.3247,
+      "step": 1674
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.526814517495787e-05,
+      "loss": 1.3159,
+      "step": 1675
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5262492599740065e-05,
+      "loss": 1.3784,
+      "step": 1676
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5256837698105047e-05,
+      "loss": 1.3529,
+      "step": 1677
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5251180472552713e-05,
+      "loss": 1.3276,
+      "step": 1678
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5245520925583984e-05,
+      "loss": 1.3906,
+      "step": 1679
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5239859059700794e-05,
+      "loss": 1.3136,
+      "step": 1680
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5234194877406122e-05,
+      "loss": 1.3991,
+      "step": 1681
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5228528381203962e-05,
+      "loss": 1.3129,
+      "step": 1682
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5222859573599329e-05,
+      "loss": 1.3421,
+      "step": 1683
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5217188457098261e-05,
+      "loss": 1.3224,
+      "step": 1684
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5211515034207823e-05,
+      "loss": 1.3974,
+      "step": 1685
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5205839307436088e-05,
+      "loss": 1.3844,
+      "step": 1686
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5200161279292154e-05,
+      "loss": 1.3794,
+      "step": 1687
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5194480952286144e-05,
+      "loss": 1.4551,
+      "step": 1688
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5188798328929179e-05,
+      "loss": 1.342,
+      "step": 1689
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5183113411733414e-05,
+      "loss": 1.3635,
+      "step": 1690
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5177426203212004e-05,
+      "loss": 1.3299,
+      "step": 1691
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5171736705879127e-05,
+      "loss": 1.2847,
+      "step": 1692
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5166044922249965e-05,
+      "loss": 1.3815,
+      "step": 1693
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5160350854840715e-05,
+      "loss": 1.3701,
+      "step": 1694
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5154654506168585e-05,
+      "loss": 1.3985,
+      "step": 1695
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5148955878751785e-05,
+      "loss": 1.3572,
+      "step": 1696
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5143254975109538e-05,
+      "loss": 1.3507,
+      "step": 1697
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5137551797762076e-05,
+      "loss": 1.3524,
+      "step": 1698
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5131846349230623e-05,
+      "loss": 1.3782,
+      "step": 1699
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5126138632037423e-05,
+      "loss": 1.308,
+      "step": 1700
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5120428648705716e-05,
+      "loss": 1.382,
+      "step": 1701
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5114716401759741e-05,
+      "loss": 1.3852,
+      "step": 1702
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5109001893724742e-05,
+      "loss": 1.3953,
+      "step": 1703
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5103285127126962e-05,
+      "loss": 1.4323,
+      "step": 1704
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5097566104493641e-05,
+      "loss": 1.3232,
+      "step": 1705
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5091844828353015e-05,
+      "loss": 1.4285,
+      "step": 1706
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5086121301234318e-05,
+      "loss": 1.4527,
+      "step": 1707
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.508039552566778e-05,
+      "loss": 1.3035,
+      "step": 1708
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5074667504184625e-05,
+      "loss": 1.3938,
+      "step": 1709
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5068937239317067e-05,
+      "loss": 1.366,
+      "step": 1710
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5063204733598316e-05,
+      "loss": 1.3656,
+      "step": 1711
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5057469989562568e-05,
+      "loss": 1.3735,
+      "step": 1712
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5051733009745013e-05,
+      "loss": 1.3274,
+      "step": 1713
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5045993796681823e-05,
+      "loss": 1.3584,
+      "step": 1714
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5040252352910168e-05,
+      "loss": 1.3811,
+      "step": 1715
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5034508680968194e-05,
+      "loss": 1.4534,
+      "step": 1716
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5028762783395035e-05,
+      "loss": 1.3637,
+      "step": 1717
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5023014662730809e-05,
+      "loss": 1.4555,
+      "step": 1718
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.501726432151662e-05,
+      "loss": 1.3977,
+      "step": 1719
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.501151176229455e-05,
+      "loss": 1.256,
+      "step": 1720
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5005756987607664e-05,
+      "loss": 1.4052,
+      "step": 1721
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 1.3667,
+      "step": 1722
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4994240802016586e-05,
+      "loss": 1.3217,
+      "step": 1723
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4988479396203413e-05,
+      "loss": 1.3986,
+      "step": 1724
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.498271578510746e-05,
+      "loss": 1.3147,
+      "step": 1725
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4976949971276675e-05,
+      "loss": 1.3474,
+      "step": 1726
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4971181957259982e-05,
+      "loss": 1.4062,
+      "step": 1727
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4965411745607275e-05,
+      "loss": 1.3813,
+      "step": 1728
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4959639338869423e-05,
+      "loss": 1.3094,
+      "step": 1729
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.495386473959826e-05,
+      "loss": 1.3805,
+      "step": 1730
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4948087950346599e-05,
+      "loss": 1.4291,
+      "step": 1731
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.494230897366821e-05,
+      "loss": 1.3759,
+      "step": 1732
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4936527812117833e-05,
+      "loss": 1.4024,
+      "step": 1733
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4930744468251183e-05,
+      "loss": 1.3901,
+      "step": 1734
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.492495894462493e-05,
+      "loss": 1.3424,
+      "step": 1735
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4919171243796706e-05,
+      "loss": 1.3358,
+      "step": 1736
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4913381368325115e-05,
+      "loss": 1.3662,
+      "step": 1737
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4907589320769715e-05,
+      "loss": 1.3793,
+      "step": 1738
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4901795103691025e-05,
+      "loss": 1.3392,
+      "step": 1739
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4895998719650526e-05,
+      "loss": 1.3841,
+      "step": 1740
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4890200171210656e-05,
+      "loss": 1.3797,
+      "step": 1741
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4884399460934806e-05,
+      "loss": 1.3822,
+      "step": 1742
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4878596591387329e-05,
+      "loss": 1.3116,
+      "step": 1743
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4872791565133528e-05,
+      "loss": 1.3426,
+      "step": 1744
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4866984384739662e-05,
+      "loss": 1.3677,
+      "step": 1745
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.486117505277294e-05,
+      "loss": 1.2893,
+      "step": 1746
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4855363571801523e-05,
+      "loss": 1.3863,
+      "step": 1747
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4849549944394524e-05,
+      "loss": 1.4951,
+      "step": 1748
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4843734173122002e-05,
+      "loss": 1.3534,
+      "step": 1749
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4837916260554966e-05,
+      "loss": 1.3691,
+      "step": 1750
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4832096209265367e-05,
+      "loss": 1.3852,
+      "step": 1751
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.482627402182611e-05,
+      "loss": 1.3122,
+      "step": 1752
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4820449700811037e-05,
+      "loss": 1.2959,
+      "step": 1753
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4814623248794935e-05,
+      "loss": 1.3133,
+      "step": 1754
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4808794668353539e-05,
+      "loss": 1.3875,
+      "step": 1755
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.480296396206351e-05,
+      "loss": 1.3661,
+      "step": 1756
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4797131132502464e-05,
+      "loss": 1.3736,
+      "step": 1757
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4791296182248952e-05,
+      "loss": 1.3676,
+      "step": 1758
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4785459113882454e-05,
+      "loss": 1.3657,
+      "step": 1759
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4779619929983397e-05,
+      "loss": 1.3974,
+      "step": 1760
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4773778633133142e-05,
+      "loss": 1.3788,
+      "step": 1761
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4767935225913976e-05,
+      "loss": 1.3083,
+      "step": 1762
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4762089710909129e-05,
+      "loss": 1.3135,
+      "step": 1763
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4756242090702756e-05,
+      "loss": 1.4572,
+      "step": 1764
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.475039236787994e-05,
+      "loss": 1.3386,
+      "step": 1765
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4744540545026703e-05,
+      "loss": 1.3948,
+      "step": 1766
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4738686624729987e-05,
+      "loss": 1.4017,
+      "step": 1767
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4732830609577672e-05,
+      "loss": 1.3661,
+      "step": 1768
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4726972502158548e-05,
+      "loss": 1.3354,
+      "step": 1769
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4721112305062341e-05,
+      "loss": 1.3828,
+      "step": 1770
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4715250020879705e-05,
+      "loss": 1.3715,
+      "step": 1771
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4709385652202204e-05,
+      "loss": 1.3772,
+      "step": 1772
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4703519201622327e-05,
+      "loss": 1.3242,
+      "step": 1773
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4697650671733495e-05,
+      "loss": 1.3177,
+      "step": 1774
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.469178006513003e-05,
+      "loss": 1.3131,
+      "step": 1775
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4685907384407187e-05,
+      "loss": 1.3478,
+      "step": 1776
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.468003263216113e-05,
+      "loss": 1.3286,
+      "step": 1777
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4674155810988944e-05,
+      "loss": 1.3608,
+      "step": 1778
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.466827692348862e-05,
+      "loss": 1.3498,
+      "step": 1779
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4662395972259074e-05,
+      "loss": 1.3871,
+      "step": 1780
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4656512959900124e-05,
+      "loss": 1.3897,
+      "step": 1781
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4650627889012507e-05,
+      "loss": 1.4108,
+      "step": 1782
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4644740762197864e-05,
+      "loss": 1.4617,
+      "step": 1783
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4638851582058746e-05,
+      "loss": 1.4215,
+      "step": 1784
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.463296035119862e-05,
+      "loss": 1.3501,
+      "step": 1785
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4627067072221845e-05,
+      "loss": 1.4415,
+      "step": 1786
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4621171747733698e-05,
+      "loss": 1.3844,
+      "step": 1787
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4615274380340353e-05,
+      "loss": 1.3362,
+      "step": 1788
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.460937497264889e-05,
+      "loss": 1.4038,
+      "step": 1789
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4603473527267294e-05,
+      "loss": 1.3401,
+      "step": 1790
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.459757004680444e-05,
+      "loss": 1.3579,
+      "step": 1791
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4591664533870118e-05,
+      "loss": 1.3201,
+      "step": 1792
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4585756991075004e-05,
+      "loss": 1.3904,
+      "step": 1793
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4579847421030677e-05,
+      "loss": 1.3051,
+      "step": 1794
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.457393582634961e-05,
+      "loss": 1.329,
+      "step": 1795
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4568022209645176e-05,
+      "loss": 1.327,
+      "step": 1796
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4562106573531632e-05,
+      "loss": 1.3287,
+      "step": 1797
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4556188920624137e-05,
+      "loss": 1.4488,
+      "step": 1798
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4550269253538739e-05,
+      "loss": 1.3714,
+      "step": 1799
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4544347574892375e-05,
+      "loss": 1.3297,
+      "step": 1800
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.453842388730287e-05,
+      "loss": 1.34,
+      "step": 1801
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4532498193388941e-05,
+      "loss": 1.3628,
+      "step": 1802
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4526570495770194e-05,
+      "loss": 1.2566,
+      "step": 1803
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4520640797067108e-05,
+      "loss": 1.3341,
+      "step": 1804
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4514709099901061e-05,
+      "loss": 1.3362,
+      "step": 1805
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4508775406894308e-05,
+      "loss": 1.3369,
+      "step": 1806
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4502839720669988e-05,
+      "loss": 1.3925,
+      "step": 1807
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4496902043852121e-05,
+      "loss": 1.4279,
+      "step": 1808
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4490962379065603e-05,
+      "loss": 1.3719,
+      "step": 1809
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4485020728936215e-05,
+      "loss": 1.3723,
+      "step": 1810
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4479077096090613e-05,
+      "loss": 1.3038,
+      "step": 1811
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4473131483156326e-05,
+      "loss": 1.3722,
+      "step": 1812
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4467183892761769e-05,
+      "loss": 1.3883,
+      "step": 1813
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4461234327536215e-05,
+      "loss": 1.3701,
+      "step": 1814
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4455282790109825e-05,
+      "loss": 1.3552,
+      "step": 1815
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4449329283113622e-05,
+      "loss": 1.3957,
+      "step": 1816
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4443373809179508e-05,
+      "loss": 1.4116,
+      "step": 1817
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4437416370940245e-05,
+      "loss": 1.435,
+      "step": 1818
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4431456971029469e-05,
+      "loss": 1.4271,
+      "step": 1819
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.442549561208169e-05,
+      "loss": 1.4286,
+      "step": 1820
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4419532296732271e-05,
+      "loss": 1.3263,
+      "step": 1821
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4413567027617442e-05,
+      "loss": 1.2997,
+      "step": 1822
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4407599807374304e-05,
+      "loss": 1.389,
+      "step": 1823
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4401630638640818e-05,
+      "loss": 1.3693,
+      "step": 1824
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4395659524055803e-05,
+      "loss": 1.3392,
+      "step": 1825
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4389686466258943e-05,
+      "loss": 1.4015,
+      "step": 1826
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4383711467890776e-05,
+      "loss": 1.4152,
+      "step": 1827
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.43777345315927e-05,
+      "loss": 1.3976,
+      "step": 1828
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4371755660006972e-05,
+      "loss": 1.4553,
+      "step": 1829
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4365774855776702e-05,
+      "loss": 1.3464,
+      "step": 1830
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4359792121545854e-05,
+      "loss": 1.4396,
+      "step": 1831
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4353807459959243e-05,
+      "loss": 1.3675,
+      "step": 1832
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4347820873662548e-05,
+      "loss": 1.462,
+      "step": 1833
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4341832365302282e-05,
+      "loss": 1.3518,
+      "step": 1834
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4335841937525819e-05,
+      "loss": 1.4108,
+      "step": 1835
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4329849592981378e-05,
+      "loss": 1.4187,
+      "step": 1836
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4323855334318026e-05,
+      "loss": 1.3683,
+      "step": 1837
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4317859164185677e-05,
+      "loss": 1.326,
+      "step": 1838
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4311861085235087e-05,
+      "loss": 1.3532,
+      "step": 1839
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4305861100117854e-05,
+      "loss": 1.3663,
+      "step": 1840
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4299859211486429e-05,
+      "loss": 1.3873,
+      "step": 1841
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4293855421994094e-05,
+      "loss": 1.4107,
+      "step": 1842
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4287849734294974e-05,
+      "loss": 1.4154,
+      "step": 1843
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.428184215104404e-05,
+      "loss": 1.3864,
+      "step": 1844
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4275832674897088e-05,
+      "loss": 1.3545,
+      "step": 1845
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4269821308510763e-05,
+      "loss": 1.5039,
+      "step": 1846
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4263808054542541e-05,
+      "loss": 1.3866,
+      "step": 1847
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4257792915650728e-05,
+      "loss": 1.3619,
+      "step": 1848
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4251775894494471e-05,
+      "loss": 1.3631,
+      "step": 1849
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4245756993733744e-05,
+      "loss": 1.3591,
+      "step": 1850
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4239736216029353e-05,
+      "loss": 1.3994,
+      "step": 1851
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4233713564042937e-05,
+      "loss": 1.3065,
+      "step": 1852
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.422768904043696e-05,
+      "loss": 1.3755,
+      "step": 1853
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4221662647874713e-05,
+      "loss": 1.299,
+      "step": 1854
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4215634389020314e-05,
+      "loss": 1.4542,
+      "step": 1855
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4209604266538706e-05,
+      "loss": 1.4048,
+      "step": 1856
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4203572283095657e-05,
+      "loss": 1.3931,
+      "step": 1857
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4197538441357757e-05,
+      "loss": 1.2934,
+      "step": 1858
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4191502743992418e-05,
+      "loss": 1.349,
+      "step": 1859
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4185465193667874e-05,
+      "loss": 1.3383,
+      "step": 1860
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4179425793053172e-05,
+      "loss": 1.3738,
+      "step": 1861
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.417338454481818e-05,
+      "loss": 1.3858,
+      "step": 1862
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.416734145163359e-05,
+      "loss": 1.3324,
+      "step": 1863
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4161296516170897e-05,
+      "loss": 1.4123,
+      "step": 1864
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.415524974110242e-05,
+      "loss": 1.3717,
+      "step": 1865
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4149201129101287e-05,
+      "loss": 1.303,
+      "step": 1866
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.414315068284144e-05,
+      "loss": 1.3844,
+      "step": 1867
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.413709840499763e-05,
+      "loss": 1.4357,
+      "step": 1868
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.413104429824542e-05,
+      "loss": 1.4244,
+      "step": 1869
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4124988365261176e-05,
+      "loss": 1.3824,
+      "step": 1870
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4118930608722084e-05,
+      "loss": 1.3893,
+      "step": 1871
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4112871031306118e-05,
+      "loss": 1.408,
+      "step": 1872
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4106809635692072e-05,
+      "loss": 1.3538,
+      "step": 1873
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4100746424559537e-05,
+      "loss": 1.3959,
+      "step": 1874
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4094681400588908e-05,
+      "loss": 1.3683,
+      "step": 1875
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.408861456646138e-05,
+      "loss": 1.3548,
+      "step": 1876
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4082545924858955e-05,
+      "loss": 1.4249,
+      "step": 1877
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.407647547846442e-05,
+      "loss": 1.3832,
+      "step": 1878
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4070403229961376e-05,
+      "loss": 1.2899,
+      "step": 1879
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.406432918203421e-05,
+      "loss": 1.3324,
+      "step": 1880
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.405825333736811e-05,
+      "loss": 1.4618,
+      "step": 1881
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4052175698649054e-05,
+      "loss": 1.4393,
+      "step": 1882
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4046096268563814e-05,
+      "loss": 1.3602,
+      "step": 1883
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4040015049799952e-05,
+      "loss": 1.3374,
+      "step": 1884
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4033932045045834e-05,
+      "loss": 1.3718,
+      "step": 1885
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4027847256990596e-05,
+      "loss": 1.3439,
+      "step": 1886
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4021760688324175e-05,
+      "loss": 1.3865,
+      "step": 1887
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4015672341737295e-05,
+      "loss": 1.3896,
+      "step": 1888
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4009582219921455e-05,
+      "loss": 1.3999,
+      "step": 1889
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4003490325568953e-05,
+      "loss": 1.3957,
+      "step": 1890
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3997396661372858e-05,
+      "loss": 1.3744,
+      "step": 1891
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3991301230027032e-05,
+      "loss": 1.4149,
+      "step": 1892
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3985204034226117e-05,
+      "loss": 1.3966,
+      "step": 1893
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3979105076665525e-05,
+      "loss": 1.3892,
+      "step": 1894
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3973004360041458e-05,
+      "loss": 1.3858,
+      "step": 1895
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.396690188705089e-05,
+      "loss": 1.368,
+      "step": 1896
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.396079766039157e-05,
+      "loss": 1.3154,
+      "step": 1897
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3954691682762028e-05,
+      "loss": 1.3414,
+      "step": 1898
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3948583956861566e-05,
+      "loss": 1.3798,
+      "step": 1899
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3942474485390251e-05,
+      "loss": 1.4666,
+      "step": 1900
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3936363271048939e-05,
+      "loss": 1.3715,
+      "step": 1901
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3930250316539237e-05,
+      "loss": 1.4104,
+      "step": 1902
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3924135624563535e-05,
+      "loss": 1.383,
+      "step": 1903
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3918019197824985e-05,
+      "loss": 1.3838,
+      "step": 1904
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3911901039027508e-05,
+      "loss": 1.3895,
+      "step": 1905
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.390578115087579e-05,
+      "loss": 1.4634,
+      "step": 1906
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.389965953607528e-05,
+      "loss": 1.3574,
+      "step": 1907
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3893536197332194e-05,
+      "loss": 1.4299,
+      "step": 1908
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3887411137353506e-05,
+      "loss": 1.3613,
+      "step": 1909
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3881284358846951e-05,
+      "loss": 1.3579,
+      "step": 1910
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3875155864521031e-05,
+      "loss": 1.3722,
+      "step": 1911
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3869025657084996e-05,
+      "loss": 1.3625,
+      "step": 1912
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3862893739248863e-05,
+      "loss": 1.3583,
+      "step": 1913
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3856760113723397e-05,
+      "loss": 1.3746,
+      "step": 1914
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.385062478322012e-05,
+      "loss": 1.3469,
+      "step": 1915
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3844487750451314e-05,
+      "loss": 1.3736,
+      "step": 1916
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3838349018130007e-05,
+      "loss": 1.3316,
+      "step": 1917
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3832208588969975e-05,
+      "loss": 1.381,
+      "step": 1918
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3826066465685756e-05,
+      "loss": 1.4175,
+      "step": 1919
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3819922650992625e-05,
+      "loss": 1.3278,
+      "step": 1920
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3813777147606614e-05,
+      "loss": 1.3446,
+      "step": 1921
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3807629958244498e-05,
+      "loss": 1.3866,
+      "step": 1922
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3801481085623792e-05,
+      "loss": 1.3862,
+      "step": 1923
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3795330532462761e-05,
+      "loss": 1.3427,
+      "step": 1924
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3789178301480415e-05,
+      "loss": 1.2975,
+      "step": 1925
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3783024395396498e-05,
+      "loss": 1.3478,
+      "step": 1926
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3776868816931501e-05,
+      "loss": 1.3304,
+      "step": 1927
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3770711568806655e-05,
+      "loss": 1.382,
+      "step": 1928
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.376455265374392e-05,
+      "loss": 1.3088,
+      "step": 1929
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3758392074466007e-05,
+      "loss": 1.4172,
+      "step": 1930
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.375222983369635e-05,
+      "loss": 1.2448,
+      "step": 1931
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3746065934159123e-05,
+      "loss": 1.4672,
+      "step": 1932
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3739900378579231e-05,
+      "loss": 1.3477,
+      "step": 1933
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.373373316968232e-05,
+      "loss": 1.3037,
+      "step": 1934
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3727564310194752e-05,
+      "loss": 1.3063,
+      "step": 1935
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.372139380284363e-05,
+      "loss": 1.3377,
+      "step": 1936
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.371522165035678e-05,
+      "loss": 1.4333,
+      "step": 1937
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3709047855462765e-05,
+      "loss": 1.3547,
+      "step": 1938
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3702872420890853e-05,
+      "loss": 1.3646,
+      "step": 1939
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3696695349371064e-05,
+      "loss": 1.4609,
+      "step": 1940
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3690516643634115e-05,
+      "loss": 1.3503,
+      "step": 1941
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3684336306411467e-05,
+      "loss": 1.3389,
+      "step": 1942
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.367815434043529e-05,
+      "loss": 1.3387,
+      "step": 1943
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3671970748438482e-05,
+      "loss": 1.3485,
+      "step": 1944
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3665785533154647e-05,
+      "loss": 1.4308,
+      "step": 1945
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3659598697318122e-05,
+      "loss": 1.4431,
+      "step": 1946
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3653410243663953e-05,
+      "loss": 1.4091,
+      "step": 1947
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3647220174927898e-05,
+      "loss": 1.408,
+      "step": 1948
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3641028493846436e-05,
+      "loss": 1.3972,
+      "step": 1949
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.363483520315675e-05,
+      "loss": 1.3673,
+      "step": 1950
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3628640305596743e-05,
+      "loss": 1.3771,
+      "step": 1951
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3622443803905028e-05,
+      "loss": 1.3601,
+      "step": 1952
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3616245700820922e-05,
+      "loss": 1.3455,
+      "step": 1953
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3610045999084457e-05,
+      "loss": 1.3347,
+      "step": 1954
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.360384470143636e-05,
+      "loss": 1.4001,
+      "step": 1955
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3597641810618071e-05,
+      "loss": 1.3577,
+      "step": 1956
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3591437329371738e-05,
+      "loss": 1.3271,
+      "step": 1957
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3585231260440207e-05,
+      "loss": 1.3417,
+      "step": 1958
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3579023606567024e-05,
+      "loss": 1.3588,
+      "step": 1959
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3572814370496441e-05,
+      "loss": 1.284,
+      "step": 1960
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3566603554973408e-05,
+      "loss": 1.3263,
+      "step": 1961
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.356039116274357e-05,
+      "loss": 1.2597,
+      "step": 1962
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3554177196553274e-05,
+      "loss": 1.3536,
+      "step": 1963
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3547961659149552e-05,
+      "loss": 1.3328,
+      "step": 1964
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3541744553280149e-05,
+      "loss": 1.3627,
+      "step": 1965
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3535525881693487e-05,
+      "loss": 1.3939,
+      "step": 1966
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3529305647138689e-05,
+      "loss": 1.3653,
+      "step": 1967
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3523083852365566e-05,
+      "loss": 1.3438,
+      "step": 1968
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3516860500124618e-05,
+      "loss": 1.3676,
+      "step": 1969
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3510635593167032e-05,
+      "loss": 1.2821,
+      "step": 1970
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3504409134244687e-05,
+      "loss": 1.3848,
+      "step": 1971
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.349818112611015e-05,
+      "loss": 1.3519,
+      "step": 1972
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.349195157151666e-05,
+      "loss": 1.2908,
+      "step": 1973
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3485720473218153e-05,
+      "loss": 1.3022,
+      "step": 1974
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3479487833969244e-05,
+      "loss": 1.3167,
+      "step": 1975
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3473253656525225e-05,
+      "loss": 1.4222,
+      "step": 1976
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3467017943642074e-05,
+      "loss": 1.3915,
+      "step": 1977
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.346078069807644e-05,
+      "loss": 1.4502,
+      "step": 1978
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3454541922585657e-05,
+      "loss": 1.3831,
+      "step": 1979
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3448301619927735e-05,
+      "loss": 1.3595,
+      "step": 1980
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3442059792861356e-05,
+      "loss": 1.3623,
+      "step": 1981
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3435816444145871e-05,
+      "loss": 1.3567,
+      "step": 1982
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3429571576541315e-05,
+      "loss": 1.3912,
+      "step": 1983
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.342332519280839e-05,
+      "loss": 1.3564,
+      "step": 1984
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3417077295708461e-05,
+      "loss": 1.3255,
+      "step": 1985
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3410827888003573e-05,
+      "loss": 1.3955,
+      "step": 1986
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.340457697245643e-05,
+      "loss": 1.4612,
+      "step": 1987
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3398324551830416e-05,
+      "loss": 1.3441,
+      "step": 1988
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3392070628889556e-05,
+      "loss": 1.3481,
+      "step": 1989
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3385815206398562e-05,
+      "loss": 1.385,
+      "step": 1990
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3379558287122802e-05,
+      "loss": 1.3691,
+      "step": 1991
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3373299873828303e-05,
+      "loss": 1.3529,
+      "step": 1992
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3367039969281753e-05,
+      "loss": 1.4029,
+      "step": 1993
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.33607785762505e-05,
+      "loss": 1.3449,
+      "step": 1994
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3354515697502552e-05,
+      "loss": 1.3751,
+      "step": 1995
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3348251335806571e-05,
+      "loss": 1.2972,
+      "step": 1996
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3341985493931877e-05,
+      "loss": 1.3334,
+      "step": 1997
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3335718174648442e-05,
+      "loss": 1.4426,
+      "step": 1998
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.332944938072689e-05,
+      "loss": 1.3627,
+      "step": 1999
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3323179114938502e-05,
+      "loss": 1.3739,
+      "step": 2000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3316907380055208e-05,
+      "loss": 1.3627,
+      "step": 2001
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3310634178849583e-05,
+      "loss": 1.3456,
+      "step": 2002
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3304359514094854e-05,
+      "loss": 1.3394,
+      "step": 2003
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3298083388564893e-05,
+      "loss": 1.3128,
+      "step": 2004
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3291805805034224e-05,
+      "loss": 1.368,
+      "step": 2005
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3285526766278005e-05,
+      "loss": 1.3273,
+      "step": 2006
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3279246275072046e-05,
+      "loss": 1.3283,
+      "step": 2007
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3272964334192797e-05,
+      "loss": 1.3939,
+      "step": 2008
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3266680946417346e-05,
+      "loss": 1.3258,
+      "step": 2009
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.326039611452342e-05,
+      "loss": 1.3849,
+      "step": 2010
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3254109841289392e-05,
+      "loss": 1.3545,
+      "step": 2011
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3247822129494265e-05,
+      "loss": 1.3496,
+      "step": 2012
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3241532981917683e-05,
+      "loss": 1.324,
+      "step": 2013
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3235242401339918e-05,
+      "loss": 1.3075,
+      "step": 2014
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3228950390541882e-05,
+      "loss": 1.3827,
+      "step": 2015
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3222656952305113e-05,
+      "loss": 1.3214,
+      "step": 2016
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3216362089411785e-05,
+      "loss": 1.36,
+      "step": 2017
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3210065804644705e-05,
+      "loss": 1.3424,
+      "step": 2018
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3203768100787298e-05,
+      "loss": 1.3804,
+      "step": 2019
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3197468980623627e-05,
+      "loss": 1.4353,
+      "step": 2020
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3191168446938372e-05,
+      "loss": 1.3371,
+      "step": 2021
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3184866502516846e-05,
+      "loss": 1.416,
+      "step": 2022
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.317856315014498e-05,
+      "loss": 1.3369,
+      "step": 2023
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.317225839260933e-05,
+      "loss": 1.325,
+      "step": 2024
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3165952232697075e-05,
+      "loss": 1.3716,
+      "step": 2025
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3159644673196007e-05,
+      "loss": 1.3585,
+      "step": 2026
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3153335716894543e-05,
+      "loss": 1.3807,
+      "step": 2027
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.314702536658172e-05,
+      "loss": 1.3487,
+      "step": 2028
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.314071362504718e-05,
+      "loss": 1.3547,
+      "step": 2029
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3134400495081197e-05,
+      "loss": 1.3432,
+      "step": 2030
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3128085979474639e-05,
+      "loss": 1.4229,
+      "step": 2031
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3121770081018998e-05,
+      "loss": 1.2984,
+      "step": 2032
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.311545280250638e-05,
+      "loss": 1.3871,
+      "step": 2033
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3109134146729494e-05,
+      "loss": 1.4368,
+      "step": 2034
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3102814116481661e-05,
+      "loss": 1.4046,
+      "step": 2035
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.309649271455681e-05,
+      "loss": 1.4112,
+      "step": 2036
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3090169943749475e-05,
+      "loss": 1.426,
+      "step": 2037
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3083845806854798e-05,
+      "loss": 1.3848,
+      "step": 2038
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3077520306668519e-05,
+      "loss": 1.3304,
+      "step": 2039
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3071193445986988e-05,
+      "loss": 1.3417,
+      "step": 2040
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.306486522760715e-05,
+      "loss": 1.3971,
+      "step": 2041
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3058535654326554e-05,
+      "loss": 1.3197,
+      "step": 2042
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.305220472894335e-05,
+      "loss": 1.3776,
+      "step": 2043
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3045872454256278e-05,
+      "loss": 1.3951,
+      "step": 2044
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3039538833064685e-05,
+      "loss": 1.3683,
+      "step": 2045
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.30332038681685e-05,
+      "loss": 1.3317,
+      "step": 2046
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3026867562368262e-05,
+      "loss": 1.335,
+      "step": 2047
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3020529918465088e-05,
+      "loss": 1.3221,
+      "step": 2048
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3014190939260694e-05,
+      "loss": 1.3992,
+      "step": 2049
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3007850627557386e-05,
+      "loss": 1.3059,
+      "step": 2050
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3001508986158057e-05,
+      "loss": 1.3562,
+      "step": 2051
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2995166017866194e-05,
+      "loss": 1.4065,
+      "step": 2052
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2988821725485861e-05,
+      "loss": 1.3131,
+      "step": 2053
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2982476111821711e-05,
+      "loss": 1.4277,
+      "step": 2054
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2976129179678987e-05,
+      "loss": 1.3234,
+      "step": 2055
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2969780931863506e-05,
+      "loss": 1.4212,
+      "step": 2056
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2963431371181672e-05,
+      "loss": 1.3548,
+      "step": 2057
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2957080500440469e-05,
+      "loss": 1.321,
+      "step": 2058
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2950728322447457e-05,
+      "loss": 1.3613,
+      "step": 2059
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2944374840010776e-05,
+      "loss": 1.5052,
+      "step": 2060
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2938020055939145e-05,
+      "loss": 1.3542,
+      "step": 2061
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2931663973041855e-05,
+      "loss": 1.3867,
+      "step": 2062
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2925306594128773e-05,
+      "loss": 1.2969,
+      "step": 2063
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2918947922010336e-05,
+      "loss": 1.3311,
+      "step": 2064
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2912587959497556e-05,
+      "loss": 1.3494,
+      "step": 2065
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2906226709402016e-05,
+      "loss": 1.357,
+      "step": 2066
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2899864174535863e-05,
+      "loss": 1.3103,
+      "step": 2067
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2893500357711818e-05,
+      "loss": 1.3927,
+      "step": 2068
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2887135261743167e-05,
+      "loss": 1.4251,
+      "step": 2069
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2880768889443763e-05,
+      "loss": 1.3677,
+      "step": 2070
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2874401243628012e-05,
+      "loss": 1.4259,
+      "step": 2071
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2868032327110904e-05,
+      "loss": 1.381,
+      "step": 2072
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2861662142707969e-05,
+      "loss": 1.3645,
+      "step": 2073
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2855290693235315e-05,
+      "loss": 1.4171,
+      "step": 2074
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2848917981509592e-05,
+      "loss": 1.3741,
+      "step": 2075
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2842544010348025e-05,
+      "loss": 1.3797,
+      "step": 2076
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2836168782568385e-05,
+      "loss": 1.3536,
+      "step": 2077
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2829792300989006e-05,
+      "loss": 1.3476,
+      "step": 2078
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2823414568428767e-05,
+      "loss": 1.4143,
+      "step": 2079
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2817035587707107e-05,
+      "loss": 1.3251,
+      "step": 2080
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2810655361644016e-05,
+      "loss": 1.3792,
+      "step": 2081
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2804273893060028e-05,
+      "loss": 1.3538,
+      "step": 2082
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2797891184776238e-05,
+      "loss": 1.3557,
+      "step": 2083
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2791507239614283e-05,
+      "loss": 1.3434,
+      "step": 2084
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2785122060396345e-05,
+      "loss": 1.4423,
+      "step": 2085
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.277873564994515e-05,
+      "loss": 1.3612,
+      "step": 2086
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2772348011083973e-05,
+      "loss": 1.3944,
+      "step": 2087
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2765959146636633e-05,
+      "loss": 1.4213,
+      "step": 2088
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2759569059427485e-05,
+      "loss": 1.3753,
+      "step": 2089
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.275317775228143e-05,
+      "loss": 1.3584,
+      "step": 2090
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2746785228023904e-05,
+      "loss": 1.3691,
+      "step": 2091
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2740391489480885e-05,
+      "loss": 1.3837,
+      "step": 2092
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2733996539478883e-05,
+      "loss": 1.3894,
+      "step": 2093
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2727600380844951e-05,
+      "loss": 1.3748,
+      "step": 2094
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2721203016406668e-05,
+      "loss": 1.3587,
+      "step": 2095
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2714804448992153e-05,
+      "loss": 1.26,
+      "step": 2096
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2708404681430054e-05,
+      "loss": 1.3169,
+      "step": 2097
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2702003716549542e-05,
+      "loss": 1.3487,
+      "step": 2098
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2695601557180333e-05,
+      "loss": 1.3484,
+      "step": 2099
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2689198206152657e-05,
+      "loss": 1.3473,
+      "step": 2100
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2682793666297281e-05,
+      "loss": 1.3556,
+      "step": 2101
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.267638794044549e-05,
+      "loss": 1.4564,
+      "step": 2102
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2669981031429095e-05,
+      "loss": 1.3252,
+      "step": 2103
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2663572942080436e-05,
+      "loss": 1.4406,
+      "step": 2104
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2657163675232367e-05,
+      "loss": 1.3816,
+      "step": 2105
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2650753233718263e-05,
+      "loss": 1.3787,
+      "step": 2106
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2644341620372025e-05,
+      "loss": 1.3204,
+      "step": 2107
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2637928838028067e-05,
+      "loss": 1.2808,
+      "step": 2108
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2631514889521319e-05,
+      "loss": 1.3445,
+      "step": 2109
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2625099777687228e-05,
+      "loss": 1.326,
+      "step": 2110
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2618683505361759e-05,
+      "loss": 1.4425,
+      "step": 2111
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2612266075381385e-05,
+      "loss": 1.382,
+      "step": 2112
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2605847490583091e-05,
+      "loss": 1.3754,
+      "step": 2113
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2599427753804377e-05,
+      "loss": 1.3799,
+      "step": 2114
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2593006867883244e-05,
+      "loss": 1.3522,
+      "step": 2115
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2586584835658208e-05,
+      "loss": 1.3422,
+      "step": 2116
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2580161659968294e-05,
+      "loss": 1.2993,
+      "step": 2117
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2573737343653026e-05,
+      "loss": 1.385,
+      "step": 2118
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2567311889552432e-05,
+      "loss": 1.375,
+      "step": 2119
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2560885300507047e-05,
+      "loss": 1.4117,
+      "step": 2120
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2554457579357906e-05,
+      "loss": 1.3349,
+      "step": 2121
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2548028728946548e-05,
+      "loss": 1.3444,
+      "step": 2122
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2541598752115006e-05,
+      "loss": 1.4276,
+      "step": 2123
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2535167651705811e-05,
+      "loss": 1.28,
+      "step": 2124
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2528735430561994e-05,
+      "loss": 1.2976,
+      "step": 2125
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.252230209152708e-05,
+      "loss": 1.389,
+      "step": 2126
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2515867637445088e-05,
+      "loss": 1.2987,
+      "step": 2127
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2509432071160527e-05,
+      "loss": 1.3328,
+      "step": 2128
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2502995395518411e-05,
+      "loss": 1.4131,
+      "step": 2129
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2496557613364229e-05,
+      "loss": 1.2912,
+      "step": 2130
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2490118727543961e-05,
+      "loss": 1.2944,
+      "step": 2131
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2483678740904081e-05,
+      "loss": 1.361,
+      "step": 2132
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.247723765629155e-05,
+      "loss": 1.3849,
+      "step": 2133
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2470795476553805e-05,
+      "loss": 1.4604,
+      "step": 2134
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.246435220453878e-05,
+      "loss": 1.3574,
+      "step": 2135
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2457907843094882e-05,
+      "loss": 1.4208,
+      "step": 2136
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2451462395071002e-05,
+      "loss": 1.381,
+      "step": 2137
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2445015863316515e-05,
+      "loss": 1.2837,
+      "step": 2138
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2438568250681273e-05,
+      "loss": 1.3547,
+      "step": 2139
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2432119560015604e-05,
+      "loss": 1.3227,
+      "step": 2140
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2425669794170313e-05,
+      "loss": 1.3946,
+      "step": 2141
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2419218955996677e-05,
+      "loss": 1.3316,
+      "step": 2142
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2412767048346461e-05,
+      "loss": 1.4022,
+      "step": 2143
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.240631407407189e-05,
+      "loss": 1.3821,
+      "step": 2144
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.239986003602566e-05,
+      "loss": 1.3779,
+      "step": 2145
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2393404937060942e-05,
+      "loss": 1.3404,
+      "step": 2146
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.238694878003138e-05,
+      "loss": 1.2961,
+      "step": 2147
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2380491567791074e-05,
+      "loss": 1.3646,
+      "step": 2148
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2374033303194597e-05,
+      "loss": 1.3601,
+      "step": 2149
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.236757398909699e-05,
+      "loss": 1.4371,
+      "step": 2150
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2361113628353757e-05,
+      "loss": 1.3723,
+      "step": 2151
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2354652223820858e-05,
+      "loss": 1.3943,
+      "step": 2152
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2348189778354724e-05,
+      "loss": 1.3818,
+      "step": 2153
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.234172629481224e-05,
+      "loss": 1.3566,
+      "step": 2154
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2335261776050751e-05,
+      "loss": 1.3221,
+      "step": 2155
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.232879622492806e-05,
+      "loss": 1.3859,
+      "step": 2156
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2322329644302426e-05,
+      "loss": 1.2803,
+      "step": 2157
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2315862037032566e-05,
+      "loss": 1.3473,
+      "step": 2158
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2309393405977645e-05,
+      "loss": 1.4352,
+      "step": 2159
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2302923753997286e-05,
+      "loss": 1.3485,
+      "step": 2160
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2296453083951563e-05,
+      "loss": 1.3889,
+      "step": 2161
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2289981398700996e-05,
+      "loss": 1.3513,
+      "step": 2162
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2283508701106559e-05,
+      "loss": 1.331,
+      "step": 2163
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2277034994029668e-05,
+      "loss": 1.3453,
+      "step": 2164
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2270560280332186e-05,
+      "loss": 1.4893,
+      "step": 2165
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2264084562876429e-05,
+      "loss": 1.3136,
+      "step": 2166
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2257607844525145e-05,
+      "loss": 1.3561,
+      "step": 2167
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2251130128141534e-05,
+      "loss": 1.3534,
+      "step": 2168
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.224465141658923e-05,
+      "loss": 1.4202,
+      "step": 2169
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2238171712732316e-05,
+      "loss": 1.3618,
+      "step": 2170
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2231691019435302e-05,
+      "loss": 1.3697,
+      "step": 2171
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2225209339563144e-05,
+      "loss": 1.3942,
+      "step": 2172
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2218726675981233e-05,
+      "loss": 1.3426,
+      "step": 2173
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2212243031555393e-05,
+      "loss": 1.4094,
+      "step": 2174
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2205758409151882e-05,
+      "loss": 1.3613,
+      "step": 2175
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2199272811637389e-05,
+      "loss": 1.3364,
+      "step": 2176
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2192786241879033e-05,
+      "loss": 1.3739,
+      "step": 2177
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2186298702744373e-05,
+      "loss": 1.3113,
+      "step": 2178
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.217981019710138e-05,
+      "loss": 1.3428,
+      "step": 2179
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2173320727818471e-05,
+      "loss": 1.338,
+      "step": 2180
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2166830297764472e-05,
+      "loss": 1.332,
+      "step": 2181
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.216033890980864e-05,
+      "loss": 1.3269,
+      "step": 2182
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.215384656682066e-05,
+      "loss": 1.4126,
+      "step": 2183
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2147353271670634e-05,
+      "loss": 1.3806,
+      "step": 2184
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2140859027229087e-05,
+      "loss": 1.4175,
+      "step": 2185
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2134363836366962e-05,
+      "loss": 1.3389,
+      "step": 2186
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2127867701955622e-05,
+      "loss": 1.3115,
+      "step": 2187
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2121370626866846e-05,
+      "loss": 1.3948,
+      "step": 2188
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2114872613972835e-05,
+      "loss": 1.3625,
+      "step": 2189
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2108373666146192e-05,
+      "loss": 1.3639,
+      "step": 2190
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.210187378625994e-05,
+      "loss": 1.3907,
+      "step": 2191
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2095372977187521e-05,
+      "loss": 1.3212,
+      "step": 2192
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2088871241802776e-05,
+      "loss": 1.3533,
+      "step": 2193
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2082368582979962e-05,
+      "loss": 1.3179,
+      "step": 2194
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.207586500359374e-05,
+      "loss": 1.2662,
+      "step": 2195
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2069360506519188e-05,
+      "loss": 1.3797,
+      "step": 2196
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2062855094631777e-05,
+      "loss": 1.3956,
+      "step": 2197
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2056348770807386e-05,
+      "loss": 1.3756,
+      "step": 2198
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2049841537922307e-05,
+      "loss": 1.315,
+      "step": 2199
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2043333398853216e-05,
+      "loss": 1.4061,
+      "step": 2200
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2036824356477206e-05,
+      "loss": 1.3635,
+      "step": 2201
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2030314413671763e-05,
+      "loss": 1.3619,
+      "step": 2202
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2023803573314767e-05,
+      "loss": 1.3453,
+      "step": 2203
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2017291838284504e-05,
+      "loss": 1.4302,
+      "step": 2204
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2010779211459649e-05,
+      "loss": 1.4125,
+      "step": 2205
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2004265695719268e-05,
+      "loss": 1.3028,
+      "step": 2206
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1997751293942828e-05,
+      "loss": 1.3652,
+      "step": 2207
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1991236009010182e-05,
+      "loss": 1.3582,
+      "step": 2208
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.198471984380158e-05,
+      "loss": 1.3341,
+      "step": 2209
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1978202801197654e-05,
+      "loss": 1.3494,
+      "step": 2210
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1971684884079424e-05,
+      "loss": 1.4087,
+      "step": 2211
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1965166095328302e-05,
+      "loss": 1.3861,
+      "step": 2212
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1958646437826082e-05,
+      "loss": 1.3634,
+      "step": 2213
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1952125914454944e-05,
+      "loss": 1.3877,
+      "step": 2214
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1945604528097447e-05,
+      "loss": 1.3658,
+      "step": 2215
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1939082281636539e-05,
+      "loss": 1.3931,
+      "step": 2216
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1932559177955533e-05,
+      "loss": 1.3835,
+      "step": 2217
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.192603521993814e-05,
+      "loss": 1.3821,
+      "step": 2218
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1919510410468435e-05,
+      "loss": 1.4097,
+      "step": 2219
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1912984752430879e-05,
+      "loss": 1.4061,
+      "step": 2220
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1906458248710297e-05,
+      "loss": 1.4167,
+      "step": 2221
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1899930902191904e-05,
+      "loss": 1.4074,
+      "step": 2222
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1893402715761264e-05,
+      "loss": 1.3717,
+      "step": 2223
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1886873692304335e-05,
+      "loss": 1.456,
+      "step": 2224
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1880343834707433e-05,
+      "loss": 1.334,
+      "step": 2225
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.187381314585725e-05,
+      "loss": 1.3312,
+      "step": 2226
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1867281628640833e-05,
+      "loss": 1.3155,
+      "step": 2227
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1860749285945611e-05,
+      "loss": 1.279,
+      "step": 2228
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.185421612065937e-05,
+      "loss": 1.4008,
+      "step": 2229
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1847682135670256e-05,
+      "loss": 1.2476,
+      "step": 2230
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1841147333866782e-05,
+      "loss": 1.4602,
+      "step": 2231
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1834611718137825e-05,
+      "loss": 1.4123,
+      "step": 2232
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1828075291372616e-05,
+      "loss": 1.4018,
+      "step": 2233
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1821538056460745e-05,
+      "loss": 1.4114,
+      "step": 2234
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1815000016292163e-05,
+      "loss": 1.3774,
+      "step": 2235
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.180846117375718e-05,
+      "loss": 1.3109,
+      "step": 2236
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1801921531746446e-05,
+      "loss": 1.2674,
+      "step": 2237
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1795381093150982e-05,
+      "loss": 1.431,
+      "step": 2238
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.178883986086215e-05,
+      "loss": 1.3823,
+      "step": 2239
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1782297837771668e-05,
+      "loss": 1.3522,
+      "step": 2240
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1775755026771598e-05,
+      "loss": 1.3445,
+      "step": 2241
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1769211430754357e-05,
+      "loss": 1.4098,
+      "step": 2242
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1762667052612706e-05,
+      "loss": 1.3662,
+      "step": 2243
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1756121895239754e-05,
+      "loss": 1.3929,
+      "step": 2244
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1749575961528949e-05,
+      "loss": 1.4047,
+      "step": 2245
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1743029254374086e-05,
+      "loss": 1.3333,
+      "step": 2246
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1736481776669307e-05,
+      "loss": 1.4619,
+      "step": 2247
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.172993353130908e-05,
+      "loss": 1.3033,
+      "step": 2248
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1723384521188227e-05,
+      "loss": 1.376,
+      "step": 2249
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1716834749201905e-05,
+      "loss": 1.3235,
+      "step": 2250
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.17102842182456e-05,
+      "loss": 1.3353,
+      "step": 2251
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1703732931215141e-05,
+      "loss": 1.3324,
+      "step": 2252
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.169718089100669e-05,
+      "loss": 1.3242,
+      "step": 2253
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.169062810051674e-05,
+      "loss": 1.3622,
+      "step": 2254
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1684074562642122e-05,
+      "loss": 1.4053,
+      "step": 2255
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1677520280279986e-05,
+      "loss": 1.3402,
+      "step": 2256
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1670965256327818e-05,
+      "loss": 1.3636,
+      "step": 2257
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.166440949368343e-05,
+      "loss": 1.4132,
+      "step": 2258
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1657852995244968e-05,
+      "loss": 1.3486,
+      "step": 2259
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.165129576391089e-05,
+      "loss": 1.3626,
+      "step": 2260
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1644737802579989e-05,
+      "loss": 1.398,
+      "step": 2261
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1638179114151378e-05,
+      "loss": 1.3698,
+      "step": 2262
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1631619701524485e-05,
+      "loss": 1.4272,
+      "step": 2263
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1625059567599071e-05,
+      "loss": 1.4179,
+      "step": 2264
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1618498715275202e-05,
+      "loss": 1.322,
+      "step": 2265
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.161193714745327e-05,
+      "loss": 1.3825,
+      "step": 2266
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1605374867033978e-05,
+      "loss": 1.3298,
+      "step": 2267
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.159881187691835e-05,
+      "loss": 1.368,
+      "step": 2268
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1592248180007722e-05,
+      "loss": 1.3258,
+      "step": 2269
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1585683779203738e-05,
+      "loss": 1.3528,
+      "step": 2270
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.157911867740836e-05,
+      "loss": 1.3274,
+      "step": 2271
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1572552877523855e-05,
+      "loss": 1.3695,
+      "step": 2272
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1565986382452797e-05,
+      "loss": 1.3114,
+      "step": 2273
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1559419195098075e-05,
+      "loss": 1.3973,
+      "step": 2274
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1552851318362876e-05,
+      "loss": 1.3871,
+      "step": 2275
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1546282755150697e-05,
+      "loss": 1.3417,
+      "step": 2276
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1539713508365336e-05,
+      "loss": 1.3402,
+      "step": 2277
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1533143580910893e-05,
+      "loss": 1.3572,
+      "step": 2278
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1526572975691772e-05,
+      "loss": 1.3443,
+      "step": 2279
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1520001695612675e-05,
+      "loss": 1.3051,
+      "step": 2280
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1513429743578594e-05,
+      "loss": 1.3772,
+      "step": 2281
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1506857122494832e-05,
+      "loss": 1.3857,
+      "step": 2282
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1500283835266983e-05,
+      "loss": 1.3536,
+      "step": 2283
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.149370988480093e-05,
+      "loss": 1.4353,
+      "step": 2284
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1487135274002854e-05,
+      "loss": 1.4221,
+      "step": 2285
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1480560005779227e-05,
+      "loss": 1.314,
+      "step": 2286
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1473984083036813e-05,
+      "loss": 1.3106,
+      "step": 2287
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1467407508682662e-05,
+      "loss": 1.3951,
+      "step": 2288
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1460830285624119e-05,
+      "loss": 1.3218,
+      "step": 2289
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1454252416768803e-05,
+      "loss": 1.4194,
+      "step": 2290
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1447673905024633e-05,
+      "loss": 1.4013,
+      "step": 2291
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1441094753299802e-05,
+      "loss": 1.4082,
+      "step": 2292
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1434514964502792e-05,
+      "loss": 1.3418,
+      "step": 2293
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1427934541542362e-05,
+      "loss": 1.3007,
+      "step": 2294
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1421353487327554e-05,
+      "loss": 1.3562,
+      "step": 2295
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.141477180476769e-05,
+      "loss": 1.3119,
+      "step": 2296
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1408189496772369e-05,
+      "loss": 1.3409,
+      "step": 2297
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.140160656625146e-05,
+      "loss": 1.3484,
+      "step": 2298
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1395023016115118e-05,
+      "loss": 1.4275,
+      "step": 2299
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1388438849273766e-05,
+      "loss": 1.3751,
+      "step": 2300
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1381854068638101e-05,
+      "loss": 1.363,
+      "step": 2301
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1375268677119089e-05,
+      "loss": 1.3515,
+      "step": 2302
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1368682677627971e-05,
+      "loss": 1.3405,
+      "step": 2303
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1362096073076251e-05,
+      "loss": 1.3569,
+      "step": 2304
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1355508866375704e-05,
+      "loss": 1.3526,
+      "step": 2305
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1348921060438372e-05,
+      "loss": 1.3292,
+      "step": 2306
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1342332658176556e-05,
+      "loss": 1.3397,
+      "step": 2307
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.133574366250283e-05,
+      "loss": 1.285,
+      "step": 2308
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1329154076330022e-05,
+      "loss": 1.3842,
+      "step": 2309
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1322563902571227e-05,
+      "loss": 1.3365,
+      "step": 2310
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1315973144139794e-05,
+      "loss": 1.3519,
+      "step": 2311
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1309381803949333e-05,
+      "loss": 1.356,
+      "step": 2312
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1302789884913717e-05,
+      "loss": 1.4074,
+      "step": 2313
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1296197389947066e-05,
+      "loss": 1.3197,
+      "step": 2314
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.128960432196376e-05,
+      "loss": 1.3777,
+      "step": 2315
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1283010683878424e-05,
+      "loss": 1.4069,
+      "step": 2316
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.127641647860595e-05,
+      "loss": 1.2469,
+      "step": 2317
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1269821709061466e-05,
+      "loss": 1.2621,
+      "step": 2318
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1263226378160357e-05,
+      "loss": 1.3634,
+      "step": 2319
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.125663048881826e-05,
+      "loss": 1.3355,
+      "step": 2320
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1250034043951051e-05,
+      "loss": 1.416,
+      "step": 2321
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1243437046474854e-05,
+      "loss": 1.3833,
+      "step": 2322
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1236839499306036e-05,
+      "loss": 1.3571,
+      "step": 2323
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1230241405361209e-05,
+      "loss": 1.4015,
+      "step": 2324
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1223642767557228e-05,
+      "loss": 1.3609,
+      "step": 2325
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1217043588811184e-05,
+      "loss": 1.3941,
+      "step": 2326
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1210443872040416e-05,
+      "loss": 1.3652,
+      "step": 2327
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1203843620162486e-05,
+      "loss": 1.3261,
+      "step": 2328
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.119724283609521e-05,
+      "loss": 1.3489,
+      "step": 2329
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1190641522756627e-05,
+      "loss": 1.3633,
+      "step": 2330
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1184039683065014e-05,
+      "loss": 1.3659,
+      "step": 2331
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1177437319938874e-05,
+      "loss": 1.3455,
+      "step": 2332
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1170834436296955e-05,
+      "loss": 1.3109,
+      "step": 2333
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1164231035058228e-05,
+      "loss": 1.4083,
+      "step": 2334
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1157627119141891e-05,
+      "loss": 1.3917,
+      "step": 2335
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.115102269146737e-05,
+      "loss": 1.4113,
+      "step": 2336
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.114441775495432e-05,
+      "loss": 1.3523,
+      "step": 2337
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1137812312522618e-05,
+      "loss": 1.3116,
+      "step": 2338
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1131206367092369e-05,
+      "loss": 1.3743,
+      "step": 2339
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1124599921583892e-05,
+      "loss": 1.4335,
+      "step": 2340
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1117992978917735e-05,
+      "loss": 1.4348,
+      "step": 2341
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1111385542014662e-05,
+      "loss": 1.4883,
+      "step": 2342
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.110477761379566e-05,
+      "loss": 1.2904,
+      "step": 2343
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1098169197181926e-05,
+      "loss": 1.3686,
+      "step": 2344
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.109156029509488e-05,
+      "loss": 1.3808,
+      "step": 2345
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1084950910456153e-05,
+      "loss": 1.3585,
+      "step": 2346
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1078341046187588e-05,
+      "loss": 1.3139,
+      "step": 2347
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.107173070521124e-05,
+      "loss": 1.3871,
+      "step": 2348
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1065119890449383e-05,
+      "loss": 1.394,
+      "step": 2349
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1058508604824487e-05,
+      "loss": 1.3383,
+      "step": 2350
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1051896851259241e-05,
+      "loss": 1.2706,
+      "step": 2351
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1045284632676535e-05,
+      "loss": 1.3306,
+      "step": 2352
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.103867195199947e-05,
+      "loss": 1.3254,
+      "step": 2353
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1032058812151344e-05,
+      "loss": 1.3645,
+      "step": 2354
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.102544521605566e-05,
+      "loss": 1.3605,
+      "step": 2355
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1018831166636133e-05,
+      "loss": 1.3184,
+      "step": 2356
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.101221666681666e-05,
+      "loss": 1.3495,
+      "step": 2357
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1005601719521352e-05,
+      "loss": 1.3471,
+      "step": 2358
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0998986327674515e-05,
+      "loss": 1.4007,
+      "step": 2359
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0992370494200644e-05,
+      "loss": 1.3877,
+      "step": 2360
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0985754222024437e-05,
+      "loss": 1.3559,
+      "step": 2361
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0979137514070783e-05,
+      "loss": 1.397,
+      "step": 2362
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0972520373264769e-05,
+      "loss": 1.2946,
+      "step": 2363
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.096590280253166e-05,
+      "loss": 1.3764,
+      "step": 2364
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0959284804796926e-05,
+      "loss": 1.3188,
+      "step": 2365
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0952666382986216e-05,
+      "loss": 1.358,
+      "step": 2366
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0946047540025373e-05,
+      "loss": 1.3552,
+      "step": 2367
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0939428278840421e-05,
+      "loss": 1.3637,
+      "step": 2368
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.093280860235757e-05,
+      "loss": 1.4131,
+      "step": 2369
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0926188513503215e-05,
+      "loss": 1.3642,
+      "step": 2370
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0919568015203933e-05,
+      "loss": 1.323,
+      "step": 2371
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0912947110386484e-05,
+      "loss": 1.3253,
+      "step": 2372
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0906325801977804e-05,
+      "loss": 1.311,
+      "step": 2373
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0899704092905007e-05,
+      "loss": 1.3938,
+      "step": 2374
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0893081986095388e-05,
+      "loss": 1.3608,
+      "step": 2375
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0886459484476414e-05,
+      "loss": 1.279,
+      "step": 2376
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0879836590975732e-05,
+      "loss": 1.315,
+      "step": 2377
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0873213308521151e-05,
+      "loss": 1.3567,
+      "step": 2378
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.086658964004067e-05,
+      "loss": 1.3844,
+      "step": 2379
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0859965588462442e-05,
+      "loss": 1.3233,
+      "step": 2380
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0853341156714799e-05,
+      "loss": 1.3085,
+      "step": 2381
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0846716347726233e-05,
+      "loss": 1.3634,
+      "step": 2382
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.084009116442541e-05,
+      "loss": 1.3878,
+      "step": 2383
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0833465609741161e-05,
+      "loss": 1.3338,
+      "step": 2384
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0826839686602475e-05,
+      "loss": 1.3524,
+      "step": 2385
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.082021339793851e-05,
+      "loss": 1.3549,
+      "step": 2386
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0813586746678584e-05,
+      "loss": 1.3146,
+      "step": 2387
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0806959735752174e-05,
+      "loss": 1.4132,
+      "step": 2388
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.080033236808892e-05,
+      "loss": 1.2766,
+      "step": 2389
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.079370464661861e-05,
+      "loss": 1.4213,
+      "step": 2390
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0787076574271201e-05,
+      "loss": 1.3964,
+      "step": 2391
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0780448153976792e-05,
+      "loss": 1.3431,
+      "step": 2392
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0773819388665652e-05,
+      "loss": 1.3781,
+      "step": 2393
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0767190281268187e-05,
+      "loss": 1.3509,
+      "step": 2394
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0760560834714964e-05,
+      "loss": 1.3164,
+      "step": 2395
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0753931051936695e-05,
+      "loss": 1.2518,
+      "step": 2396
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0747300935864245e-05,
+      "loss": 1.3255,
+      "step": 2397
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.074067048942862e-05,
+      "loss": 1.2808,
+      "step": 2398
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0734039715560973e-05,
+      "loss": 1.3347,
+      "step": 2399
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0727408617192612e-05,
+      "loss": 1.3279,
+      "step": 2400
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0720777197254974e-05,
+      "loss": 1.36,
+      "step": 2401
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.071414545867965e-05,
+      "loss": 1.3606,
+      "step": 2402
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0707513404398365e-05,
+      "loss": 1.411,
+      "step": 2403
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0700881037342985e-05,
+      "loss": 1.3474,
+      "step": 2404
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0694248360445514e-05,
+      "loss": 1.3515,
+      "step": 2405
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0687615376638094e-05,
+      "loss": 1.4539,
+      "step": 2406
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0680982088853003e-05,
+      "loss": 1.3821,
+      "step": 2407
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0674348500022653e-05,
+      "loss": 1.3148,
+      "step": 2408
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0667714613079587e-05,
+      "loss": 1.3872,
+      "step": 2409
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0661080430956483e-05,
+      "loss": 1.3986,
+      "step": 2410
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0654445956586148e-05,
+      "loss": 1.3934,
+      "step": 2411
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0647811192901518e-05,
+      "loss": 1.3278,
+      "step": 2412
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0641176142835659e-05,
+      "loss": 1.3525,
+      "step": 2413
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.063454080932176e-05,
+      "loss": 1.3122,
+      "step": 2414
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0627905195293135e-05,
+      "loss": 1.329,
+      "step": 2415
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0621269303683227e-05,
+      "loss": 1.3146,
+      "step": 2416
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0614633137425599e-05,
+      "loss": 1.3715,
+      "step": 2417
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0607996699453932e-05,
+      "loss": 1.3401,
+      "step": 2418
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0601359992702034e-05,
+      "loss": 1.4031,
+      "step": 2419
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0594723020103826e-05,
+      "loss": 1.3483,
+      "step": 2420
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.058808578459335e-05,
+      "loss": 1.381,
+      "step": 2421
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0581448289104759e-05,
+      "loss": 1.4052,
+      "step": 2422
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0574810536572331e-05,
+      "loss": 1.4025,
+      "step": 2423
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0568172529930448e-05,
+      "loss": 1.4193,
+      "step": 2424
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0561534272113607e-05,
+      "loss": 1.3744,
+      "step": 2425
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0554895766056419e-05,
+      "loss": 1.325,
+      "step": 2426
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0548257014693602e-05,
+      "loss": 1.348,
+      "step": 2427
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0541618020959981e-05,
+      "loss": 1.3738,
+      "step": 2428
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0534978787790494e-05,
+      "loss": 1.3473,
+      "step": 2429
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0528339318120175e-05,
+      "loss": 1.3645,
+      "step": 2430
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0521699614884178e-05,
+      "loss": 1.3245,
+      "step": 2431
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.051505968101774e-05,
+      "loss": 1.3791,
+      "step": 2432
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0508419519456219e-05,
+      "loss": 1.3297,
+      "step": 2433
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0501779133135062e-05,
+      "loss": 1.3843,
+      "step": 2434
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0495138524989816e-05,
+      "loss": 1.3792,
+      "step": 2435
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0488497697956134e-05,
+      "loss": 1.408,
+      "step": 2436
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0481856654969758e-05,
+      "loss": 1.3762,
+      "step": 2437
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0475215398966527e-05,
+      "loss": 1.3445,
+      "step": 2438
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0468573932882382e-05,
+      "loss": 1.4093,
+      "step": 2439
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0461932259653342e-05,
+      "loss": 1.4032,
+      "step": 2440
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0455290382215529e-05,
+      "loss": 1.3139,
+      "step": 2441
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.044864830350515e-05,
+      "loss": 1.36,
+      "step": 2442
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0442006026458506e-05,
+      "loss": 1.3388,
+      "step": 2443
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0435363554011981e-05,
+      "loss": 1.3144,
+      "step": 2444
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0428720889102048e-05,
+      "loss": 1.3384,
+      "step": 2445
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0422078034665262e-05,
+      "loss": 1.3953,
+      "step": 2446
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0415434993638269e-05,
+      "loss": 1.3317,
+      "step": 2447
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0408791768957785e-05,
+      "loss": 1.3116,
+      "step": 2448
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0402148363560618e-05,
+      "loss": 1.3513,
+      "step": 2449
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0395504780383653e-05,
+      "loss": 1.3388,
+      "step": 2450
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.038886102236385e-05,
+      "loss": 1.3147,
+      "step": 2451
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0382217092438256e-05,
+      "loss": 1.3711,
+      "step": 2452
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0375572993543981e-05,
+      "loss": 1.2904,
+      "step": 2453
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0368928728618219e-05,
+      "loss": 1.3078,
+      "step": 2454
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0362284300598232e-05,
+      "loss": 1.3194,
+      "step": 2455
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0355639712421363e-05,
+      "loss": 1.436,
+      "step": 2456
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0348994967025012e-05,
+      "loss": 1.3735,
+      "step": 2457
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0342350067346659e-05,
+      "loss": 1.3712,
+      "step": 2458
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0335705016323845e-05,
+      "loss": 1.2845,
+      "step": 2459
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0329059816894187e-05,
+      "loss": 1.3541,
+      "step": 2460
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.032241447199536e-05,
+      "loss": 1.4521,
+      "step": 2461
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.031576898456511e-05,
+      "loss": 1.3767,
+      "step": 2462
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0309123357541237e-05,
+      "loss": 1.2807,
+      "step": 2463
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0302477593861608e-05,
+      "loss": 1.3221,
+      "step": 2464
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.029583169646415e-05,
+      "loss": 1.3926,
+      "step": 2465
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0289185668286854e-05,
+      "loss": 1.4272,
+      "step": 2466
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0282539512267758e-05,
+      "loss": 1.2841,
+      "step": 2467
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0275893231344964e-05,
+      "loss": 1.3281,
+      "step": 2468
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.026924682845663e-05,
+      "loss": 1.374,
+      "step": 2469
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0262600306540962e-05,
+      "loss": 1.2809,
+      "step": 2470
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0255953668536223e-05,
+      "loss": 1.3322,
+      "step": 2471
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0249306917380731e-05,
+      "loss": 1.4129,
+      "step": 2472
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0242660056012844e-05,
+      "loss": 1.3868,
+      "step": 2473
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0236013087370975e-05,
+      "loss": 1.3022,
+      "step": 2474
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0229366014393586e-05,
+      "loss": 1.3092,
+      "step": 2475
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0222718840019181e-05,
+      "loss": 1.3151,
+      "step": 2476
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0216071567186312e-05,
+      "loss": 1.3568,
+      "step": 2477
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0209424198833571e-05,
+      "loss": 1.3403,
+      "step": 2478
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0202776737899599e-05,
+      "loss": 1.3506,
+      "step": 2479
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0196129187323066e-05,
+      "loss": 1.3012,
+      "step": 2480
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0189481550042695e-05,
+      "loss": 1.4217,
+      "step": 2481
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0182833828997238e-05,
+      "loss": 1.3664,
+      "step": 2482
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0176186027125487e-05,
+      "loss": 1.3184,
+      "step": 2483
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0169538147366276e-05,
+      "loss": 1.4474,
+      "step": 2484
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0162890192658459e-05,
+      "loss": 1.324,
+      "step": 2485
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0156242165940937e-05,
+      "loss": 1.32,
+      "step": 2486
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0149594070152638e-05,
+      "loss": 1.3036,
+      "step": 2487
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0142945908232515e-05,
+      "loss": 1.3174,
+      "step": 2488
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.013629768311956e-05,
+      "loss": 1.3603,
+      "step": 2489
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0129649397752785e-05,
+      "loss": 1.3395,
+      "step": 2490
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0123001055071233e-05,
+      "loss": 1.3833,
+      "step": 2491
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0116352658013973e-05,
+      "loss": 1.3754,
+      "step": 2492
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0109704209520093e-05,
+      "loss": 1.3185,
+      "step": 2493
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0103055712528706e-05,
+      "loss": 1.3247,
+      "step": 2494
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.009640716997895e-05,
+      "loss": 1.3632,
+      "step": 2495
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.008975858480998e-05,
+      "loss": 1.3174,
+      "step": 2496
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0083109959960974e-05,
+      "loss": 1.3289,
+      "step": 2497
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0076461298371114e-05,
+      "loss": 1.2626,
+      "step": 2498
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0069812602979617e-05,
+      "loss": 1.3213,
+      "step": 2499
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0063163876725703e-05,
+      "loss": 1.35,
+      "step": 2500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0056515122548607e-05,
+      "loss": 1.3201,
+      "step": 2501
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0049866343387582e-05,
+      "loss": 1.3966,
+      "step": 2502
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0043217542181885e-05,
+      "loss": 1.3428,
+      "step": 2503
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0036568721870786e-05,
+      "loss": 1.3445,
+      "step": 2504
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0029919885393564e-05,
+      "loss": 1.3061,
+      "step": 2505
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0023271035689506e-05,
+      "loss": 1.3885,
+      "step": 2506
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0016622175697898e-05,
+      "loss": 1.36,
+      "step": 2507
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.000997330835804e-05,
+      "loss": 1.4367,
+      "step": 2508
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0003324436609232e-05,
+      "loss": 1.4312,
+      "step": 2509
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.996675563390773e-06,
+      "loss": 1.3465,
+      "step": 2510
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.990026691641963e-06,
+      "loss": 1.3481,
+      "step": 2511
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.983377824302107e-06,
+      "loss": 1.3652,
+      "step": 2512
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.976728964310499e-06,
+      "loss": 1.3708,
+      "step": 2513
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.97008011460644e-06,
+      "loss": 1.3576,
+      "step": 2514
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.963431278129218e-06,
+      "loss": 1.2612,
+      "step": 2515
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.956782457818118e-06,
+      "loss": 1.3621,
+      "step": 2516
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.950133656612421e-06,
+      "loss": 1.3553,
+      "step": 2517
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.943484877451395e-06,
+      "loss": 1.4296,
+      "step": 2518
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.9368361232743e-06,
+      "loss": 1.4249,
+      "step": 2519
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.930187397020385e-06,
+      "loss": 1.2579,
+      "step": 2520
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.92353870162889e-06,
+      "loss": 1.36,
+      "step": 2521
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.916890040039031e-06,
+      "loss": 1.3994,
+      "step": 2522
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.910241415190022e-06,
+      "loss": 1.3352,
+      "step": 2523
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.903592830021052e-06,
+      "loss": 1.3547,
+      "step": 2524
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.896944287471295e-06,
+      "loss": 1.3601,
+      "step": 2525
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.890295790479912e-06,
+      "loss": 1.3541,
+      "step": 2526
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.883647341986032e-06,
+      "loss": 1.3886,
+      "step": 2527
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.87699894492877e-06,
+      "loss": 1.3647,
+      "step": 2528
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.870350602247218e-06,
+      "loss": 1.2716,
+      "step": 2529
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.863702316880444e-06,
+      "loss": 1.3392,
+      "step": 2530
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.857054091767489e-06,
+      "loss": 1.3678,
+      "step": 2531
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.850405929847367e-06,
+      "loss": 1.3431,
+      "step": 2532
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.843757834059065e-06,
+      "loss": 1.2785,
+      "step": 2533
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.837109807341543e-06,
+      "loss": 1.4181,
+      "step": 2534
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.830461852633727e-06,
+      "loss": 1.3593,
+      "step": 2535
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.823813972874514e-06,
+      "loss": 1.351,
+      "step": 2536
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.817166171002766e-06,
+      "loss": 1.3304,
+      "step": 2537
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.81051844995731e-06,
+      "loss": 1.3258,
+      "step": 2538
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.803870812676938e-06,
+      "loss": 1.3087,
+      "step": 2539
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.797223262100406e-06,
+      "loss": 1.3652,
+      "step": 2540
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.790575801166432e-06,
+      "loss": 1.3379,
+      "step": 2541
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.783928432813688e-06,
+      "loss": 1.26,
+      "step": 2542
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.777281159980824e-06,
+      "loss": 1.4223,
+      "step": 2543
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.770633985606419e-06,
+      "loss": 1.3263,
+      "step": 2544
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.76398691262903e-06,
+      "loss": 1.3536,
+      "step": 2545
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.757339943987161e-06,
+      "loss": 1.2658,
+      "step": 2546
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.750693082619274e-06,
+      "loss": 1.4624,
+      "step": 2547
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.74404633146378e-06,
+      "loss": 1.2998,
+      "step": 2548
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.737399693459042e-06,
+      "loss": 1.3224,
+      "step": 2549
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.730753171543374e-06,
+      "loss": 1.2854,
+      "step": 2550
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.724106768655039e-06,
+      "loss": 1.3855,
+      "step": 2551
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.717460487732246e-06,
+      "loss": 1.3238,
+      "step": 2552
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.710814331713149e-06,
+      "loss": 1.3073,
+      "step": 2553
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.704168303535851e-06,
+      "loss": 1.324,
+      "step": 2554
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.697522406138395e-06,
+      "loss": 1.396,
+      "step": 2555
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.690876642458764e-06,
+      "loss": 1.3962,
+      "step": 2556
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.684231015434891e-06,
+      "loss": 1.3894,
+      "step": 2557
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.677585528004638e-06,
+      "loss": 1.366,
+      "step": 2558
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.670940183105813e-06,
+      "loss": 1.3164,
+      "step": 2559
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.664294983676159e-06,
+      "loss": 1.3719,
+      "step": 2560
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.657649932653348e-06,
+      "loss": 1.3809,
+      "step": 2561
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.651005032974994e-06,
+      "loss": 1.3688,
+      "step": 2562
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.644360287578642e-06,
+      "loss": 1.3794,
+      "step": 2563
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.63771569940177e-06,
+      "loss": 1.4428,
+      "step": 2564
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.631071271381785e-06,
+      "loss": 1.3174,
+      "step": 2565
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.624427006456022e-06,
+      "loss": 1.3881,
+      "step": 2566
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.617782907561748e-06,
+      "loss": 1.364,
+      "step": 2567
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.611138977636154e-06,
+      "loss": 1.3575,
+      "step": 2568
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.60449521961635e-06,
+      "loss": 1.3623,
+      "step": 2569
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.597851636439385e-06,
+      "loss": 1.3432,
+      "step": 2570
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.591208231042218e-06,
+      "loss": 1.4136,
+      "step": 2571
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.584565006361735e-06,
+      "loss": 1.4289,
+      "step": 2572
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.577921965334738e-06,
+      "loss": 1.3873,
+      "step": 2573
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.571279110897953e-06,
+      "loss": 1.4511,
+      "step": 2574
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.564636445988019e-06,
+      "loss": 1.3899,
+      "step": 2575
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.557993973541494e-06,
+      "loss": 1.3612,
+      "step": 2576
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.551351696494854e-06,
+      "loss": 1.3522,
+      "step": 2577
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.544709617784476e-06,
+      "loss": 1.3288,
+      "step": 2578
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.538067740346663e-06,
+      "loss": 1.312,
+      "step": 2579
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.531426067117623e-06,
+      "loss": 1.3072,
+      "step": 2580
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.524784601033475e-06,
+      "loss": 1.3445,
+      "step": 2581
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.518143345030247e-06,
+      "loss": 1.3548,
+      "step": 2582
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.511502302043867e-06,
+      "loss": 1.3586,
+      "step": 2583
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.504861475010186e-06,
+      "loss": 1.4068,
+      "step": 2584
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.498220866864943e-06,
+      "loss": 1.3553,
+      "step": 2585
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.491580480543784e-06,
+      "loss": 1.4406,
+      "step": 2586
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.484940318982261e-06,
+      "loss": 1.364,
+      "step": 2587
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.478300385115825e-06,
+      "loss": 1.4159,
+      "step": 2588
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.471660681879825e-06,
+      "loss": 1.352,
+      "step": 2589
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.465021212209508e-06,
+      "loss": 1.3089,
+      "step": 2590
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.45838197904002e-06,
+      "loss": 1.3765,
+      "step": 2591
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.4517429853064e-06,
+      "loss": 1.3163,
+      "step": 2592
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.445104233943586e-06,
+      "loss": 1.2992,
+      "step": 2593
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.438465727886398e-06,
+      "loss": 1.381,
+      "step": 2594
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.431827470069557e-06,
+      "loss": 1.3694,
+      "step": 2595
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.425189463427672e-06,
+      "loss": 1.3566,
+      "step": 2596
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.418551710895243e-06,
+      "loss": 1.3355,
+      "step": 2597
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.411914215406654e-06,
+      "loss": 1.3828,
+      "step": 2598
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.405276979896178e-06,
+      "loss": 1.3303,
+      "step": 2599
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.39864000729797e-06,
+      "loss": 1.4071,
+      "step": 2600
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.392003300546071e-06,
+      "loss": 1.3348,
+      "step": 2601
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.385366862574405e-06,
+      "loss": 1.326,
+      "step": 2602
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.378730696316774e-06,
+      "loss": 1.378,
+      "step": 2603
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.372094804706867e-06,
+      "loss": 1.3921,
+      "step": 2604
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.365459190678243e-06,
+      "loss": 1.3895,
+      "step": 2605
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.358823857164343e-06,
+      "loss": 1.2911,
+      "step": 2606
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.352188807098482e-06,
+      "loss": 1.3188,
+      "step": 2607
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.345554043413852e-06,
+      "loss": 1.3941,
+      "step": 2608
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.338919569043517e-06,
+      "loss": 1.3772,
+      "step": 2609
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.332285386920416e-06,
+      "loss": 1.4172,
+      "step": 2610
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.32565149997735e-06,
+      "loss": 1.3093,
+      "step": 2611
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.319017911147e-06,
+      "loss": 1.3745,
+      "step": 2612
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.312384623361909e-06,
+      "loss": 1.2245,
+      "step": 2613
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.30575163955449e-06,
+      "loss": 1.3002,
+      "step": 2614
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.299118962657018e-06,
+      "loss": 1.3841,
+      "step": 2615
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.292486595601637e-06,
+      "loss": 1.3119,
+      "step": 2616
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.285854541320352e-06,
+      "loss": 1.3678,
+      "step": 2617
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.279222802745028e-06,
+      "loss": 1.3547,
+      "step": 2618
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.272591382807391e-06,
+      "loss": 1.3349,
+      "step": 2619
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.265960284439028e-06,
+      "loss": 1.2314,
+      "step": 2620
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.259329510571385e-06,
+      "loss": 1.3663,
+      "step": 2621
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.252699064135759e-06,
+      "loss": 1.325,
+      "step": 2622
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.246068948063305e-06,
+      "loss": 1.3989,
+      "step": 2623
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.239439165285037e-06,
+      "loss": 1.2965,
+      "step": 2624
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.232809718731815e-06,
+      "loss": 1.3041,
+      "step": 2625
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.22618061133435e-06,
+      "loss": 1.294,
+      "step": 2626
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.219551846023211e-06,
+      "loss": 1.3136,
+      "step": 2627
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.212923425728806e-06,
+      "loss": 1.3653,
+      "step": 2628
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.206295353381395e-06,
+      "loss": 1.2886,
+      "step": 2629
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.199667631911086e-06,
+      "loss": 1.2739,
+      "step": 2630
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.19304026424783e-06,
+      "loss": 1.3584,
+      "step": 2631
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.18641325332142e-06,
+      "loss": 1.27,
+      "step": 2632
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.179786602061493e-06,
+      "loss": 1.3603,
+      "step": 2633
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.173160313397528e-06,
+      "loss": 1.3035,
+      "step": 2634
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.166534390258842e-06,
+      "loss": 1.3649,
+      "step": 2635
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.159908835574591e-06,
+      "loss": 1.3514,
+      "step": 2636
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.153283652273768e-06,
+      "loss": 1.3576,
+      "step": 2637
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.146658843285203e-06,
+      "loss": 1.3243,
+      "step": 2638
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.140034411537558e-06,
+      "loss": 1.3674,
+      "step": 2639
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.13341035995933e-06,
+      "loss": 1.3683,
+      "step": 2640
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.126786691478849e-06,
+      "loss": 1.4601,
+      "step": 2641
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.120163409024272e-06,
+      "loss": 1.3904,
+      "step": 2642
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.113540515523587e-06,
+      "loss": 1.3219,
+      "step": 2643
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.106918013904617e-06,
+      "loss": 1.3125,
+      "step": 2644
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.100295907094998e-06,
+      "loss": 1.3585,
+      "step": 2645
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.093674198022201e-06,
+      "loss": 1.2543,
+      "step": 2646
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.087052889613519e-06,
+      "loss": 1.3189,
+      "step": 2647
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.08043198479607e-06,
+      "loss": 1.3106,
+      "step": 2648
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.073811486496788e-06,
+      "loss": 1.3484,
+      "step": 2649
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.067191397642435e-06,
+      "loss": 1.3893,
+      "step": 2650
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.060571721159584e-06,
+      "loss": 1.4561,
+      "step": 2651
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.05395245997463e-06,
+      "loss": 1.3322,
+      "step": 2652
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.047333617013786e-06,
+      "loss": 1.4217,
+      "step": 2653
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.040715195203077e-06,
+      "loss": 1.2855,
+      "step": 2654
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.034097197468343e-06,
+      "loss": 1.428,
+      "step": 2655
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.027479626735235e-06,
+      "loss": 1.3954,
+      "step": 2656
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.020862485929219e-06,
+      "loss": 1.4476,
+      "step": 2657
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.014245777975565e-06,
+      "loss": 1.289,
+      "step": 2658
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.007629505799359e-06,
+      "loss": 1.3231,
+      "step": 2659
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.001013672325491e-06,
+      "loss": 1.3282,
+      "step": 2660
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.994398280478653e-06,
+      "loss": 1.3835,
+      "step": 2661
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.987783333183345e-06,
+      "loss": 1.3016,
+      "step": 2662
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.981168833363874e-06,
+      "loss": 1.345,
+      "step": 2663
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.974554783944342e-06,
+      "loss": 1.2928,
+      "step": 2664
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.967941187848662e-06,
+      "loss": 1.3819,
+      "step": 2665
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.961328048000535e-06,
+      "loss": 1.3852,
+      "step": 2666
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.954715367323468e-06,
+      "loss": 1.3234,
+      "step": 2667
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.948103148740762e-06,
+      "loss": 1.2993,
+      "step": 2668
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.941491395175516e-06,
+      "loss": 1.3668,
+      "step": 2669
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.934880109550622e-06,
+      "loss": 1.2675,
+      "step": 2670
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.928269294788763e-06,
+      "loss": 1.365,
+      "step": 2671
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.921658953812416e-06,
+      "loss": 1.3791,
+      "step": 2672
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.91504908954385e-06,
+      "loss": 1.3343,
+      "step": 2673
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.90843970490512e-06,
+      "loss": 1.2983,
+      "step": 2674
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.901830802818074e-06,
+      "loss": 1.3781,
+      "step": 2675
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.895222386204341e-06,
+      "loss": 1.3743,
+      "step": 2676
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.888614457985343e-06,
+      "loss": 1.2337,
+      "step": 2677
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.88200702108227e-06,
+      "loss": 1.336,
+      "step": 2678
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.875400078416113e-06,
+      "loss": 1.2622,
+      "step": 2679
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.868793632907636e-06,
+      "loss": 1.4168,
+      "step": 2680
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.862187687477386e-06,
+      "loss": 1.3352,
+      "step": 2681
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.855582245045682e-06,
+      "loss": 1.3983,
+      "step": 2682
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.848977308532632e-06,
+      "loss": 1.3891,
+      "step": 2683
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.84237288085811e-06,
+      "loss": 1.3425,
+      "step": 2684
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.835768964941773e-06,
+      "loss": 1.3625,
+      "step": 2685
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.829165563703046e-06,
+      "loss": 1.3283,
+      "step": 2686
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.822562680061127e-06,
+      "loss": 1.3765,
+      "step": 2687
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.815960316934991e-06,
+      "loss": 1.3649,
+      "step": 2688
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.809358477243375e-06,
+      "loss": 1.4085,
+      "step": 2689
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.80275716390479e-06,
+      "loss": 1.324,
+      "step": 2690
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.796156379837512e-06,
+      "loss": 1.3827,
+      "step": 2691
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.789556127959586e-06,
+      "loss": 1.3974,
+      "step": 2692
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.782956411188815e-06,
+      "loss": 1.3817,
+      "step": 2693
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.776357232442778e-06,
+      "loss": 1.3578,
+      "step": 2694
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.769758594638796e-06,
+      "loss": 1.3327,
+      "step": 2695
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.76316050069397e-06,
+      "loss": 1.3332,
+      "step": 2696
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.756562953525151e-06,
+      "loss": 1.395,
+      "step": 2697
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.749965956048952e-06,
+      "loss": 1.3836,
+      "step": 2698
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.743369511181743e-06,
+      "loss": 1.2899,
+      "step": 2699
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.736773621839645e-06,
+      "loss": 1.3347,
+      "step": 2700
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.730178290938538e-06,
+      "loss": 1.4186,
+      "step": 2701
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.723583521394054e-06,
+      "loss": 1.3942,
+      "step": 2702
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.71698931612158e-06,
+      "loss": 1.3009,
+      "step": 2703
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.710395678036246e-06,
+      "loss": 1.3062,
+      "step": 2704
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.703802610052936e-06,
+      "loss": 1.2831,
+      "step": 2705
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.697210115086285e-06,
+      "loss": 1.3474,
+      "step": 2706
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.690618196050667e-06,
+      "loss": 1.4024,
+      "step": 2707
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.684026855860208e-06,
+      "loss": 1.4029,
+      "step": 2708
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.677436097428775e-06,
+      "loss": 1.4186,
+      "step": 2709
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.670845923669984e-06,
+      "loss": 1.4295,
+      "step": 2710
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.664256337497175e-06,
+      "loss": 1.314,
+      "step": 2711
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.657667341823449e-06,
+      "loss": 1.3559,
+      "step": 2712
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.651078939561635e-06,
+      "loss": 1.3594,
+      "step": 2713
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.644491133624298e-06,
+      "loss": 1.3793,
+      "step": 2714
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.637903926923752e-06,
+      "loss": 1.2873,
+      "step": 2715
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.631317322372032e-06,
+      "loss": 1.3015,
+      "step": 2716
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.624731322880913e-06,
+      "loss": 1.3692,
+      "step": 2717
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.618145931361902e-06,
+      "loss": 1.4386,
+      "step": 2718
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.611561150726236e-06,
+      "loss": 1.3836,
+      "step": 2719
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.604976983884883e-06,
+      "loss": 1.3283,
+      "step": 2720
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.598393433748542e-06,
+      "loss": 1.3586,
+      "step": 2721
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.591810503227634e-06,
+      "loss": 1.3952,
+      "step": 2722
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.585228195232311e-06,
+      "loss": 1.3782,
+      "step": 2723
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.578646512672446e-06,
+      "loss": 1.3274,
+      "step": 2724
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.57206545845764e-06,
+      "loss": 1.2959,
+      "step": 2725
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.565485035497208e-06,
+      "loss": 1.3669,
+      "step": 2726
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.558905246700202e-06,
+      "loss": 1.3413,
+      "step": 2727
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.552326094975369e-06,
+      "loss": 1.3883,
+      "step": 2728
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.545747583231199e-06,
+      "loss": 1.3256,
+      "step": 2729
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.539169714375885e-06,
+      "loss": 1.3211,
+      "step": 2730
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.53259249131734e-06,
+      "loss": 1.3323,
+      "step": 2731
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.52601591696319e-06,
+      "loss": 1.3413,
+      "step": 2732
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.519439994220777e-06,
+      "loss": 1.3947,
+      "step": 2733
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.51286472599715e-06,
+      "loss": 1.3253,
+      "step": 2734
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.506290115199074e-06,
+      "loss": 1.3889,
+      "step": 2735
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.49971616473302e-06,
+      "loss": 1.2885,
+      "step": 2736
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.49314287750517e-06,
+      "loss": 1.3326,
+      "step": 2737
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.486570256421407e-06,
+      "loss": 1.3561,
+      "step": 2738
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.479998304387329e-06,
+      "loss": 1.2791,
+      "step": 2739
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.47342702430823e-06,
+      "loss": 1.4414,
+      "step": 2740
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.466856419089107e-06,
+      "loss": 1.3788,
+      "step": 2741
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.460286491634664e-06,
+      "loss": 1.3483,
+      "step": 2742
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.453717244849303e-06,
+      "loss": 1.382,
+      "step": 2743
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.447148681637127e-06,
+      "loss": 1.4323,
+      "step": 2744
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.440580804901927e-06,
+      "loss": 1.3203,
+      "step": 2745
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.434013617547205e-06,
+      "loss": 1.3472,
+      "step": 2746
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.427447122476148e-06,
+      "loss": 1.3202,
+      "step": 2747
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.420881322591643e-06,
+      "loss": 1.4043,
+      "step": 2748
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.414316220796265e-06,
+      "loss": 1.3064,
+      "step": 2749
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.407751819992281e-06,
+      "loss": 1.3121,
+      "step": 2750
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.401188123081653e-06,
+      "loss": 1.3739,
+      "step": 2751
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.394625132966025e-06,
+      "loss": 1.4417,
+      "step": 2752
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.388062852546735e-06,
+      "loss": 1.3446,
+      "step": 2753
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.381501284724801e-06,
+      "loss": 1.3481,
+      "step": 2754
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.37494043240093e-06,
+      "loss": 1.3582,
+      "step": 2755
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.368380298475515e-06,
+      "loss": 1.3763,
+      "step": 2756
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.361820885848623e-06,
+      "loss": 1.3357,
+      "step": 2757
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.355262197420011e-06,
+      "loss": 1.3787,
+      "step": 2758
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.348704236089112e-06,
+      "loss": 1.3165,
+      "step": 2759
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.342147004755038e-06,
+      "loss": 1.3629,
+      "step": 2760
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.335590506316573e-06,
+      "loss": 1.3442,
+      "step": 2761
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.329034743672187e-06,
+      "loss": 1.3339,
+      "step": 2762
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.32247971972002e-06,
+      "loss": 1.3195,
+      "step": 2763
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.315925437357883e-06,
+      "loss": 1.3292,
+      "step": 2764
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.309371899483261e-06,
+      "loss": 1.3732,
+      "step": 2765
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.302819108993311e-06,
+      "loss": 1.3121,
+      "step": 2766
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.296267068784862e-06,
+      "loss": 1.2918,
+      "step": 2767
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.289715781754404e-06,
+      "loss": 1.3372,
+      "step": 2768
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.283165250798099e-06,
+      "loss": 1.3799,
+      "step": 2769
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.276615478811774e-06,
+      "loss": 1.3122,
+      "step": 2770
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.270066468690922e-06,
+      "loss": 1.2509,
+      "step": 2771
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.263518223330698e-06,
+      "loss": 1.3368,
+      "step": 2772
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.256970745625916e-06,
+      "loss": 1.3776,
+      "step": 2773
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.250424038471054e-06,
+      "loss": 1.3701,
+      "step": 2774
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.24387810476025e-06,
+      "loss": 1.2937,
+      "step": 2775
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.237332947387296e-06,
+      "loss": 1.4188,
+      "step": 2776
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.230788569245648e-06,
+      "loss": 1.3213,
+      "step": 2777
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.224244973228409e-06,
+      "loss": 1.454,
+      "step": 2778
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.217702162228337e-06,
+      "loss": 1.3079,
+      "step": 2779
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.211160139137854e-06,
+      "loss": 1.282,
+      "step": 2780
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.204618906849022e-06,
+      "loss": 1.3692,
+      "step": 2781
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.198078468253556e-06,
+      "loss": 1.3165,
+      "step": 2782
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.191538826242823e-06,
+      "loss": 1.3168,
+      "step": 2783
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.184999983707838e-06,
+      "loss": 1.4428,
+      "step": 2784
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.178461943539258e-06,
+      "loss": 1.2911,
+      "step": 2785
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.171924708627387e-06,
+      "loss": 1.4577,
+      "step": 2786
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.165388281862177e-06,
+      "loss": 1.3111,
+      "step": 2787
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.15885266613322e-06,
+      "loss": 1.3748,
+      "step": 2788
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.152317864329748e-06,
+      "loss": 1.3789,
+      "step": 2789
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.145783879340635e-06,
+      "loss": 1.3683,
+      "step": 2790
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.13925071405439e-06,
+      "loss": 1.3186,
+      "step": 2791
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.132718371359168e-06,
+      "loss": 1.3357,
+      "step": 2792
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.126186854142752e-06,
+      "loss": 1.362,
+      "step": 2793
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.119656165292572e-06,
+      "loss": 1.3326,
+      "step": 2794
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.11312630769567e-06,
+      "loss": 1.327,
+      "step": 2795
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.10659728423874e-06,
+      "loss": 1.3575,
+      "step": 2796
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.100069097808103e-06,
+      "loss": 1.3929,
+      "step": 2797
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.093541751289705e-06,
+      "loss": 1.3311,
+      "step": 2798
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.087015247569125e-06,
+      "loss": 1.314,
+      "step": 2799
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.080489589531567e-06,
+      "loss": 1.3197,
+      "step": 2800
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.073964780061863e-06,
+      "loss": 1.3702,
+      "step": 2801
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.06744082204447e-06,
+      "loss": 1.3786,
+      "step": 2802
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.060917718363466e-06,
+      "loss": 1.3617,
+      "step": 2803
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.054395471902555e-06,
+      "loss": 1.3671,
+      "step": 2804
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.047874085545058e-06,
+      "loss": 1.3974,
+      "step": 2805
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.041353562173918e-06,
+      "loss": 1.416,
+      "step": 2806
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.034833904671698e-06,
+      "loss": 1.3261,
+      "step": 2807
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.028315115920577e-06,
+      "loss": 1.3069,
+      "step": 2808
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.021797198802348e-06,
+      "loss": 1.3214,
+      "step": 2809
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.01528015619842e-06,
+      "loss": 1.3622,
+      "step": 2810
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.008763990989821e-06,
+      "loss": 1.3848,
+      "step": 2811
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.002248706057177e-06,
+      "loss": 1.3505,
+      "step": 2812
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.995734304280737e-06,
+      "loss": 1.3503,
+      "step": 2813
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.989220788540356e-06,
+      "loss": 1.378,
+      "step": 2814
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.982708161715499e-06,
+      "loss": 1.4068,
+      "step": 2815
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.976196426685235e-06,
+      "loss": 1.3789,
+      "step": 2816
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.96968558632824e-06,
+      "loss": 1.3124,
+      "step": 2817
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.963175643522797e-06,
+      "loss": 1.3401,
+      "step": 2818
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.956666601146786e-06,
+      "loss": 1.4172,
+      "step": 2819
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.950158462077697e-06,
+      "loss": 1.272,
+      "step": 2820
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.943651229192615e-06,
+      "loss": 1.3181,
+      "step": 2821
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.937144905368226e-06,
+      "loss": 1.3466,
+      "step": 2822
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.930639493480814e-06,
+      "loss": 1.337,
+      "step": 2823
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.92413499640626e-06,
+      "loss": 1.2738,
+      "step": 2824
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.91763141702004e-06,
+      "loss": 1.3331,
+      "step": 2825
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.911128758197226e-06,
+      "loss": 1.3224,
+      "step": 2826
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.904627022812484e-06,
+      "loss": 1.3779,
+      "step": 2827
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.898126213740063e-06,
+      "loss": 1.3437,
+      "step": 2828
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.891626333853813e-06,
+      "loss": 1.3518,
+      "step": 2829
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.885127386027169e-06,
+      "loss": 1.3513,
+      "step": 2830
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.878629373133155e-06,
+      "loss": 1.4093,
+      "step": 2831
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.872132298044382e-06,
+      "loss": 1.3843,
+      "step": 2832
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.86563616363304e-06,
+      "loss": 1.3124,
+      "step": 2833
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.859140972770916e-06,
+      "loss": 1.3201,
+      "step": 2834
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.852646728329368e-06,
+      "loss": 1.3147,
+      "step": 2835
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.846153433179342e-06,
+      "loss": 1.3251,
+      "step": 2836
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.839661090191362e-06,
+      "loss": 1.4232,
+      "step": 2837
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.833169702235531e-06,
+      "loss": 1.3687,
+      "step": 2838
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.82667927218153e-06,
+      "loss": 1.3423,
+      "step": 2839
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.820189802898619e-06,
+      "loss": 1.3589,
+      "step": 2840
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.813701297255629e-06,
+      "loss": 1.3437,
+      "step": 2841
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.807213758120965e-06,
+      "loss": 1.3915,
+      "step": 2842
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.800727188362613e-06,
+      "loss": 1.3598,
+      "step": 2843
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.794241590848123e-06,
+      "loss": 1.3688,
+      "step": 2844
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.78775696844461e-06,
+      "loss": 1.2588,
+      "step": 2845
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.781273324018768e-06,
+      "loss": 1.341,
+      "step": 2846
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.774790660436857e-06,
+      "loss": 1.3705,
+      "step": 2847
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.7683089805647e-06,
+      "loss": 1.381,
+      "step": 2848
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.761828287267688e-06,
+      "loss": 1.447,
+      "step": 2849
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.755348583410773e-06,
+      "loss": 1.3671,
+      "step": 2850
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.74886987185847e-06,
+      "loss": 1.2371,
+      "step": 2851
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.742392155474858e-06,
+      "loss": 1.4518,
+      "step": 2852
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.735915437123574e-06,
+      "loss": 1.3373,
+      "step": 2853
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.729439719667816e-06,
+      "loss": 1.3231,
+      "step": 2854
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.722965005970335e-06,
+      "loss": 1.3405,
+      "step": 2855
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.716491298893443e-06,
+      "loss": 1.3083,
+      "step": 2856
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.710018601299004e-06,
+      "loss": 1.3418,
+      "step": 2857
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.703546916048437e-06,
+      "loss": 1.2647,
+      "step": 2858
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.697076246002712e-06,
+      "loss": 1.3263,
+      "step": 2859
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.690606594022355e-06,
+      "loss": 1.3914,
+      "step": 2860
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.684137962967437e-06,
+      "loss": 1.3477,
+      "step": 2861
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.677670355697577e-06,
+      "loss": 1.3968,
+      "step": 2862
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.671203775071942e-06,
+      "loss": 1.3671,
+      "step": 2863
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.664738223949252e-06,
+      "loss": 1.329,
+      "step": 2864
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.658273705187762e-06,
+      "loss": 1.3692,
+      "step": 2865
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.651810221645279e-06,
+      "loss": 1.3779,
+      "step": 2866
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.645347776179144e-06,
+      "loss": 1.3362,
+      "step": 2867
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.638886371646245e-06,
+      "loss": 1.3404,
+      "step": 2868
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.632426010903013e-06,
+      "loss": 1.2887,
+      "step": 2869
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.625966696805406e-06,
+      "loss": 1.2984,
+      "step": 2870
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.6195084322089306e-06,
+      "loss": 1.2973,
+      "step": 2871
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.613051219968624e-06,
+      "loss": 1.3494,
+      "step": 2872
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.606595062939058e-06,
+      "loss": 1.352,
+      "step": 2873
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.600139963974341e-06,
+      "loss": 1.3966,
+      "step": 2874
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.593685925928111e-06,
+      "loss": 1.3202,
+      "step": 2875
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.587232951653539e-06,
+      "loss": 1.3222,
+      "step": 2876
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.580781044003324e-06,
+      "loss": 1.314,
+      "step": 2877
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.574330205829694e-06,
+      "loss": 1.3287,
+      "step": 2878
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.567880439984402e-06,
+      "loss": 1.396,
+      "step": 2879
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.561431749318732e-06,
+      "loss": 1.3038,
+      "step": 2880
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.554984136683488e-06,
+      "loss": 1.3477,
+      "step": 2881
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5485376049290014e-06,
+      "loss": 1.3321,
+      "step": 2882
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.542092156905123e-06,
+      "loss": 1.387,
+      "step": 2883
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.535647795461224e-06,
+      "loss": 1.3689,
+      "step": 2884
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.529204523446198e-06,
+      "loss": 1.4277,
+      "step": 2885
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.522762343708454e-06,
+      "loss": 1.3356,
+      "step": 2886
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.516321259095921e-06,
+      "loss": 1.3324,
+      "step": 2887
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.509881272456041e-06,
+      "loss": 1.3965,
+      "step": 2888
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.503442386635774e-06,
+      "loss": 1.3612,
+      "step": 2889
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4970046044815905e-06,
+      "loss": 1.3602,
+      "step": 2890
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.490567928839472e-06,
+      "loss": 1.2628,
+      "step": 2891
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.484132362554915e-06,
+      "loss": 1.3377,
+      "step": 2892
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.477697908472925e-06,
+      "loss": 1.3242,
+      "step": 2893
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.471264569438012e-06,
+      "loss": 1.3794,
+      "step": 2894
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.464832348294195e-06,
+      "loss": 1.3754,
+      "step": 2895
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.458401247885e-06,
+      "loss": 1.329,
+      "step": 2896
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.451971271053455e-06,
+      "loss": 1.3025,
+      "step": 2897
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.445542420642097e-06,
+      "loss": 1.3305,
+      "step": 2898
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.439114699492957e-06,
+      "loss": 1.4133,
+      "step": 2899
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4326881104475725e-06,
+      "loss": 1.3345,
+      "step": 2900
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.426262656346979e-06,
+      "loss": 1.4194,
+      "step": 2901
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.419838340031709e-06,
+      "loss": 1.3446,
+      "step": 2902
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.413415164341793e-06,
+      "loss": 1.2861,
+      "step": 2903
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.406993132116759e-06,
+      "loss": 1.3928,
+      "step": 2904
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.400572246195628e-06,
+      "loss": 1.3791,
+      "step": 2905
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.394152509416911e-06,
+      "loss": 1.3002,
+      "step": 2906
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.387733924618617e-06,
+      "loss": 1.3612,
+      "step": 2907
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.381316494638243e-06,
+      "loss": 1.2954,
+      "step": 2908
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.374900222312774e-06,
+      "loss": 1.4205,
+      "step": 2909
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.368485110478685e-06,
+      "loss": 1.2581,
+      "step": 2910
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.36207116197194e-06,
+      "loss": 1.4378,
+      "step": 2911
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.355658379627981e-06,
+      "loss": 1.3514,
+      "step": 2912
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.349246766281741e-06,
+      "loss": 1.2996,
+      "step": 2913
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.342836324767639e-06,
+      "loss": 1.3611,
+      "step": 2914
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.336427057919569e-06,
+      "loss": 1.284,
+      "step": 2915
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3300189685709065e-06,
+      "loss": 1.4082,
+      "step": 2916
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.323612059554514e-06,
+      "loss": 1.2968,
+      "step": 2917
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.317206333702722e-06,
+      "loss": 1.3279,
+      "step": 2918
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.310801793847344e-06,
+      "loss": 1.3678,
+      "step": 2919
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.30439844281967e-06,
+      "loss": 1.368,
+      "step": 2920
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.29799628345046e-06,
+      "loss": 1.3673,
+      "step": 2921
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.291595318569951e-06,
+      "loss": 1.4069,
+      "step": 2922
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.285195551007848e-06,
+      "loss": 1.3966,
+      "step": 2923
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.27879698359333e-06,
+      "loss": 1.2521,
+      "step": 2924
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.272399619155048e-06,
+      "loss": 1.3253,
+      "step": 2925
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.266003460521116e-06,
+      "loss": 1.3018,
+      "step": 2926
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.259608510519121e-06,
+      "loss": 1.3304,
+      "step": 2927
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2532147719761e-06,
+      "loss": 1.2389,
+      "step": 2928
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.246822247718574e-06,
+      "loss": 1.3187,
+      "step": 2929
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.240430940572519e-06,
+      "loss": 1.3685,
+      "step": 2930
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.234040853363371e-06,
+      "loss": 1.3876,
+      "step": 2931
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.227651988916032e-06,
+      "loss": 1.2823,
+      "step": 2932
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.221264350054855e-06,
+      "loss": 1.3711,
+      "step": 2933
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.214877939603659e-06,
+      "loss": 1.3676,
+      "step": 2934
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.208492760385719e-06,
+      "loss": 1.4052,
+      "step": 2935
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.202108815223763e-06,
+      "loss": 1.3691,
+      "step": 2936
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1957261069399745e-06,
+      "loss": 1.3669,
+      "step": 2937
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.189344638355987e-06,
+      "loss": 1.3414,
+      "step": 2938
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.182964412292894e-06,
+      "loss": 1.3816,
+      "step": 2939
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.176585431571235e-06,
+      "loss": 1.3887,
+      "step": 2940
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.170207699010996e-06,
+      "loss": 1.3027,
+      "step": 2941
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.163831217431615e-06,
+      "loss": 1.302,
+      "step": 2942
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.157455989651976e-06,
+      "loss": 1.358,
+      "step": 2943
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.151082018490413e-06,
+      "loss": 1.3864,
+      "step": 2944
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.144709306764692e-06,
+      "loss": 1.379,
+      "step": 2945
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1383378572920344e-06,
+      "loss": 1.298,
+      "step": 2946
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.131967672889101e-06,
+      "loss": 1.4433,
+      "step": 2947
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.12559875637199e-06,
+      "loss": 1.3341,
+      "step": 2948
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.119231110556241e-06,
+      "loss": 1.324,
+      "step": 2949
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.112864738256835e-06,
+      "loss": 1.2821,
+      "step": 2950
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.1064996422881825e-06,
+      "loss": 1.3478,
+      "step": 2951
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.100135825464138e-06,
+      "loss": 1.3444,
+      "step": 2952
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0937732905979875e-06,
+      "loss": 1.4211,
+      "step": 2953
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.087412040502446e-06,
+      "loss": 1.3235,
+      "step": 2954
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.081052077989668e-06,
+      "loss": 1.2945,
+      "step": 2955
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.07469340587123e-06,
+      "loss": 1.3376,
+      "step": 2956
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0683360269581465e-06,
+      "loss": 1.2522,
+      "step": 2957
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.061979944060856e-06,
+      "loss": 1.3815,
+      "step": 2958
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0556251599892245e-06,
+      "loss": 1.2606,
+      "step": 2959
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.049271677552545e-06,
+      "loss": 1.3326,
+      "step": 2960
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.042919499559538e-06,
+      "loss": 1.2791,
+      "step": 2961
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.036568628818332e-06,
+      "loss": 1.3459,
+      "step": 2962
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.030219068136499e-06,
+      "loss": 1.334,
+      "step": 2963
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.023870820321017e-06,
+      "loss": 1.3585,
+      "step": 2964
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.017523888178291e-06,
+      "loss": 1.3591,
+      "step": 2965
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.011178274514142e-06,
+      "loss": 1.3064,
+      "step": 2966
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.004833982133808e-06,
+      "loss": 1.3229,
+      "step": 2967
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9984910138419434e-06,
+      "loss": 1.3601,
+      "step": 2968
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.992149372442617e-06,
+      "loss": 1.3712,
+      "step": 2969
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.98580906073931e-06,
+      "loss": 1.3189,
+      "step": 2970
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.979470081534916e-06,
+      "loss": 1.3793,
+      "step": 2971
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.973132437631743e-06,
+      "loss": 1.2831,
+      "step": 2972
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.966796131831501e-06,
+      "loss": 1.2748,
+      "step": 2973
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.960461166935317e-06,
+      "loss": 1.3902,
+      "step": 2974
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9541275457437215e-06,
+      "loss": 1.2848,
+      "step": 2975
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.947795271056651e-06,
+      "loss": 1.3006,
+      "step": 2976
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.94146434567345e-06,
+      "loss": 1.4037,
+      "step": 2977
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.935134772392853e-06,
+      "loss": 1.4364,
+      "step": 2978
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.928806554013016e-06,
+      "loss": 1.2984,
+      "step": 2979
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.922479693331483e-06,
+      "loss": 1.3376,
+      "step": 2980
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.916154193145206e-06,
+      "loss": 1.3063,
+      "step": 2981
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.909830056250527e-06,
+      "loss": 1.3292,
+      "step": 2982
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.9035072854431915e-06,
+      "loss": 1.3335,
+      "step": 2983
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.897185883518341e-06,
+      "loss": 1.402,
+      "step": 2984
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8908658532705085e-06,
+      "loss": 1.3743,
+      "step": 2985
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.884547197493623e-06,
+      "loss": 1.4265,
+      "step": 2986
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.878229918981003e-06,
+      "loss": 1.3681,
+      "step": 2987
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.871914020525364e-06,
+      "loss": 1.317,
+      "step": 2988
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.865599504918805e-06,
+      "loss": 1.283,
+      "step": 2989
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8592863749528185e-06,
+      "loss": 1.3442,
+      "step": 2990
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.852974633418281e-06,
+      "loss": 1.2939,
+      "step": 2991
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.846664283105455e-06,
+      "loss": 1.391,
+      "step": 2992
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.840355326803993e-06,
+      "loss": 1.3289,
+      "step": 2993
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.834047767302929e-06,
+      "loss": 1.2508,
+      "step": 2994
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8277416073906725e-06,
+      "loss": 1.3456,
+      "step": 2995
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.821436849855023e-06,
+      "loss": 1.3142,
+      "step": 2996
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.815133497483157e-06,
+      "loss": 1.2941,
+      "step": 2997
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.808831553061631e-06,
+      "loss": 1.376,
+      "step": 2998
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.802531019376377e-06,
+      "loss": 1.3037,
+      "step": 2999
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.796231899212704e-06,
+      "loss": 1.3802,
+      "step": 3000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.789934195355299e-06,
+      "loss": 1.3474,
+      "step": 3001
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.783637910588216e-06,
+      "loss": 1.3742,
+      "step": 3002
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.777343047694891e-06,
+      "loss": 1.3217,
+      "step": 3003
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.771049609458123e-06,
+      "loss": 1.3241,
+      "step": 3004
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.764757598660084e-06,
+      "loss": 1.3746,
+      "step": 3005
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.75846701808232e-06,
+      "loss": 1.3615,
+      "step": 3006
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.752177870505736e-06,
+      "loss": 1.3592,
+      "step": 3007
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.74589015871061e-06,
+      "loss": 1.242,
+      "step": 3008
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7396038854765825e-06,
+      "loss": 1.3861,
+      "step": 3009
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.733319053582659e-06,
+      "loss": 1.3061,
+      "step": 3010
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.727035665807209e-06,
+      "loss": 1.4539,
+      "step": 3011
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.720753724927957e-06,
+      "loss": 1.392,
+      "step": 3012
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.714473233721999e-06,
+      "loss": 1.2796,
+      "step": 3013
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.7081941949657805e-06,
+      "loss": 1.3267,
+      "step": 3014
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.701916611435109e-06,
+      "loss": 1.4095,
+      "step": 3015
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6956404859051505e-06,
+      "loss": 1.3982,
+      "step": 3016
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.689365821150421e-06,
+      "loss": 1.3689,
+      "step": 3017
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.683092619944796e-06,
+      "loss": 1.2933,
+      "step": 3018
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.676820885061499e-06,
+      "loss": 1.271,
+      "step": 3019
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.670550619273112e-06,
+      "loss": 1.2865,
+      "step": 3020
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.664281825351561e-06,
+      "loss": 1.3239,
+      "step": 3021
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6580145060681255e-06,
+      "loss": 1.3103,
+      "step": 3022
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6517486641934315e-06,
+      "loss": 1.2957,
+      "step": 3023
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.645484302497452e-06,
+      "loss": 1.2818,
+      "step": 3024
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.639221423749502e-06,
+      "loss": 1.3537,
+      "step": 3025
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.632960030718251e-06,
+      "loss": 1.3834,
+      "step": 3026
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6267001261717015e-06,
+      "loss": 1.2778,
+      "step": 3027
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.620441712877203e-06,
+      "loss": 1.4349,
+      "step": 3028
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.614184793601443e-06,
+      "loss": 1.2929,
+      "step": 3029
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.607929371110449e-06,
+      "loss": 1.3328,
+      "step": 3030
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.601675448169591e-06,
+      "loss": 1.3576,
+      "step": 3031
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.595423027543572e-06,
+      "loss": 1.372,
+      "step": 3032
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.589172111996431e-06,
+      "loss": 1.3069,
+      "step": 3033
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5829227042915414e-06,
+      "loss": 1.3796,
+      "step": 3034
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.576674807191614e-06,
+      "loss": 1.3747,
+      "step": 3035
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.570428423458687e-06,
+      "loss": 1.3785,
+      "step": 3036
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5641835558541314e-06,
+      "loss": 1.354,
+      "step": 3037
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5579402071386485e-06,
+      "loss": 1.3552,
+      "step": 3038
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.551698380072266e-06,
+      "loss": 1.3784,
+      "step": 3039
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.545458077414344e-06,
+      "loss": 1.2755,
+      "step": 3040
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.539219301923563e-06,
+      "loss": 1.3054,
+      "step": 3041
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.532982056357928e-06,
+      "loss": 1.3274,
+      "step": 3042
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5267463434747745e-06,
+      "loss": 1.292,
+      "step": 3043
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.520512166030762e-06,
+      "loss": 1.4054,
+      "step": 3044
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5142795267818505e-06,
+      "loss": 1.3356,
+      "step": 3045
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.508048428483344e-06,
+      "loss": 1.2892,
+      "step": 3046
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.501818873889856e-06,
+      "loss": 1.3735,
+      "step": 3047
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.495590865755315e-06,
+      "loss": 1.3225,
+      "step": 3048
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.489364406832971e-06,
+      "loss": 1.3694,
+      "step": 3049
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.483139499875386e-06,
+      "loss": 1.3798,
+      "step": 3050
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.476916147634437e-06,
+      "loss": 1.4099,
+      "step": 3051
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4706943528613135e-06,
+      "loss": 1.4158,
+      "step": 3052
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.464474118306515e-06,
+      "loss": 1.4447,
+      "step": 3053
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.458255446719854e-06,
+      "loss": 1.3206,
+      "step": 3054
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.452038340850451e-06,
+      "loss": 1.3389,
+      "step": 3055
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.445822803446731e-06,
+      "loss": 1.4211,
+      "step": 3056
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.439608837256432e-06,
+      "loss": 1.3085,
+      "step": 3057
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.433396445026594e-06,
+      "loss": 1.3453,
+      "step": 3058
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.427185629503561e-06,
+      "loss": 1.4144,
+      "step": 3059
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.420976393432977e-06,
+      "loss": 1.3237,
+      "step": 3060
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.414768739559799e-06,
+      "loss": 1.3197,
+      "step": 3061
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.408562670628267e-06,
+      "loss": 1.4558,
+      "step": 3062
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.4023581893819345e-06,
+      "loss": 1.3758,
+      "step": 3063
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.396155298563646e-06,
+      "loss": 1.3342,
+      "step": 3064
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.389954000915548e-06,
+      "loss": 1.3053,
+      "step": 3065
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.383754299179079e-06,
+      "loss": 1.3409,
+      "step": 3066
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.377556196094974e-06,
+      "loss": 1.3316,
+      "step": 3067
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.37135969440326e-06,
+      "loss": 1.3351,
+      "step": 3068
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.365164796843253e-06,
+      "loss": 1.3298,
+      "step": 3069
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.358971506153568e-06,
+      "loss": 1.2752,
+      "step": 3070
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.352779825072105e-06,
+      "loss": 1.3313,
+      "step": 3071
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.34658975633605e-06,
+      "loss": 1.3115,
+      "step": 3072
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.340401302681879e-06,
+      "loss": 1.4146,
+      "step": 3073
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.334214466845354e-06,
+      "loss": 1.3597,
+      "step": 3074
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.328029251561521e-06,
+      "loss": 1.3537,
+      "step": 3075
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.321845659564709e-06,
+      "loss": 1.3218,
+      "step": 3076
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3156636935885344e-06,
+      "loss": 1.3121,
+      "step": 3077
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.30948335636589e-06,
+      "loss": 1.3335,
+      "step": 3078
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.303304650628943e-06,
+      "loss": 1.3843,
+      "step": 3079
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.29712757910915e-06,
+      "loss": 1.3665,
+      "step": 3080
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.290952144537241e-06,
+      "loss": 1.347,
+      "step": 3081
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.284778349643221e-06,
+      "loss": 1.1857,
+      "step": 3082
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2786061971563714e-06,
+      "loss": 1.3488,
+      "step": 3083
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.272435689805249e-06,
+      "loss": 1.4492,
+      "step": 3084
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.266266830317683e-06,
+      "loss": 1.4205,
+      "step": 3085
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.26009962142077e-06,
+      "loss": 1.3258,
+      "step": 3086
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.25393406584088e-06,
+      "loss": 1.303,
+      "step": 3087
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.247770166303653e-06,
+      "loss": 1.1825,
+      "step": 3088
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2416079255339945e-06,
+      "loss": 1.3115,
+      "step": 3089
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.23544734625608e-06,
+      "loss": 1.3868,
+      "step": 3090
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.229288431193348e-06,
+      "loss": 1.3222,
+      "step": 3091
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2231311830684995e-06,
+      "loss": 1.3628,
+      "step": 3092
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.216975604603503e-06,
+      "loss": 1.3337,
+      "step": 3093
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.210821698519592e-06,
+      "loss": 1.3404,
+      "step": 3094
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.204669467537243e-06,
+      "loss": 1.2843,
+      "step": 3095
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.198518914376211e-06,
+      "loss": 1.3736,
+      "step": 3096
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.192370041755505e-06,
+      "loss": 1.3259,
+      "step": 3097
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.186222852393386e-06,
+      "loss": 1.3499,
+      "step": 3098
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.180077349007376e-06,
+      "loss": 1.3434,
+      "step": 3099
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.173933534314247e-06,
+      "loss": 1.3324,
+      "step": 3100
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.167791411030027e-06,
+      "loss": 1.4288,
+      "step": 3101
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1616509818699975e-06,
+      "loss": 1.3092,
+      "step": 3102
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.15551224954869e-06,
+      "loss": 1.2669,
+      "step": 3103
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1493752167798825e-06,
+      "loss": 1.2497,
+      "step": 3104
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.143239886276606e-06,
+      "loss": 1.3019,
+      "step": 3105
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.13710626075114e-06,
+      "loss": 1.3271,
+      "step": 3106
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1309743429150045e-06,
+      "loss": 1.3519,
+      "step": 3107
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.124844135478971e-06,
+      "loss": 1.2922,
+      "step": 3108
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.118715641153049e-06,
+      "loss": 1.2968,
+      "step": 3109
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1125888626464955e-06,
+      "loss": 1.2808,
+      "step": 3110
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.106463802667811e-06,
+      "loss": 1.392,
+      "step": 3111
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.1003404639247234e-06,
+      "loss": 1.38,
+      "step": 3112
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.094218849124212e-06,
+      "loss": 1.3927,
+      "step": 3113
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.088098960972494e-06,
+      "loss": 1.3278,
+      "step": 3114
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.081980802175016e-06,
+      "loss": 1.292,
+      "step": 3115
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0758643754364665e-06,
+      "loss": 1.3473,
+      "step": 3116
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.069749683460765e-06,
+      "loss": 1.3528,
+      "step": 3117
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.063636728951065e-06,
+      "loss": 1.3755,
+      "step": 3118
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.05752551460975e-06,
+      "loss": 1.331,
+      "step": 3119
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.051416043138438e-06,
+      "loss": 1.3169,
+      "step": 3120
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.045308317237975e-06,
+      "loss": 1.326,
+      "step": 3121
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.039202339608432e-06,
+      "loss": 1.3202,
+      "step": 3122
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.033098112949114e-06,
+      "loss": 1.369,
+      "step": 3123
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.026995639958544e-06,
+      "loss": 1.3401,
+      "step": 3124
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.020894923334476e-06,
+      "loss": 1.2449,
+      "step": 3125
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.014795965773884e-06,
+      "loss": 1.3139,
+      "step": 3126
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.008698769972967e-06,
+      "loss": 1.3735,
+      "step": 3127
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.002603338627144e-06,
+      "loss": 1.3858,
+      "step": 3128
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.996509674431053e-06,
+      "loss": 1.3457,
+      "step": 3129
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.990417780078549e-06,
+      "loss": 1.2865,
+      "step": 3130
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.98432765826271e-06,
+      "loss": 1.3036,
+      "step": 3131
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.978239311675826e-06,
+      "loss": 1.3415,
+      "step": 3132
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.972152743009406e-06,
+      "loss": 1.3465,
+      "step": 3133
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.966067954954169e-06,
+      "loss": 1.2487,
+      "step": 3134
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.959984950200048e-06,
+      "loss": 1.2923,
+      "step": 3135
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.953903731436191e-06,
+      "loss": 1.3059,
+      "step": 3136
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.94782430135095e-06,
+      "loss": 1.3656,
+      "step": 3137
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.941746662631893e-06,
+      "loss": 1.4605,
+      "step": 3138
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9356708179657905e-06,
+      "loss": 1.3525,
+      "step": 3139
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.929596770038626e-06,
+      "loss": 1.3139,
+      "step": 3140
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.923524521535582e-06,
+      "loss": 1.3384,
+      "step": 3141
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.917454075141049e-06,
+      "loss": 1.3494,
+      "step": 3142
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.911385433538621e-06,
+      "loss": 1.3019,
+      "step": 3143
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.9053185994110975e-06,
+      "loss": 1.2902,
+      "step": 3144
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.899253575440469e-06,
+      "loss": 1.3368,
+      "step": 3145
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.893190364307933e-06,
+      "loss": 1.2903,
+      "step": 3146
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.887128968693887e-06,
+      "loss": 1.3103,
+      "step": 3147
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.881069391277922e-06,
+      "loss": 1.3059,
+      "step": 3148
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8750116347388255e-06,
+      "loss": 1.3058,
+      "step": 3149
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.868955701754584e-06,
+      "loss": 1.4183,
+      "step": 3150
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.862901595002373e-06,
+      "loss": 1.3579,
+      "step": 3151
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8568493171585625e-06,
+      "loss": 1.4056,
+      "step": 3152
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.850798870898715e-06,
+      "loss": 1.2906,
+      "step": 3153
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.844750258897582e-06,
+      "loss": 1.2881,
+      "step": 3154
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.838703483829107e-06,
+      "loss": 1.3344,
+      "step": 3155
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.832658548366412e-06,
+      "loss": 1.4042,
+      "step": 3156
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8266154551818225e-06,
+      "loss": 1.3459,
+      "step": 3157
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.82057420694683e-06,
+      "loss": 1.3977,
+      "step": 3158
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8145348063321295e-06,
+      "loss": 1.3986,
+      "step": 3159
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8084972560075815e-06,
+      "loss": 1.3556,
+      "step": 3160
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.80246155864225e-06,
+      "loss": 1.3201,
+      "step": 3161
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.796427716904347e-06,
+      "loss": 1.3106,
+      "step": 3162
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.790395733461297e-06,
+      "loss": 1.3114,
+      "step": 3163
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.784365610979692e-06,
+      "loss": 1.261,
+      "step": 3164
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.778337352125289e-06,
+      "loss": 1.329,
+      "step": 3165
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.772310959563043e-06,
+      "loss": 1.2432,
+      "step": 3166
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.766286435957063e-06,
+      "loss": 1.3617,
+      "step": 3167
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.760263783970647e-06,
+      "loss": 1.3611,
+      "step": 3168
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.754243006266256e-06,
+      "loss": 1.4051,
+      "step": 3169
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.748224105505532e-06,
+      "loss": 1.238,
+      "step": 3170
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.742207084349274e-06,
+      "loss": 1.3538,
+      "step": 3171
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.736191945457463e-06,
+      "loss": 1.4064,
+      "step": 3172
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7301786914892374e-06,
+      "loss": 1.2979,
+      "step": 3173
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7241673251029155e-06,
+      "loss": 1.3503,
+      "step": 3174
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.718157848955963e-06,
+      "loss": 1.319,
+      "step": 3175
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.712150265705028e-06,
+      "loss": 1.2536,
+      "step": 3176
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.706144578005908e-06,
+      "loss": 1.3507,
+      "step": 3177
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.700140788513575e-06,
+      "loss": 1.3253,
+      "step": 3178
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.694138899882151e-06,
+      "loss": 1.3673,
+      "step": 3179
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.688138914764918e-06,
+      "loss": 1.3612,
+      "step": 3180
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.682140835814327e-06,
+      "loss": 1.29,
+      "step": 3181
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6761446656819745e-06,
+      "loss": 1.3784,
+      "step": 3182
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.670150407018624e-06,
+      "loss": 1.2727,
+      "step": 3183
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.664158062474181e-06,
+      "loss": 1.3134,
+      "step": 3184
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.65816763469772e-06,
+      "loss": 1.3143,
+      "step": 3185
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6521791263374534e-06,
+      "loss": 1.31,
+      "step": 3186
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.646192540040758e-06,
+      "loss": 1.3135,
+      "step": 3187
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.640207878454149e-06,
+      "loss": 1.261,
+      "step": 3188
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.634225144223302e-06,
+      "loss": 1.3402,
+      "step": 3189
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.62824433999303e-06,
+      "loss": 1.337,
+      "step": 3190
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.622265468407302e-06,
+      "loss": 1.3738,
+      "step": 3191
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.616288532109225e-06,
+      "loss": 1.2841,
+      "step": 3192
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.61031353374106e-06,
+      "loss": 1.3633,
+      "step": 3193
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.604340475944196e-06,
+      "loss": 1.3192,
+      "step": 3194
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.598369361359184e-06,
+      "loss": 1.3386,
+      "step": 3195
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5924001926257e-06,
+      "loss": 1.3379,
+      "step": 3196
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.586432972382561e-06,
+      "loss": 1.3499,
+      "step": 3197
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.580467703267736e-06,
+      "loss": 1.3691,
+      "step": 3198
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.574504387918311e-06,
+      "loss": 1.3379,
+      "step": 3199
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.568543028970532e-06,
+      "loss": 1.3346,
+      "step": 3200
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.562583629059757e-06,
+      "loss": 1.3245,
+      "step": 3201
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.556626190820497e-06,
+      "loss": 1.3007,
+      "step": 3202
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5506707168863795e-06,
+      "loss": 1.3338,
+      "step": 3203
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5447172098901805e-06,
+      "loss": 1.3757,
+      "step": 3204
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.538765672463786e-06,
+      "loss": 1.3659,
+      "step": 3205
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5328161072382355e-06,
+      "loss": 1.4095,
+      "step": 3206
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.526868516843673e-06,
+      "loss": 1.3751,
+      "step": 3207
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.52092290390939e-06,
+      "loss": 1.3547,
+      "step": 3208
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.514979271063785e-06,
+      "loss": 1.2757,
+      "step": 3209
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.509037620934399e-06,
+      "loss": 1.2956,
+      "step": 3210
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.503097956147884e-06,
+      "loss": 1.3164,
+      "step": 3211
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4971602793300134e-06,
+      "loss": 1.3185,
+      "step": 3212
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.491224593105695e-06,
+      "loss": 1.3333,
+      "step": 3213
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4852909000989405e-06,
+      "loss": 1.3441,
+      "step": 3214
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.479359202932897e-06,
+      "loss": 1.4571,
+      "step": 3215
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4734295042298105e-06,
+      "loss": 1.3257,
+      "step": 3216
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.467501806611062e-06,
+      "loss": 1.3497,
+      "step": 3217
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.461576112697132e-06,
+      "loss": 1.3117,
+      "step": 3218
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.455652425107631e-06,
+      "loss": 1.3636,
+      "step": 3219
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.449730746461264e-06,
+      "loss": 1.3563,
+      "step": 3220
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.443811079375867e-06,
+      "loss": 1.2475,
+      "step": 3221
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.43789342646837e-06,
+      "loss": 1.3472,
+      "step": 3222
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.43197779035483e-06,
+      "loss": 1.3314,
+      "step": 3223
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.426064173650392e-06,
+      "loss": 1.308,
+      "step": 3224
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.420152578969327e-06,
+      "loss": 1.3642,
+      "step": 3225
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.414243008924997e-06,
+      "loss": 1.3431,
+      "step": 3226
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.4083354661298816e-06,
+      "loss": 1.3665,
+      "step": 3227
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.4024299531955645e-06,
+      "loss": 1.3793,
+      "step": 3228
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.396526472732711e-06,
+      "loss": 1.2731,
+      "step": 3229
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3906250273511154e-06,
+      "loss": 1.3226,
+      "step": 3230
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.38472561965965e-06,
+      "loss": 1.3765,
+      "step": 3231
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3788282522663085e-06,
+      "loss": 1.3086,
+      "step": 3232
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3729329277781585e-06,
+      "loss": 1.4264,
+      "step": 3233
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.367039648801386e-06,
+      "loss": 1.3768,
+      "step": 3234
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.361148417941255e-06,
+      "loss": 1.3639,
+      "step": 3235
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3552592378021415e-06,
+      "loss": 1.3796,
+      "step": 3236
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.349372110987496e-06,
+      "loss": 1.3163,
+      "step": 3237
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.34348704009988e-06,
+      "loss": 1.3156,
+      "step": 3238
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.337604027740928e-06,
+      "loss": 1.3105,
+      "step": 3239
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.33172307651138e-06,
+      "loss": 1.2918,
+      "step": 3240
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.325844189011058e-06,
+      "loss": 1.3232,
+      "step": 3241
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.319967367838868e-06,
+      "loss": 1.3628,
+      "step": 3242
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.314092615592814e-06,
+      "loss": 1.2999,
+      "step": 3243
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.308219934869969e-06,
+      "loss": 1.4364,
+      "step": 3244
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3023493282665115e-06,
+      "loss": 1.3237,
+      "step": 3245
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.296480798377674e-06,
+      "loss": 1.2859,
+      "step": 3246
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.290614347797802e-06,
+      "loss": 1.357,
+      "step": 3247
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.284749979120299e-06,
+      "loss": 1.3359,
+      "step": 3248
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.278887694937662e-06,
+      "loss": 1.2917,
+      "step": 3249
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2730274978414555e-06,
+      "loss": 1.1378,
+      "step": 3250
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.267169390422334e-06,
+      "loss": 1.209,
+      "step": 3251
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2613133752700145e-06,
+      "loss": 1.0321,
+      "step": 3252
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.255459454973301e-06,
+      "loss": 1.125,
+      "step": 3253
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.249607632120064e-06,
+      "loss": 1.1145,
+      "step": 3254
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.243757909297247e-06,
+      "loss": 1.0625,
+      "step": 3255
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.237910289090873e-06,
+      "loss": 1.0987,
+      "step": 3256
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.232064774086022e-06,
+      "loss": 1.0578,
+      "step": 3257
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.226221366866859e-06,
+      "loss": 1.1266,
+      "step": 3258
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.220380070016601e-06,
+      "loss": 1.1482,
+      "step": 3259
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.214540886117547e-06,
+      "loss": 1.0773,
+      "step": 3260
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.208703817751053e-06,
+      "loss": 1.1095,
+      "step": 3261
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.202868867497542e-06,
+      "loss": 0.9985,
+      "step": 3262
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.197036037936494e-06,
+      "loss": 1.0481,
+      "step": 3263
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1912053316464694e-06,
+      "loss": 1.11,
+      "step": 3264
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.185376751205067e-06,
+      "loss": 1.1117,
+      "step": 3265
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.179550299188968e-06,
+      "loss": 1.0155,
+      "step": 3266
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1737259781738934e-06,
+      "loss": 1.1121,
+      "step": 3267
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.167903790734634e-06,
+      "loss": 1.1508,
+      "step": 3268
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.162083739445038e-06,
+      "loss": 1.0962,
+      "step": 3269
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.156265826878e-06,
+      "loss": 1.0492,
+      "step": 3270
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.150450055605479e-06,
+      "loss": 1.1382,
+      "step": 3271
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.144636428198477e-06,
+      "loss": 1.0714,
+      "step": 3272
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.138824947227063e-06,
+      "loss": 1.0628,
+      "step": 3273
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.133015615260339e-06,
+      "loss": 1.1931,
+      "step": 3274
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.1272084348664754e-06,
+      "loss": 1.1115,
+      "step": 3275
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.121403408612672e-06,
+      "loss": 1.1041,
+      "step": 3276
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.115600539065197e-06,
+      "loss": 1.1117,
+      "step": 3277
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.10979982878935e-06,
+      "loss": 1.0369,
+      "step": 3278
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.10400128034948e-06,
+      "loss": 1.0386,
+      "step": 3279
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.098204896308978e-06,
+      "loss": 1.0825,
+      "step": 3280
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.092410679230286e-06,
+      "loss": 1.046,
+      "step": 3281
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.086618631674888e-06,
+      "loss": 1.1081,
+      "step": 3282
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.080828756203294e-06,
+      "loss": 1.0935,
+      "step": 3283
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.075041055375074e-06,
+      "loss": 1.1276,
+      "step": 3284
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0692555317488166e-06,
+      "loss": 1.0655,
+      "step": 3285
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.063472187882168e-06,
+      "loss": 1.0657,
+      "step": 3286
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.057691026331792e-06,
+      "loss": 1.0327,
+      "step": 3287
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0519120496534044e-06,
+      "loss": 1.0154,
+      "step": 3288
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.046135260401739e-06,
+      "loss": 1.1195,
+      "step": 3289
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.04036066113058e-06,
+      "loss": 1.0238,
+      "step": 3290
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.034588254392725e-06,
+      "loss": 1.1033,
+      "step": 3291
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0288180427400205e-06,
+      "loss": 1.0021,
+      "step": 3292
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0230500287233244e-06,
+      "loss": 1.0182,
+      "step": 3293
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.017284214892541e-06,
+      "loss": 1.1205,
+      "step": 3294
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.011520603796591e-06,
+      "loss": 1.0591,
+      "step": 3295
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.005759197983418e-06,
+      "loss": 1.0197,
+      "step": 3296
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 1.0707,
+      "step": 3297
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.994243012392338e-06,
+      "loss": 1.0977,
+      "step": 3298
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.988488237705452e-06,
+      "loss": 1.0616,
+      "step": 3299
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.98273567848338e-06,
+      "loss": 1.1782,
+      "step": 3300
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9769853372691934e-06,
+      "loss": 1.1339,
+      "step": 3301
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.971237216604967e-06,
+      "loss": 1.0653,
+      "step": 3302
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.96549131903181e-06,
+      "loss": 1.0832,
+      "step": 3303
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.959747647089833e-06,
+      "loss": 1.0559,
+      "step": 3304
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.954006203318179e-06,
+      "loss": 1.0535,
+      "step": 3305
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9482669902549896e-06,
+      "loss": 1.0632,
+      "step": 3306
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.942530010437435e-06,
+      "loss": 1.0288,
+      "step": 3307
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.936795266401686e-06,
+      "loss": 1.0908,
+      "step": 3308
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.931062760682935e-06,
+      "loss": 1.1272,
+      "step": 3309
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.925332495815376e-06,
+      "loss": 1.1202,
+      "step": 3310
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.919604474332223e-06,
+      "loss": 1.0664,
+      "step": 3311
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9138786987656865e-06,
+      "loss": 1.0831,
+      "step": 3312
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9081551716469875e-06,
+      "loss": 1.1266,
+      "step": 3313
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.902433895506362e-06,
+      "loss": 1.0873,
+      "step": 3314
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.896714872873038e-06,
+      "loss": 1.0937,
+      "step": 3315
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8909981062752586e-06,
+      "loss": 1.0845,
+      "step": 3316
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.885283598240259e-06,
+      "loss": 1.0582,
+      "step": 3317
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.879571351294287e-06,
+      "loss": 1.105,
+      "step": 3318
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.873861367962578e-06,
+      "loss": 1.1221,
+      "step": 3319
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.868153650769381e-06,
+      "loss": 1.0543,
+      "step": 3320
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8624482022379286e-06,
+      "loss": 1.1154,
+      "step": 3321
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.856745024890466e-06,
+      "loss": 1.0298,
+      "step": 3322
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.851044121248217e-06,
+      "loss": 1.0729,
+      "step": 3323
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.845345493831419e-06,
+      "loss": 1.0592,
+      "step": 3324
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8396491451592855e-06,
+      "loss": 1.1121,
+      "step": 3325
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.833955077750038e-06,
+      "loss": 1.045,
+      "step": 3326
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8282632941208725e-06,
+      "loss": 1.0779,
+      "step": 3327
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.822573796787997e-06,
+      "loss": 1.2065,
+      "step": 3328
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.81688658826659e-06,
+      "loss": 1.1033,
+      "step": 3329
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.811201671070822e-06,
+      "loss": 1.0951,
+      "step": 3330
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.805519047713862e-06,
+      "loss": 1.1879,
+      "step": 3331
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.799838720707847e-06,
+      "loss": 1.069,
+      "step": 3332
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.794160692563917e-06,
+      "loss": 1.1378,
+      "step": 3333
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.78848496579218e-06,
+      "loss": 1.0554,
+      "step": 3334
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.782811542901742e-06,
+      "loss": 1.0676,
+      "step": 3335
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.777140426400674e-06,
+      "loss": 1.1179,
+      "step": 3336
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.771471618796043e-06,
+      "loss": 1.037,
+      "step": 3337
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.765805122593879e-06,
+      "loss": 1.0421,
+      "step": 3338
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.76014094029921e-06,
+      "loss": 0.9662,
+      "step": 3339
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.754479074416021e-06,
+      "loss": 1.0393,
+      "step": 3340
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.748819527447289e-06,
+      "loss": 1.0922,
+      "step": 3341
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.743162301894952e-06,
+      "loss": 1.1213,
+      "step": 3342
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.737507400259939e-06,
+      "loss": 1.1026,
+      "step": 3343
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.731854825042129e-06,
+      "loss": 1.149,
+      "step": 3344
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.726204578740396e-06,
+      "loss": 1.1144,
+      "step": 3345
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.720556663852569e-06,
+      "loss": 1.1154,
+      "step": 3346
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.714911082875446e-06,
+      "loss": 1.0777,
+      "step": 3347
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.709267838304808e-06,
+      "loss": 1.0626,
+      "step": 3348
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.703626932635383e-06,
+      "loss": 1.103,
+      "step": 3349
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.697988368360884e-06,
+      "loss": 1.0216,
+      "step": 3350
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.692352147973973e-06,
+      "loss": 1.0656,
+      "step": 3351
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.686718273966291e-06,
+      "loss": 1.0315,
+      "step": 3352
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.681086748828424e-06,
+      "loss": 1.0364,
+      "step": 3353
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.675457575049939e-06,
+      "loss": 1.0434,
+      "step": 3354
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.669830755119345e-06,
+      "loss": 1.1107,
+      "step": 3355
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.664206291524126e-06,
+      "loss": 1.1651,
+      "step": 3356
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.658584186750713e-06,
+      "loss": 1.0598,
+      "step": 3357
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6529644432845e-06,
+      "loss": 1.1047,
+      "step": 3358
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6473470636098405e-06,
+      "loss": 1.0905,
+      "step": 3359
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.641732050210032e-06,
+      "loss": 1.0735,
+      "step": 3360
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.636119405567339e-06,
+      "loss": 1.1138,
+      "step": 3361
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.630509132162967e-06,
+      "loss": 1.0382,
+      "step": 3362
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.62490123247708e-06,
+      "loss": 1.0546,
+      "step": 3363
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.619295708988788e-06,
+      "loss": 1.0879,
+      "step": 3364
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.61369256417616e-06,
+      "loss": 1.0631,
+      "step": 3365
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6080918005162e-06,
+      "loss": 1.0479,
+      "step": 3366
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6024934204848745e-06,
+      "loss": 1.1018,
+      "step": 3367
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.596897426557079e-06,
+      "loss": 1.0659,
+      "step": 3368
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.591303821206673e-06,
+      "loss": 1.121,
+      "step": 3369
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.585712606906441e-06,
+      "loss": 1.1281,
+      "step": 3370
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.580123786128129e-06,
+      "loss": 1.0553,
+      "step": 3371
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5745373613424075e-06,
+      "loss": 1.0508,
+      "step": 3372
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.568953335018899e-06,
+      "loss": 1.1577,
+      "step": 3373
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.563371709626167e-06,
+      "loss": 1.0923,
+      "step": 3374
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.557792487631704e-06,
+      "loss": 1.0649,
+      "step": 3375
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.552215671501951e-06,
+      "loss": 1.0508,
+      "step": 3376
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.546641263702271e-06,
+      "loss": 1.1316,
+      "step": 3377
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.541069266696985e-06,
+      "loss": 1.0885,
+      "step": 3378
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5354996829493185e-06,
+      "loss": 1.1017,
+      "step": 3379
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.529932514921457e-06,
+      "loss": 1.1009,
+      "step": 3380
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.524367765074499e-06,
+      "loss": 1.0786,
+      "step": 3381
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.518805435868492e-06,
+      "loss": 1.1031,
+      "step": 3382
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.513245529762394e-06,
+      "loss": 1.0065,
+      "step": 3383
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.507688049214108e-06,
+      "loss": 1.1328,
+      "step": 3384
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.502132996680452e-06,
+      "loss": 1.0862,
+      "step": 3385
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.496580374617182e-06,
+      "loss": 1.128,
+      "step": 3386
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.491030185478976e-06,
+      "loss": 1.0346,
+      "step": 3387
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4854824317194266e-06,
+      "loss": 1.1062,
+      "step": 3388
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.479937115791066e-06,
+      "loss": 1.0595,
+      "step": 3389
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.474394240145334e-06,
+      "loss": 1.0618,
+      "step": 3390
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.468853807232606e-06,
+      "loss": 1.1061,
+      "step": 3391
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.46331581950216e-06,
+      "loss": 1.0999,
+      "step": 3392
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.457780279402212e-06,
+      "loss": 1.1085,
+      "step": 3393
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.452247189379878e-06,
+      "loss": 1.0663,
+      "step": 3394
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.446716551881213e-06,
+      "loss": 1.0711,
+      "step": 3395
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.441188369351157e-06,
+      "loss": 1.0762,
+      "step": 3396
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.435662644233594e-06,
+      "loss": 1.0208,
+      "step": 3397
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.430139378971303e-06,
+      "loss": 1.0727,
+      "step": 3398
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4246185760059856e-06,
+      "loss": 1.0855,
+      "step": 3399
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.419100237778257e-06,
+      "loss": 1.0394,
+      "step": 3400
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.413584366727626e-06,
+      "loss": 1.0689,
+      "step": 3401
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.408070965292534e-06,
+      "loss": 1.1183,
+      "step": 3402
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.402560035910307e-06,
+      "loss": 1.169,
+      "step": 3403
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.397051581017201e-06,
+      "loss": 1.0345,
+      "step": 3404
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3915456030483585e-06,
+      "loss": 1.0181,
+      "step": 3405
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.386042104437842e-06,
+      "loss": 1.1506,
+      "step": 3406
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.380541087618606e-06,
+      "loss": 1.0702,
+      "step": 3407
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.375042555022518e-06,
+      "loss": 1.0997,
+      "step": 3408
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.369546509080338e-06,
+      "loss": 1.0461,
+      "step": 3409
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.364052952221737e-06,
+      "loss": 1.0493,
+      "step": 3410
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.358561886875273e-06,
+      "loss": 1.1099,
+      "step": 3411
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.353073315468417e-06,
+      "loss": 1.1344,
+      "step": 3412
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.347587240427526e-06,
+      "loss": 1.0411,
+      "step": 3413
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.342103664177856e-06,
+      "loss": 1.1037,
+      "step": 3414
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.336622589143563e-06,
+      "loss": 1.094,
+      "step": 3415
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.33114401774769e-06,
+      "loss": 1.1154,
+      "step": 3416
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.325667952412184e-06,
+      "loss": 1.1092,
+      "step": 3417
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3201943955578684e-06,
+      "loss": 1.1297,
+      "step": 3418
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.314723349604476e-06,
+      "loss": 1.0338,
+      "step": 3419
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3092548169706125e-06,
+      "loss": 1.0597,
+      "step": 3420
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.303788800073787e-06,
+      "loss": 1.1014,
+      "step": 3421
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.298325301330383e-06,
+      "loss": 1.1232,
+      "step": 3422
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.292864323155684e-06,
+      "loss": 1.0213,
+      "step": 3423
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2874058679638464e-06,
+      "loss": 1.1269,
+      "step": 3424
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.281949938167925e-06,
+      "loss": 1.0893,
+      "step": 3425
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.276496536179843e-06,
+      "loss": 1.0755,
+      "step": 3426
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.27104566441042e-06,
+      "loss": 1.042,
+      "step": 3427
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.26559732526935e-06,
+      "loss": 1.0699,
+      "step": 3428
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.260151521165204e-06,
+      "loss": 1.0203,
+      "step": 3429
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.25470825450544e-06,
+      "loss": 1.0611,
+      "step": 3430
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2492675276963904e-06,
+      "loss": 1.1201,
+      "step": 3431
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2438293431432665e-06,
+      "loss": 1.0381,
+      "step": 3432
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.238393703250151e-06,
+      "loss": 1.1494,
+      "step": 3433
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2329606104200115e-06,
+      "loss": 1.1153,
+      "step": 3434
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.227530067054676e-06,
+      "loss": 1.0232,
+      "step": 3435
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.222102075554859e-06,
+      "loss": 1.039,
+      "step": 3436
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.216676638320135e-06,
+      "loss": 1.0581,
+      "step": 3437
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.211253757748961e-06,
+      "loss": 1.064,
+      "step": 3438
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.205833436238652e-06,
+      "loss": 1.1306,
+      "step": 3439
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.200415676185403e-06,
+      "loss": 1.0554,
+      "step": 3440
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.195000479984264e-06,
+      "loss": 1.1161,
+      "step": 3441
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.189587850029169e-06,
+      "loss": 1.081,
+      "step": 3442
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.184177788712895e-06,
+      "loss": 1.0748,
+      "step": 3443
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.178770298427107e-06,
+      "loss": 1.0935,
+      "step": 3444
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.173365381562317e-06,
+      "loss": 1.1214,
+      "step": 3445
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1679630405079e-06,
+      "loss": 1.0527,
+      "step": 3446
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.162563277652104e-06,
+      "loss": 1.1158,
+      "step": 3447
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.157166095382024e-06,
+      "loss": 1.0989,
+      "step": 3448
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1517714960836264e-06,
+      "loss": 1.0912,
+      "step": 3449
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1463794821417236e-06,
+      "loss": 1.1028,
+      "step": 3450
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.140990055939997e-06,
+      "loss": 1.0693,
+      "step": 3451
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.135603219860971e-06,
+      "loss": 1.128,
+      "step": 3452
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.130218976286039e-06,
+      "loss": 1.1241,
+      "step": 3453
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.124837327595435e-06,
+      "loss": 1.0784,
+      "step": 3454
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.119458276168258e-06,
+      "loss": 1.1417,
+      "step": 3455
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.114081824382447e-06,
+      "loss": 1.0667,
+      "step": 3456
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.108707974614804e-06,
+      "loss": 1.0514,
+      "step": 3457
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.103336729240967e-06,
+      "loss": 1.1074,
+      "step": 3458
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0979680906354395e-06,
+      "loss": 1.1157,
+      "step": 3459
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.092602061171553e-06,
+      "loss": 1.0523,
+      "step": 3460
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.087238643221505e-06,
+      "loss": 1.0427,
+      "step": 3461
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0818778391563255e-06,
+      "loss": 1.0846,
+      "step": 3462
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0765196513458885e-06,
+      "loss": 1.1809,
+      "step": 3463
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.071164082158924e-06,
+      "loss": 1.0848,
+      "step": 3464
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.065811133962987e-06,
+      "loss": 1.1421,
+      "step": 3465
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.060460809124491e-06,
+      "loss": 1.0363,
+      "step": 3466
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.055113110008675e-06,
+      "loss": 1.0564,
+      "step": 3467
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.049768038979631e-06,
+      "loss": 1.0319,
+      "step": 3468
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.044425598400274e-06,
+      "loss": 0.9957,
+      "step": 3469
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.039085790632373e-06,
+      "loss": 1.0988,
+      "step": 3470
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.033748618036515e-06,
+      "loss": 1.1212,
+      "step": 3471
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.028414082972141e-06,
+      "loss": 1.0314,
+      "step": 3472
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0230821877975064e-06,
+      "loss": 1.0797,
+      "step": 3473
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.017752934869719e-06,
+      "loss": 1.0926,
+      "step": 3474
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.012426326544701e-06,
+      "loss": 1.0676,
+      "step": 3475
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0071023651772215e-06,
+      "loss": 1.0859,
+      "step": 3476
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.001781053120863e-06,
+      "loss": 1.0728,
+      "step": 3477
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.996462392728053e-06,
+      "loss": 1.0997,
+      "step": 3478
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9911463863500365e-06,
+      "loss": 1.1056,
+      "step": 3479
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.985833036336884e-06,
+      "loss": 1.1021,
+      "step": 3480
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.980522345037501e-06,
+      "loss": 1.1309,
+      "step": 3481
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.975214314799607e-06,
+      "loss": 1.116,
+      "step": 3482
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9699089479697564e-06,
+      "loss": 1.1792,
+      "step": 3483
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.964606246893313e-06,
+      "loss": 1.0717,
+      "step": 3484
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.959306213914477e-06,
+      "loss": 1.0751,
+      "step": 3485
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.954008851376252e-06,
+      "loss": 0.976,
+      "step": 3486
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9487141616204804e-06,
+      "loss": 1.113,
+      "step": 3487
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.943422146987805e-06,
+      "loss": 1.0664,
+      "step": 3488
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.938132809817699e-06,
+      "loss": 1.1186,
+      "step": 3489
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.932846152448442e-06,
+      "loss": 1.1571,
+      "step": 3490
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.927562177217137e-06,
+      "loss": 1.0237,
+      "step": 3491
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.922280886459701e-06,
+      "loss": 1.0322,
+      "step": 3492
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.917002282510854e-06,
+      "loss": 1.0318,
+      "step": 3493
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.911726367704146e-06,
+      "loss": 1.0908,
+      "step": 3494
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.90645314437192e-06,
+      "loss": 1.1048,
+      "step": 3495
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.901182614845339e-06,
+      "loss": 1.1629,
+      "step": 3496
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.89591478145437e-06,
+      "loss": 1.0677,
+      "step": 3497
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.890649646527798e-06,
+      "loss": 1.1078,
+      "step": 3498
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.8853872123932e-06,
+      "loss": 1.0588,
+      "step": 3499
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.880127481376975e-06,
+      "loss": 1.0732,
+      "step": 3500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.874870455804314e-06,
+      "loss": 1.0386,
+      "step": 3501
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8696161379992225e-06,
+      "loss": 1.0517,
+      "step": 3502
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.864364530284498e-06,
+      "loss": 1.0896,
+      "step": 3503
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.859115634981748e-06,
+      "loss": 1.0975,
+      "step": 3504
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.853869454411383e-06,
+      "loss": 1.124,
+      "step": 3505
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.848625990892603e-06,
+      "loss": 1.0491,
+      "step": 3506
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8433852467434175e-06,
+      "loss": 1.0839,
+      "step": 3507
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.838147224280625e-06,
+      "loss": 1.1059,
+      "step": 3508
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.83291192581983e-06,
+      "loss": 1.0463,
+      "step": 3509
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.827679353675421e-06,
+      "loss": 1.067,
+      "step": 3510
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.822449510160596e-06,
+      "loss": 1.0838,
+      "step": 3511
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8172223975873355e-06,
+      "loss": 1.0687,
+      "step": 3512
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8119980182664163e-06,
+      "loss": 1.1338,
+      "step": 3513
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8067763745074017e-06,
+      "loss": 1.0775,
+      "step": 3514
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8015574686186585e-06,
+      "loss": 1.0469,
+      "step": 3515
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.7963413029073282e-06,
+      "loss": 1.0655,
+      "step": 3516
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.7911278796793518e-06,
+      "loss": 1.0715,
+      "step": 3517
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7859172012394576e-06,
+      "loss": 1.0114,
+      "step": 3518
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7807092698911495e-06,
+      "loss": 1.0102,
+      "step": 3519
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7755040879367323e-06,
+      "loss": 1.1454,
+      "step": 3520
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.77030165767728e-06,
+      "loss": 1.0647,
+      "step": 3521
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7651019814126656e-06,
+      "loss": 0.9591,
+      "step": 3522
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.759905061441529e-06,
+      "loss": 1.0832,
+      "step": 3523
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.754710900061307e-06,
+      "loss": 1.0571,
+      "step": 3524
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7495194995682007e-06,
+      "loss": 1.0321,
+      "step": 3525
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.744330862257208e-06,
+      "loss": 1.0316,
+      "step": 3526
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.739144990422089e-06,
+      "loss": 1.0873,
+      "step": 3527
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7339618863553983e-06,
+      "loss": 1.1043,
+      "step": 3528
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7287815523484463e-06,
+      "loss": 1.1003,
+      "step": 3529
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.723603990691337e-06,
+      "loss": 1.032,
+      "step": 3530
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7184292036729363e-06,
+      "loss": 1.0045,
+      "step": 3531
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7132571935808924e-06,
+      "loss": 1.125,
+      "step": 3532
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.708087962701625e-06,
+      "loss": 1.0134,
+      "step": 3533
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.702921513320318e-06,
+      "loss": 1.1721,
+      "step": 3534
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6977578477209352e-06,
+      "loss": 1.1392,
+      "step": 3535
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.692596968186198e-06,
+      "loss": 1.0357,
+      "step": 3536
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.687438876997612e-06,
+      "loss": 1.1194,
+      "step": 3537
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6822835764354313e-06,
+      "loss": 1.0485,
+      "step": 3538
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6771310687786943e-06,
+      "loss": 1.0931,
+      "step": 3539
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.671981356305191e-06,
+      "loss": 1.0735,
+      "step": 3540
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6668344412914857e-06,
+      "loss": 1.075,
+      "step": 3541
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.661690326012897e-06,
+      "loss": 1.1305,
+      "step": 3542
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6565490127435153e-06,
+      "loss": 1.1253,
+      "step": 3543
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.651410503756182e-06,
+      "loss": 1.147,
+      "step": 3544
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6462748013225113e-06,
+      "loss": 1.0487,
+      "step": 3545
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6411419077128653e-06,
+      "loss": 1.0775,
+      "step": 3546
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.636011825196365e-06,
+      "loss": 1.1304,
+      "step": 3547
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6308845560409012e-06,
+      "loss": 1.0948,
+      "step": 3548
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.625760102513103e-06,
+      "loss": 1.044,
+      "step": 3549
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6206384668783724e-06,
+      "loss": 1.0232,
+      "step": 3550
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6155196514008496e-06,
+      "loss": 1.0996,
+      "step": 3551
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.610403658343443e-06,
+      "loss": 1.1108,
+      "step": 3552
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.605290489967799e-06,
+      "loss": 1.0565,
+      "step": 3553
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.600180148534328e-06,
+      "loss": 1.0086,
+      "step": 3554
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5950726363021802e-06,
+      "loss": 1.0649,
+      "step": 3555
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5899679555292654e-06,
+      "loss": 1.046,
+      "step": 3556
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5848661084722302e-06,
+      "loss": 1.0209,
+      "step": 3557
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5797670973864795e-06,
+      "loss": 1.0782,
+      "step": 3558
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5746709245261547e-06,
+      "loss": 1.0884,
+      "step": 3559
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5695775921441524e-06,
+      "loss": 1.117,
+      "step": 3560
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5644871024921024e-06,
+      "loss": 1.0789,
+      "step": 3561
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5593994578203893e-06,
+      "loss": 1.0215,
+      "step": 3562
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.554314660378133e-06,
+      "loss": 1.0473,
+      "step": 3563
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.549232712413191e-06,
+      "loss": 1.0544,
+      "step": 3564
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5441536161721736e-06,
+      "loss": 1.0655,
+      "step": 3565
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.539077373900416e-06,
+      "loss": 1.1072,
+      "step": 3566
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.534003987842005e-06,
+      "loss": 1.0664,
+      "step": 3567
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.528933460239754e-06,
+      "loss": 1.0272,
+      "step": 3568
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5238657933352215e-06,
+      "loss": 1.107,
+      "step": 3569
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5188009893686916e-06,
+      "loss": 1.1242,
+      "step": 3570
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5137390505791947e-06,
+      "loss": 1.0777,
+      "step": 3571
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5086799792044812e-06,
+      "loss": 1.0431,
+      "step": 3572
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5036237774810478e-06,
+      "loss": 1.0785,
+      "step": 3573
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4985704476441096e-06,
+      "loss": 1.0469,
+      "step": 3574
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.493519991927623e-06,
+      "loss": 1.022,
+      "step": 3575
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4884724125642646e-06,
+      "loss": 1.0633,
+      "step": 3576
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.483427711785449e-06,
+      "loss": 1.0541,
+      "step": 3577
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4783858918213066e-06,
+      "loss": 1.1626,
+      "step": 3578
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4733469549007083e-06,
+      "loss": 1.0466,
+      "step": 3579
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4683109032512386e-06,
+      "loss": 1.057,
+      "step": 3580
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.463277739099207e-06,
+      "loss": 1.0658,
+      "step": 3581
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4582474646696575e-06,
+      "loss": 1.1495,
+      "step": 3582
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.453220082186344e-06,
+      "loss": 1.0878,
+      "step": 3583
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4481955938717514e-06,
+      "loss": 1.099,
+      "step": 3584
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4431740019470772e-06,
+      "loss": 1.0963,
+      "step": 3585
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4381553086322483e-06,
+      "loss": 1.123,
+      "step": 3586
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.433139516145896e-06,
+      "loss": 1.1111,
+      "step": 3587
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.428126626705386e-06,
+      "loss": 1.0564,
+      "step": 3588
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.423116642526785e-06,
+      "loss": 1.105,
+      "step": 3589
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4181095658248888e-06,
+      "loss": 1.0366,
+      "step": 3590
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4131053988131947e-06,
+      "loss": 1.0369,
+      "step": 3591
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4081041437039288e-06,
+      "loss": 1.0514,
+      "step": 3592
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4031058027080134e-06,
+      "loss": 1.1055,
+      "step": 3593
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3981103780350975e-06,
+      "loss": 1.0907,
+      "step": 3594
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3931178718935332e-06,
+      "loss": 1.0748,
+      "step": 3595
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3881282864903763e-06,
+      "loss": 1.1138,
+      "step": 3596
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3831416240314085e-06,
+      "loss": 1.0874,
+      "step": 3597
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3781578867211016e-06,
+      "loss": 1.0779,
+      "step": 3598
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.373177076762648e-06,
+      "loss": 1.0782,
+      "step": 3599
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.368199196357934e-06,
+      "loss": 1.0735,
+      "step": 3600
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.363224247707563e-06,
+      "loss": 1.144,
+      "step": 3601
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.35825223301083e-06,
+      "loss": 1.0892,
+      "step": 3602
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3532831544657464e-06,
+      "loss": 1.0655,
+      "step": 3603
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3483170142690103e-06,
+      "loss": 1.0459,
+      "step": 3604
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.343353814616036e-06,
+      "loss": 1.0974,
+      "step": 3605
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.338393557700924e-06,
+      "loss": 1.0864,
+      "step": 3606
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.333436245716488e-06,
+      "loss": 1.1188,
+      "step": 3607
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3284818808542263e-06,
+      "loss": 1.0942,
+      "step": 3608
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.323530465304341e-06,
+      "loss": 1.0504,
+      "step": 3609
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3185820012557367e-06,
+      "loss": 1.109,
+      "step": 3610
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3136364908959973e-06,
+      "loss": 1.0995,
+      "step": 3611
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.308693936411421e-06,
+      "loss": 1.1058,
+      "step": 3612
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.303754339986975e-06,
+      "loss": 1.0684,
+      "step": 3613
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.298817703806342e-06,
+      "loss": 1.1607,
+      "step": 3614
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.2938840300518793e-06,
+      "loss": 1.1203,
+      "step": 3615
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2889533209046475e-06,
+      "loss": 1.0924,
+      "step": 3616
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.284025578544385e-06,
+      "loss": 1.1107,
+      "step": 3617
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.279100805149532e-06,
+      "loss": 1.0463,
+      "step": 3618
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2741790028972e-06,
+      "loss": 1.1211,
+      "step": 3619
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2692601739632014e-06,
+      "loss": 1.0717,
+      "step": 3620
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.264344320522024e-06,
+      "loss": 0.9906,
+      "step": 3621
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2594314447468457e-06,
+      "loss": 1.0707,
+      "step": 3622
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2545215488095307e-06,
+      "loss": 1.0979,
+      "step": 3623
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2496146348806167e-06,
+      "loss": 1.0383,
+      "step": 3624
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2447107051293324e-06,
+      "loss": 1.1846,
+      "step": 3625
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.239809761723579e-06,
+      "loss": 1.0538,
+      "step": 3626
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.234911806829948e-06,
+      "loss": 1.0472,
+      "step": 3627
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.230016842613697e-06,
+      "loss": 1.0834,
+      "step": 3628
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.225124871238776e-06,
+      "loss": 1.0258,
+      "step": 3629
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.220235894867794e-06,
+      "loss": 1.0437,
+      "step": 3630
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2153499156620527e-06,
+      "loss": 1.137,
+      "step": 3631
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2104669357815167e-06,
+      "loss": 1.1626,
+      "step": 3632
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2055869573848374e-06,
+      "loss": 1.008,
+      "step": 3633
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2007099826293253e-06,
+      "loss": 1.0689,
+      "step": 3634
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.195836013670972e-06,
+      "loss": 1.0671,
+      "step": 3635
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.190965052664443e-06,
+      "loss": 1.1336,
+      "step": 3636
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1860971017630605e-06,
+      "loss": 1.0729,
+      "step": 3637
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1812321631188324e-06,
+      "loss": 1.0643,
+      "step": 3638
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.176370238882421e-06,
+      "loss": 1.0479,
+      "step": 3639
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1715113312031674e-06,
+      "loss": 1.111,
+      "step": 3640
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1666554422290687e-06,
+      "loss": 1.1304,
+      "step": 3641
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.161802574106799e-06,
+      "loss": 1.0728,
+      "step": 3642
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1569527289816837e-06,
+      "loss": 1.0713,
+      "step": 3643
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1521059089977257e-06,
+      "loss": 1.0701,
+      "step": 3644
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.14726211629758e-06,
+      "loss": 1.1114,
+      "step": 3645
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.142421353022568e-06,
+      "loss": 1.1052,
+      "step": 3646
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1375836213126653e-06,
+      "loss": 1.0638,
+      "step": 3647
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1327489233065224e-06,
+      "loss": 1.1264,
+      "step": 3648
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.127917261141431e-06,
+      "loss": 1.0433,
+      "step": 3649
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1230886369533518e-06,
+      "loss": 1.065,
+      "step": 3650
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.118263052876904e-06,
+      "loss": 1.131,
+      "step": 3651
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1134405110453512e-06,
+      "loss": 1.0673,
+      "step": 3652
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.108621013590626e-06,
+      "loss": 1.092,
+      "step": 3653
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.103804562643302e-06,
+      "loss": 1.1085,
+      "step": 3654
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0989911603326184e-06,
+      "loss": 1.0902,
+      "step": 3655
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0941808087864546e-06,
+      "loss": 1.0947,
+      "step": 3656
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.089373510131354e-06,
+      "loss": 1.025,
+      "step": 3657
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0845692664924978e-06,
+      "loss": 1.0723,
+      "step": 3658
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.079768079993729e-06,
+      "loss": 1.1974,
+      "step": 3659
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.074969952757526e-06,
+      "loss": 1.0937,
+      "step": 3660
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0701748869050285e-06,
+      "loss": 1.077,
+      "step": 3661
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.065382884556012e-06,
+      "loss": 1.0663,
+      "step": 3662
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0605939478288994e-06,
+      "loss": 1.0343,
+      "step": 3663
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.055808078840766e-06,
+      "loss": 1.0975,
+      "step": 3664
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0510252797073204e-06,
+      "loss": 1.0635,
+      "step": 3665
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.046245552542926e-06,
+      "loss": 1.0799,
+      "step": 3666
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0414688994605724e-06,
+      "loss": 1.0537,
+      "step": 3667
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0366953225719076e-06,
+      "loss": 1.0737,
+      "step": 3668
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0319248239872056e-06,
+      "loss": 1.0216,
+      "step": 3669
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0271574058153895e-06,
+      "loss": 1.1093,
+      "step": 3670
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0223930701640124e-06,
+      "loss": 1.1762,
+      "step": 3671
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.017631819139273e-06,
+      "loss": 1.071,
+      "step": 3672
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.012873654845997e-06,
+      "loss": 1.1303,
+      "step": 3673
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0081185793876567e-06,
+      "loss": 1.1337,
+      "step": 3674
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.003366594866345e-06,
+      "loss": 1.0942,
+      "step": 3675
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9986177033828046e-06,
+      "loss": 1.0473,
+      "step": 3676
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9938719070363954e-06,
+      "loss": 1.1009,
+      "step": 3677
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9891292079251223e-06,
+      "loss": 1.108,
+      "step": 3678
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9843896081456115e-06,
+      "loss": 1.0695,
+      "step": 3679
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.979653109793119e-06,
+      "loss": 1.0865,
+      "step": 3680
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9749197149615393e-06,
+      "loss": 1.0497,
+      "step": 3681
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.970189425743383e-06,
+      "loss": 1.078,
+      "step": 3682
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9654622442297997e-06,
+      "loss": 1.0673,
+      "step": 3683
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.960738172510551e-06,
+      "loss": 1.1011,
+      "step": 3684
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9560172126740395e-06,
+      "loss": 1.0749,
+      "step": 3685
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9512993668072775e-06,
+      "loss": 1.0989,
+      "step": 3686
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9465846369959126e-06,
+      "loss": 1.0883,
+      "step": 3687
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9418730253242057e-06,
+      "loss": 1.0953,
+      "step": 3688
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9371645338750477e-06,
+      "loss": 1.042,
+      "step": 3689
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9324591647299404e-06,
+      "loss": 1.1414,
+      "step": 3690
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.927756919969017e-06,
+      "loss": 1.1102,
+      "step": 3691
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9230578016710154e-06,
+      "loss": 1.0617,
+      "step": 3692
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.918361811913306e-06,
+      "loss": 1.1333,
+      "step": 3693
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.913668952771863e-06,
+      "loss": 1.1019,
+      "step": 3694
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.908979226321289e-06,
+      "loss": 1.061,
+      "step": 3695
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9042926346347932e-06,
+      "loss": 1.0279,
+      "step": 3696
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8996091797841976e-06,
+      "loss": 1.0742,
+      "step": 3697
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8949288638399465e-06,
+      "loss": 1.0868,
+      "step": 3698
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8902516888710863e-06,
+      "loss": 1.0178,
+      "step": 3699
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.885577656945285e-06,
+      "loss": 1.13,
+      "step": 3700
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8809067701288117e-06,
+      "loss": 1.0773,
+      "step": 3701
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.876239030486554e-06,
+      "loss": 1.0794,
+      "step": 3702
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8715744400819976e-06,
+      "loss": 1.0322,
+      "step": 3703
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8669130009772483e-06,
+      "loss": 1.099,
+      "step": 3704
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.862254715233006e-06,
+      "loss": 1.041,
+      "step": 3705
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.857599584908589e-06,
+      "loss": 1.1071,
+      "step": 3706
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8529476120619102e-06,
+      "loss": 1.0734,
+      "step": 3707
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.848298798749496e-06,
+      "loss": 1.0557,
+      "step": 3708
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.843653147026466e-06,
+      "loss": 1.0557,
+      "step": 3709
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8390106589465514e-06,
+      "loss": 1.0175,
+      "step": 3710
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.834371336562077e-06,
+      "loss": 1.0286,
+      "step": 3711
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.829735181923978e-06,
+      "loss": 1.0731,
+      "step": 3712
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8251021970817794e-06,
+      "loss": 1.1254,
+      "step": 3713
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8204723840836044e-06,
+      "loss": 1.1411,
+      "step": 3714
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8158457449761865e-06,
+      "loss": 1.1195,
+      "step": 3715
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8112222818048407e-06,
+      "loss": 1.1257,
+      "step": 3716
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8066019966134907e-06,
+      "loss": 1.0594,
+      "step": 3717
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8019848914446444e-06,
+      "loss": 1.1226,
+      "step": 3718
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.797370968339417e-06,
+      "loss": 1.0318,
+      "step": 3719
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7927602293375e-06,
+      "loss": 1.1019,
+      "step": 3720
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7881526764771948e-06,
+      "loss": 1.1016,
+      "step": 3721
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.783548311795379e-06,
+      "loss": 1.0572,
+      "step": 3722
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7789471373275345e-06,
+      "loss": 1.0821,
+      "step": 3723
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7743491551077197e-06,
+      "loss": 1.1395,
+      "step": 3724
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7697543671685946e-06,
+      "loss": 1.0221,
+      "step": 3725
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7651627755413946e-06,
+      "loss": 1.0696,
+      "step": 3726
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7605743822559504e-06,
+      "loss": 1.0859,
+      "step": 3727
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.755989189340682e-06,
+      "loss": 1.0622,
+      "step": 3728
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.751407198822583e-06,
+      "loss": 1.0684,
+      "step": 3729
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7468284127272403e-06,
+      "loss": 1.0485,
+      "step": 3730
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.742252833078818e-06,
+      "loss": 1.0405,
+      "step": 3731
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7376804619000706e-06,
+      "loss": 1.1156,
+      "step": 3732
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7331113012123257e-06,
+      "loss": 1.0852,
+      "step": 3733
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.728545353035501e-06,
+      "loss": 1.1522,
+      "step": 3734
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7239826193880815e-06,
+      "loss": 1.0652,
+      "step": 3735
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7194231022871466e-06,
+      "loss": 1.115,
+      "step": 3736
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.714866803748337e-06,
+      "loss": 1.0287,
+      "step": 3737
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7103137257858867e-06,
+      "loss": 1.0832,
+      "step": 3738
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7057638704125923e-06,
+      "loss": 1.0602,
+      "step": 3739
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.701217239639835e-06,
+      "loss": 1.0474,
+      "step": 3740
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.696673835477569e-06,
+      "loss": 1.0353,
+      "step": 3741
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6921336599343153e-06,
+      "loss": 1.0569,
+      "step": 3742
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.687596715017179e-06,
+      "loss": 1.0331,
+      "step": 3743
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.683063002731824e-06,
+      "loss": 1.1867,
+      "step": 3744
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.678532525082498e-06,
+      "loss": 1.0327,
+      "step": 3745
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.67400528407201e-06,
+      "loss": 1.0525,
+      "step": 3746
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.669481281701739e-06,
+      "loss": 1.1003,
+      "step": 3747
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6649605199716344e-06,
+      "loss": 1.08,
+      "step": 3748
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6604430008802164e-06,
+      "loss": 1.0885,
+      "step": 3749
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6559287264245614e-06,
+      "loss": 1.0826,
+      "step": 3750
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.651417698600325e-06,
+      "loss": 1.0989,
+      "step": 3751
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6469099194017144e-06,
+      "loss": 1.0203,
+      "step": 3752
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.642405390821513e-06,
+      "loss": 1.0618,
+      "step": 3753
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6379041148510543e-06,
+      "loss": 1.0855,
+      "step": 3754
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.633406093480244e-06,
+      "loss": 1.0846,
+      "step": 3755
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6289113286975485e-06,
+      "loss": 1.1666,
+      "step": 3756
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.624419822489985e-06,
+      "loss": 1.1509,
+      "step": 3757
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6199315768431445e-06,
+      "loss": 1.0467,
+      "step": 3758
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.615446593741161e-06,
+      "loss": 1.0259,
+      "step": 3759
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6109648751667404e-06,
+      "loss": 1.0565,
+      "step": 3760
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6064864231011344e-06,
+      "loss": 1.0702,
+      "step": 3761
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6020112395241627e-06,
+      "loss": 1.018,
+      "step": 3762
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5975393264141824e-06,
+      "loss": 1.1084,
+      "step": 3763
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5930706857481224e-06,
+      "loss": 1.0615,
+      "step": 3764
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5886053195014537e-06,
+      "loss": 1.0222,
+      "step": 3765
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.584143229648207e-06,
+      "loss": 1.06,
+      "step": 3766
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5796844181609583e-06,
+      "loss": 1.0792,
+      "step": 3767
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.575228887010839e-06,
+      "loss": 1.1307,
+      "step": 3768
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5707766381675314e-06,
+      "loss": 1.125,
+      "step": 3769
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.566327673599259e-06,
+      "loss": 1.1088,
+      "step": 3770
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5618819952728035e-06,
+      "loss": 1.0599,
+      "step": 3771
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5574396051534835e-06,
+      "loss": 1.0501,
+      "step": 3772
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5530005052051742e-06,
+      "loss": 1.0557,
+      "step": 3773
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5485646973902867e-06,
+      "loss": 0.9825,
+      "step": 3774
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5441321836697874e-06,
+      "loss": 1.0986,
+      "step": 3775
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5397029660031734e-06,
+      "loss": 1.1004,
+      "step": 3776
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5352770463484986e-06,
+      "loss": 1.0603,
+      "step": 3777
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5308544266623457e-06,
+      "loss": 1.1604,
+      "step": 3778
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.526435108899854e-06,
+      "loss": 1.0765,
+      "step": 3779
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.522019095014683e-06,
+      "loss": 1.0462,
+      "step": 3780
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.517606386959047e-06,
+      "loss": 1.1106,
+      "step": 3781
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.513196986683699e-06,
+      "loss": 1.0182,
+      "step": 3782
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.508790896137918e-06,
+      "loss": 1.0815,
+      "step": 3783
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.504388117269535e-06,
+      "loss": 1.044,
+      "step": 3784
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4999886520249016e-06,
+      "loss": 1.0615,
+      "step": 3785
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.495592502348918e-06,
+      "loss": 1.0856,
+      "step": 3786
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4911996701850083e-06,
+      "loss": 1.1208,
+      "step": 3787
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.486810157475138e-06,
+      "loss": 1.008,
+      "step": 3788
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4824239661597983e-06,
+      "loss": 1.0801,
+      "step": 3789
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.478041098178019e-06,
+      "loss": 1.11,
+      "step": 3790
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4736615554673527e-06,
+      "loss": 1.0623,
+      "step": 3791
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.469285339963892e-06,
+      "loss": 1.0979,
+      "step": 3792
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.464912453602246e-06,
+      "loss": 1.039,
+      "step": 3793
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4605428983155667e-06,
+      "loss": 1.0783,
+      "step": 3794
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4561766760355186e-06,
+      "loss": 1.0818,
+      "step": 3795
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4518137886923076e-06,
+      "loss": 1.0582,
+      "step": 3796
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.447454238214654e-06,
+      "loss": 1.0062,
+      "step": 3797
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4430980265298032e-06,
+      "loss": 1.0467,
+      "step": 3798
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.438745155563533e-06,
+      "loss": 1.0767,
+      "step": 3799
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.434395627240137e-06,
+      "loss": 1.0494,
+      "step": 3800
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4300494434824373e-06,
+      "loss": 1.045,
+      "step": 3801
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4257066062117675e-06,
+      "loss": 1.115,
+      "step": 3802
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4213671173479957e-06,
+      "loss": 1.087,
+      "step": 3803
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4170309788094935e-06,
+      "loss": 1.1366,
+      "step": 3804
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4126981925131686e-06,
+      "loss": 1.0538,
+      "step": 3805
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.408368760374431e-06,
+      "loss": 1.1262,
+      "step": 3806
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4040426843072206e-06,
+      "loss": 1.0552,
+      "step": 3807
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3997199662239825e-06,
+      "loss": 1.0717,
+      "step": 3808
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3954006080356894e-06,
+      "loss": 1.0895,
+      "step": 3809
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.391084611651816e-06,
+      "loss": 1.0555,
+      "step": 3810
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3867719789803622e-06,
+      "loss": 1.0438,
+      "step": 3811
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3824627119278344e-06,
+      "loss": 1.1033,
+      "step": 3812
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.378156812399248e-06,
+      "loss": 0.996,
+      "step": 3813
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3738542822981414e-06,
+      "loss": 1.075,
+      "step": 3814
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3695551235265492e-06,
+      "loss": 1.1702,
+      "step": 3815
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3652593379850284e-06,
+      "loss": 1.0503,
+      "step": 3816
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3609669275726353e-06,
+      "loss": 1.0718,
+      "step": 3817
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3566778941869417e-06,
+      "loss": 1.0585,
+      "step": 3818
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3523922397240163e-06,
+      "loss": 1.1166,
+      "step": 3819
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.348109966078448e-06,
+      "loss": 1.0717,
+      "step": 3820
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.343831075143318e-06,
+      "loss": 1.0539,
+      "step": 3821
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.339555568810221e-06,
+      "loss": 1.1598,
+      "step": 3822
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3352834489692488e-06,
+      "loss": 1.0207,
+      "step": 3823
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.331014717509004e-06,
+      "loss": 1.0813,
+      "step": 3824
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3267493763165803e-06,
+      "loss": 1.0879,
+      "step": 3825
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3224874272775875e-06,
+      "loss": 0.9716,
+      "step": 3826
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.318228872276118e-06,
+      "loss": 1.034,
+      "step": 3827
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3139737131947825e-06,
+      "loss": 1.0534,
+      "step": 3828
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.309721951914675e-06,
+      "loss": 1.0573,
+      "step": 3829
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3054735903153926e-06,
+      "loss": 1.1718,
+      "step": 3830
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3012286302750354e-06,
+      "loss": 1.1013,
+      "step": 3831
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.296987073670189e-06,
+      "loss": 1.0464,
+      "step": 3832
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.292748922375947e-06,
+      "loss": 1.0,
+      "step": 3833
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.288514178265884e-06,
+      "loss": 1.0684,
+      "step": 3834
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.28428284321208e-06,
+      "loss": 1.083,
+      "step": 3835
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2800549190850997e-06,
+      "loss": 1.1414,
+      "step": 3836
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.275830407754006e-06,
+      "loss": 1.065,
+      "step": 3837
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.271609311086348e-06,
+      "loss": 1.1146,
+      "step": 3838
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2673916309481712e-06,
+      "loss": 1.0551,
+      "step": 3839
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2631773692040017e-06,
+      "loss": 1.0912,
+      "step": 3840
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.258966527716866e-06,
+      "loss": 1.0595,
+      "step": 3841
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.254759108348267e-06,
+      "loss": 1.1349,
+      "step": 3842
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2505551129582047e-06,
+      "loss": 1.0645,
+      "step": 3843
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.246354543405158e-06,
+      "loss": 1.1396,
+      "step": 3844
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2421574015460947e-06,
+      "loss": 1.0767,
+      "step": 3845
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.237963689236472e-06,
+      "loss": 1.075,
+      "step": 3846
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2337734083302164e-06,
+      "loss": 1.0805,
+      "step": 3847
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2295865606797564e-06,
+      "loss": 1.0235,
+      "step": 3848
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.225403148135985e-06,
+      "loss": 1.1065,
+      "step": 3849
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2212231725482914e-06,
+      "loss": 1.0307,
+      "step": 3850
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2170466357645336e-06,
+      "loss": 1.1574,
+      "step": 3851
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2128735396310606e-06,
+      "loss": 1.0991,
+      "step": 3852
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.20870388599269e-06,
+      "loss": 1.1174,
+      "step": 3853
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2045376766927253e-06,
+      "loss": 1.0446,
+      "step": 3854
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.200374913572939e-06,
+      "loss": 1.03,
+      "step": 3855
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.196215598473592e-06,
+      "loss": 1.1024,
+      "step": 3856
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.192059733233408e-06,
+      "loss": 0.9996,
+      "step": 3857
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1879073196895973e-06,
+      "loss": 1.0842,
+      "step": 3858
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.183758359677832e-06,
+      "loss": 1.0957,
+      "step": 3859
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1796128550322703e-06,
+      "loss": 1.0675,
+      "step": 3860
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.175470807585536e-06,
+      "loss": 1.0864,
+      "step": 3861
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1713322191687234e-06,
+      "loss": 1.0896,
+      "step": 3862
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1671970916114014e-06,
+      "loss": 1.0834,
+      "step": 3863
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.163065426741603e-06,
+      "loss": 1.027,
+      "step": 3864
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.15893722638584e-06,
+      "loss": 1.047,
+      "step": 3865
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.154812492369083e-06,
+      "loss": 1.0617,
+      "step": 3866
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1506912265147772e-06,
+      "loss": 1.1139,
+      "step": 3867
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1465734306448283e-06,
+      "loss": 1.092,
+      "step": 3868
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1424591065796174e-06,
+      "loss": 1.1053,
+      "step": 3869
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1383482561379788e-06,
+      "loss": 1.1008,
+      "step": 3870
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1342408811372217e-06,
+      "loss": 1.0986,
+      "step": 3871
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.130136983393112e-06,
+      "loss": 1.0325,
+      "step": 3872
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.12603656471988e-06,
+      "loss": 1.0808,
+      "step": 3873
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1219396269302238e-06,
+      "loss": 1.1165,
+      "step": 3874
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.117846171835293e-06,
+      "loss": 1.0704,
+      "step": 3875
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1137562012447055e-06,
+      "loss": 1.0381,
+      "step": 3876
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1096697169665312e-06,
+      "loss": 1.0341,
+      "step": 3877
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.10558672080731e-06,
+      "loss": 1.0196,
+      "step": 3878
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1015072145720283e-06,
+      "loss": 1.0448,
+      "step": 3879
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.097431200064136e-06,
+      "loss": 1.0641,
+      "step": 3880
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0933586790855354e-06,
+      "loss": 1.0795,
+      "step": 3881
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.08928965343659e-06,
+      "loss": 1.072,
+      "step": 3882
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.085224124916112e-06,
+      "loss": 0.9914,
+      "step": 3883
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.081162095321374e-06,
+      "loss": 1.0171,
+      "step": 3884
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0771035664480944e-06,
+      "loss": 1.1096,
+      "step": 3885
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0730485400904498e-06,
+      "loss": 1.055,
+      "step": 3886
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.068997018041069e-06,
+      "loss": 1.0874,
+      "step": 3887
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0649490020910247e-06,
+      "loss": 1.0978,
+      "step": 3888
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0609044940298495e-06,
+      "loss": 1.0692,
+      "step": 3889
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0568634956455145e-06,
+      "loss": 1.0939,
+      "step": 3890
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.052826008724449e-06,
+      "loss": 1.0482,
+      "step": 3891
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.048792035051521e-06,
+      "loss": 1.0905,
+      "step": 3892
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.044761576410056e-06,
+      "loss": 0.9978,
+      "step": 3893
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0407346345818137e-06,
+      "loss": 1.1401,
+      "step": 3894
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0367112113470112e-06,
+      "loss": 1.0597,
+      "step": 3895
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.032691308484299e-06,
+      "loss": 1.0966,
+      "step": 3896
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0286749277707783e-06,
+      "loss": 1.0348,
+      "step": 3897
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0246620709819874e-06,
+      "loss": 1.0215,
+      "step": 3898
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.020652739891914e-06,
+      "loss": 1.0208,
+      "step": 3899
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0166469362729868e-06,
+      "loss": 1.0877,
+      "step": 3900
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.012644661896065e-06,
+      "loss": 1.1001,
+      "step": 3901
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.008645918530462e-06,
+      "loss": 1.0655,
+      "step": 3902
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.004650707943916e-06,
+      "loss": 1.0889,
+      "step": 3903
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0006590319026164e-06,
+      "loss": 1.0143,
+      "step": 3904
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9966708921711796e-06,
+      "loss": 1.1155,
+      "step": 3905
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9926862905126663e-06,
+      "loss": 1.0712,
+      "step": 3906
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9887052286885654e-06,
+      "loss": 1.1292,
+      "step": 3907
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9847277084588114e-06,
+      "loss": 1.0111,
+      "step": 3908
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9807537315817606e-06,
+      "loss": 1.0369,
+      "step": 3909
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9767832998142167e-06,
+      "loss": 1.1131,
+      "step": 3910
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9728164149114015e-06,
+      "loss": 1.052,
+      "step": 3911
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9688530786269854e-06,
+      "loss": 1.0191,
+      "step": 3912
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.964893292713049e-06,
+      "loss": 1.0592,
+      "step": 3913
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9609370589201227e-06,
+      "loss": 1.1401,
+      "step": 3914
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9569843789971597e-06,
+      "loss": 1.0524,
+      "step": 3915
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.953035254691539e-06,
+      "loss": 1.1541,
+      "step": 3916
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9490896877490715e-06,
+      "loss": 1.0714,
+      "step": 3917
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9451476799139935e-06,
+      "loss": 1.0582,
+      "step": 3918
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.94120923292897e-06,
+      "loss": 1.0573,
+      "step": 3919
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9372743485350887e-06,
+      "loss": 1.0746,
+      "step": 3920
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.933343028471867e-06,
+      "loss": 1.0945,
+      "step": 3921
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.929415274477239e-06,
+      "loss": 1.1376,
+      "step": 3922
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.925491088287573e-06,
+      "loss": 1.1373,
+      "step": 3923
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9215704716376495e-06,
+      "loss": 1.1669,
+      "step": 3924
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9176534262606796e-06,
+      "loss": 1.1261,
+      "step": 3925
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9137399538882872e-06,
+      "loss": 1.0953,
+      "step": 3926
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9098300562505266e-06,
+      "loss": 1.0257,
+      "step": 3927
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9059237350758608e-06,
+      "loss": 1.0455,
+      "step": 3928
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9020209920911825e-06,
+      "loss": 1.1035,
+      "step": 3929
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8981218290217952e-06,
+      "loss": 1.0441,
+      "step": 3930
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8942262475914186e-06,
+      "loss": 1.1267,
+      "step": 3931
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8903342495221977e-06,
+      "loss": 1.0268,
+      "step": 3932
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.886445836534684e-06,
+      "loss": 1.0456,
+      "step": 3933
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8825610103478531e-06,
+      "loss": 1.0869,
+      "step": 3934
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.878679772679084e-06,
+      "loss": 1.0164,
+      "step": 3935
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8748021252441816e-06,
+      "loss": 1.0666,
+      "step": 3936
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.870928069757353e-06,
+      "loss": 1.0954,
+      "step": 3937
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8670576079312242e-06,
+      "loss": 1.0395,
+      "step": 3938
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.863190741476828e-06,
+      "loss": 1.1456,
+      "step": 3939
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8593274721036136e-06,
+      "loss": 1.1253,
+      "step": 3940
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8554678015194316e-06,
+      "loss": 1.0586,
+      "step": 3941
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8516117314305526e-06,
+      "loss": 1.135,
+      "step": 3942
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8477592635416418e-06,
+      "loss": 1.0747,
+      "step": 3943
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8439103995557872e-06,
+      "loss": 1.1007,
+      "step": 3944
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8400651411744685e-06,
+      "loss": 1.0434,
+      "step": 3945
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8362234900975862e-06,
+      "loss": 1.0716,
+      "step": 3946
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8323854480234348e-06,
+      "loss": 1.1658,
+      "step": 3947
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8285510166487154e-06,
+      "loss": 1.0541,
+      "step": 3948
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8247201976685402e-06,
+      "loss": 1.0168,
+      "step": 3949
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.820892992776413e-06,
+      "loss": 1.0764,
+      "step": 3950
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.817069403664251e-06,
+      "loss": 1.1257,
+      "step": 3951
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8132494320223636e-06,
+      "loss": 1.1104,
+      "step": 3952
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.809433079539471e-06,
+      "loss": 1.0738,
+      "step": 3953
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8056203479026812e-06,
+      "loss": 1.0555,
+      "step": 3954
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8018112387975139e-06,
+      "loss": 1.1258,
+      "step": 3955
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7980057539078776e-06,
+      "loss": 1.1906,
+      "step": 3956
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7942038949160857e-06,
+      "loss": 1.1437,
+      "step": 3957
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7904056635028421e-06,
+      "loss": 1.0299,
+      "step": 3958
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7866110613472543e-06,
+      "loss": 1.0322,
+      "step": 3959
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7828200901268177e-06,
+      "loss": 1.0532,
+      "step": 3960
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.779032751517431e-06,
+      "loss": 1.0987,
+      "step": 3961
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7752490471933769e-06,
+      "loss": 1.0307,
+      "step": 3962
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.771468978827343e-06,
+      "loss": 1.0958,
+      "step": 3963
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7676925480904016e-06,
+      "loss": 1.0664,
+      "step": 3964
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7639197566520138e-06,
+      "loss": 1.0679,
+      "step": 3965
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7601506061800456e-06,
+      "loss": 1.1159,
+      "step": 3966
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.756385098340736e-06,
+      "loss": 1.1671,
+      "step": 3967
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7526232347987304e-06,
+      "loss": 1.1125,
+      "step": 3968
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7488650172170496e-06,
+      "loss": 1.0888,
+      "step": 3969
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7451104472571113e-06,
+      "loss": 1.1324,
+      "step": 3970
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7413595265787142e-06,
+      "loss": 0.9949,
+      "step": 3971
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7376122568400533e-06,
+      "loss": 1.0996,
+      "step": 3972
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7338686396976956e-06,
+      "loss": 1.113,
+      "step": 3973
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7301286768066083e-06,
+      "loss": 1.0451,
+      "step": 3974
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7263923698201302e-06,
+      "loss": 1.054,
+      "step": 3975
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7226597203899941e-06,
+      "loss": 1.0829,
+      "step": 3976
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7189307301663082e-06,
+      "loss": 1.0396,
+      "step": 3977
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7152054007975683e-06,
+      "loss": 1.1289,
+      "step": 3978
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7114837339306557e-06,
+      "loss": 1.0436,
+      "step": 3979
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7077657312108143e-06,
+      "loss": 0.9504,
+      "step": 3980
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7040513942816905e-06,
+      "loss": 1.0038,
+      "step": 3981
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7003407247852944e-06,
+      "loss": 1.1044,
+      "step": 3982
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6966337243620267e-06,
+      "loss": 1.1126,
+      "step": 3983
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6929303946506537e-06,
+      "loss": 1.0031,
+      "step": 3984
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.689230737288332e-06,
+      "loss": 1.0541,
+      "step": 3985
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6855347539105826e-06,
+      "loss": 1.101,
+      "step": 3986
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6818424461513129e-06,
+      "loss": 1.0412,
+      "step": 3987
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6781538156427946e-06,
+      "loss": 1.104,
+      "step": 3988
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6744688640156847e-06,
+      "loss": 1.1143,
+      "step": 3989
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6707875928990059e-06,
+      "loss": 1.0834,
+      "step": 3990
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6671100039201559e-06,
+      "loss": 0.9871,
+      "step": 3991
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6634360987049113e-06,
+      "loss": 1.0692,
+      "step": 3992
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6597658788774064e-06,
+      "loss": 1.0818,
+      "step": 3993
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6560993460601614e-06,
+      "loss": 1.0629,
+      "step": 3994
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6524365018740529e-06,
+      "loss": 1.0253,
+      "step": 3995
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6487773479383406e-06,
+      "loss": 1.0044,
+      "step": 3996
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6451218858706374e-06,
+      "loss": 1.0944,
+      "step": 3997
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6414701172869375e-06,
+      "loss": 1.0579,
+      "step": 3998
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6378220438015935e-06,
+      "loss": 1.0646,
+      "step": 3999
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6341776670273324e-06,
+      "loss": 1.0885,
+      "step": 4000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6305369885752377e-06,
+      "loss": 1.0394,
+      "step": 4001
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6269000100547682e-06,
+      "loss": 1.0236,
+      "step": 4002
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6232667330737362e-06,
+      "loss": 1.1071,
+      "step": 4003
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6196371592383264e-06,
+      "loss": 1.0919,
+      "step": 4004
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6160112901530855e-06,
+      "loss": 1.1421,
+      "step": 4005
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6123891274209147e-06,
+      "loss": 1.0655,
+      "step": 4006
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6087706726430874e-06,
+      "loss": 1.0696,
+      "step": 4007
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6051559274192275e-06,
+      "loss": 1.0304,
+      "step": 4008
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6015448933473288e-06,
+      "loss": 1.0474,
+      "step": 4009
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5979375720237344e-06,
+      "loss": 1.0978,
+      "step": 4010
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5943339650431578e-06,
+      "loss": 1.0263,
+      "step": 4011
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5907340739986577e-06,
+      "loss": 1.1231,
+      "step": 4012
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5871379004816645e-06,
+      "loss": 1.1018,
+      "step": 4013
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5835454460819478e-06,
+      "loss": 1.1004,
+      "step": 4014
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.579956712387649e-06,
+      "loss": 1.0664,
+      "step": 4015
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5763717009852542e-06,
+      "loss": 1.004,
+      "step": 4016
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5727904134596084e-06,
+      "loss": 1.1059,
+      "step": 4017
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5692128513939142e-06,
+      "loss": 1.031,
+      "step": 4018
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5656390163697165e-06,
+      "loss": 1.0951,
+      "step": 4019
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.562068909966925e-06,
+      "loss": 1.1003,
+      "step": 4020
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5585025337637905e-06,
+      "loss": 1.1174,
+      "step": 4021
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5549398893369216e-06,
+      "loss": 1.0647,
+      "step": 4022
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5513809782612731e-06,
+      "loss": 1.0491,
+      "step": 4023
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5478258021101543e-06,
+      "loss": 1.0643,
+      "step": 4024
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.544274362455216e-06,
+      "loss": 1.1472,
+      "step": 4025
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.540726660866466e-06,
+      "loss": 1.0669,
+      "step": 4026
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5371826989122507e-06,
+      "loss": 1.056,
+      "step": 4027
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5336424781592707e-06,
+      "loss": 1.191,
+      "step": 4028
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5301060001725698e-06,
+      "loss": 1.0969,
+      "step": 4029
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5265732665155353e-06,
+      "loss": 1.0493,
+      "step": 4030
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.523044278749899e-06,
+      "loss": 1.1584,
+      "step": 4031
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5195190384357405e-06,
+      "loss": 1.0498,
+      "step": 4032
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5159975471314848e-06,
+      "loss": 1.054,
+      "step": 4033
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5124798063938895e-06,
+      "loss": 1.1067,
+      "step": 4034
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5089658177780653e-06,
+      "loss": 1.1158,
+      "step": 4035
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5054555828374552e-06,
+      "loss": 1.0647,
+      "step": 4036
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.501949103123852e-06,
+      "loss": 1.1025,
+      "step": 4037
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4984463801873773e-06,
+      "loss": 1.0522,
+      "step": 4038
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.494947415576502e-06,
+      "loss": 1.1745,
+      "step": 4039
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.491452210838029e-06,
+      "loss": 1.0411,
+      "step": 4040
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4879607675171048e-06,
+      "loss": 1.0179,
+      "step": 4041
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4844730871572045e-06,
+      "loss": 1.1714,
+      "step": 4042
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.480989171300151e-06,
+      "loss": 0.9835,
+      "step": 4043
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.477509021486091e-06,
+      "loss": 1.082,
+      "step": 4044
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4740326392535175e-06,
+      "loss": 1.0419,
+      "step": 4045
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4705600261392505e-06,
+      "loss": 1.0378,
+      "step": 4046
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.467091183678444e-06,
+      "loss": 1.0226,
+      "step": 4047
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4636261134045904e-06,
+      "loss": 1.0732,
+      "step": 4048
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4601648168495074e-06,
+      "loss": 1.1312,
+      "step": 4049
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4567072955433526e-06,
+      "loss": 1.1175,
+      "step": 4050
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4532535510146061e-06,
+      "loss": 1.0943,
+      "step": 4051
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.449803584790086e-06,
+      "loss": 0.9719,
+      "step": 4052
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.446357398394934e-06,
+      "loss": 1.1608,
+      "step": 4053
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4429149933526264e-06,
+      "loss": 1.0825,
+      "step": 4054
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4394763711849601e-06,
+      "loss": 1.1066,
+      "step": 4055
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4360415334120704e-06,
+      "loss": 1.0426,
+      "step": 4056
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.432610481552409e-06,
+      "loss": 1.1496,
+      "step": 4057
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4291832171227616e-06,
+      "loss": 1.0787,
+      "step": 4058
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.425759741638234e-06,
+      "loss": 1.0421,
+      "step": 4059
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4223400566122635e-06,
+      "loss": 1.0752,
+      "step": 4060
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.418924163556603e-06,
+      "loss": 1.0341,
+      "step": 4061
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4155120639813392e-06,
+      "loss": 1.0076,
+      "step": 4062
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4121037593948738e-06,
+      "loss": 1.0791,
+      "step": 4063
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.408699251303931e-06,
+      "loss": 1.067,
+      "step": 4064
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4052985412135646e-06,
+      "loss": 1.1076,
+      "step": 4065
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4019016306271382e-06,
+      "loss": 1.0414,
+      "step": 4066
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3985085210463479e-06,
+      "loss": 1.0759,
+      "step": 4067
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3951192139711967e-06,
+      "loss": 1.0695,
+      "step": 4068
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3917337109000196e-06,
+      "loss": 1.1166,
+      "step": 4069
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3883520133294558e-06,
+      "loss": 1.0364,
+      "step": 4070
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3849741227544777e-06,
+      "loss": 1.0934,
+      "step": 4071
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3816000406683604e-06,
+      "loss": 1.1313,
+      "step": 4072
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3782297685627066e-06,
+      "loss": 1.0623,
+      "step": 4073
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3748633079274254e-06,
+      "loss": 1.0746,
+      "step": 4074
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3715006602507497e-06,
+      "loss": 1.1082,
+      "step": 4075
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3681418270192182e-06,
+      "loss": 1.1434,
+      "step": 4076
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.364786809717692e-06,
+      "loss": 1.044,
+      "step": 4077
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3614356098293368e-06,
+      "loss": 1.0287,
+      "step": 4078
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3580882288356379e-06,
+      "loss": 1.1266,
+      "step": 4079
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.354744668216389e-06,
+      "loss": 0.9852,
+      "step": 4080
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3514049294496911e-06,
+      "loss": 1.0054,
+      "step": 4081
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3480690140119657e-06,
+      "loss": 1.0652,
+      "step": 4082
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3447369233779328e-06,
+      "loss": 1.0487,
+      "step": 4083
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3414086590206322e-06,
+      "loss": 1.1222,
+      "step": 4084
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.338084222411401e-06,
+      "loss": 1.0469,
+      "step": 4085
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.334763615019895e-06,
+      "loss": 1.0761,
+      "step": 4086
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3314468383140687e-06,
+      "loss": 1.1031,
+      "step": 4087
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3281338937601895e-06,
+      "loss": 1.0551,
+      "step": 4088
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3248247828228244e-06,
+      "loss": 1.0871,
+      "step": 4089
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3215195069648546e-06,
+      "loss": 1.132,
+      "step": 4090
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3182180676474533e-06,
+      "loss": 1.0688,
+      "step": 4091
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3149204663301118e-06,
+      "loss": 1.004,
+      "step": 4092
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.311626704470612e-06,
+      "loss": 1.0459,
+      "step": 4093
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3083367835250481e-06,
+      "loss": 1.0309,
+      "step": 4094
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.30505070494781e-06,
+      "loss": 1.0679,
+      "step": 4095
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3017684701915933e-06,
+      "loss": 1.0955,
+      "step": 4096
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2984900807073919e-06,
+      "loss": 1.1425,
+      "step": 4097
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2952155379444975e-06,
+      "loss": 1.0561,
+      "step": 4098
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2919448433505088e-06,
+      "loss": 1.1079,
+      "step": 4099
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.288677998371314e-06,
+      "loss": 1.0942,
+      "step": 4100
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2854150044511094e-06,
+      "loss": 1.0836,
+      "step": 4101
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.282155863032377e-06,
+      "loss": 1.069,
+      "step": 4102
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2789005755559082e-06,
+      "loss": 1.0797,
+      "step": 4103
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2756491434607798e-06,
+      "loss": 1.1188,
+      "step": 4104
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2724015681843737e-06,
+      "loss": 1.1163,
+      "step": 4105
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2691578511623582e-06,
+      "loss": 1.058,
+      "step": 4106
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2659179938287035e-06,
+      "loss": 1.0833,
+      "step": 4107
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2626819976156668e-06,
+      "loss": 1.0484,
+      "step": 4108
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2594498639538032e-06,
+      "loss": 0.9634,
+      "step": 4109
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2562215942719602e-06,
+      "loss": 1.036,
+      "step": 4110
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2529971899972726e-06,
+      "loss": 1.0566,
+      "step": 4111
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2497766525551724e-06,
+      "loss": 0.9899,
+      "step": 4112
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2465599833693776e-06,
+      "loss": 1.0706,
+      "step": 4113
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2433471838618981e-06,
+      "loss": 1.0754,
+      "step": 4114
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2401382554530307e-06,
+      "loss": 1.1181,
+      "step": 4115
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2369331995613664e-06,
+      "loss": 1.0897,
+      "step": 4116
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.233732017603776e-06,
+      "loss": 1.0368,
+      "step": 4117
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.230534710995428e-06,
+      "loss": 1.1438,
+      "step": 4118
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2273412811497675e-06,
+      "loss": 1.0473,
+      "step": 4119
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2241517294785355e-06,
+      "loss": 1.0259,
+      "step": 4120
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2209660573917481e-06,
+      "loss": 0.97,
+      "step": 4121
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2177842662977136e-06,
+      "loss": 1.0003,
+      "step": 4122
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2146063576030265e-06,
+      "loss": 1.0679,
+      "step": 4123
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2114323327125567e-06,
+      "loss": 1.0223,
+      "step": 4124
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2082621930294635e-06,
+      "loss": 1.0759,
+      "step": 4125
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2050959399551865e-06,
+      "loss": 1.0858,
+      "step": 4126
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2019335748894489e-06,
+      "loss": 1.0303,
+      "step": 4127
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1987750992302505e-06,
+      "loss": 1.1015,
+      "step": 4128
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1956205143738786e-06,
+      "loss": 1.043,
+      "step": 4129
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1924698217148955e-06,
+      "loss": 1.0352,
+      "step": 4130
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.189323022646144e-06,
+      "loss": 1.0796,
+      "step": 4131
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.186180118558743e-06,
+      "loss": 1.0608,
+      "step": 4132
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1830411108420969e-06,
+      "loss": 1.0707,
+      "step": 4133
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1799060008838792e-06,
+      "loss": 1.0872,
+      "step": 4134
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1767747900700476e-06,
+      "loss": 1.0539,
+      "step": 4135
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1736474797848286e-06,
+      "loss": 1.0378,
+      "step": 4136
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1705240714107301e-06,
+      "loss": 0.9959,
+      "step": 4137
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.167404566328536e-06,
+      "loss": 1.1067,
+      "step": 4138
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1642889659172985e-06,
+      "loss": 1.0932,
+      "step": 4139
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.16117727155435e-06,
+      "loss": 0.9578,
+      "step": 4140
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1580694846152896e-06,
+      "loss": 1.1571,
+      "step": 4141
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1549656064739966e-06,
+      "loss": 1.0564,
+      "step": 4142
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.151865638502615e-06,
+      "loss": 1.1145,
+      "step": 4143
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1487695820715672e-06,
+      "loss": 1.1194,
+      "step": 4144
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1456774385495385e-06,
+      "loss": 1.101,
+      "step": 4145
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.142589209303494e-06,
+      "loss": 1.0526,
+      "step": 4146
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1395048956986577e-06,
+      "loss": 1.0748,
+      "step": 4147
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1364244990985308e-06,
+      "loss": 1.077,
+      "step": 4148
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1333480208648784e-06,
+      "loss": 1.1491,
+      "step": 4149
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1302754623577373e-06,
+      "loss": 1.0169,
+      "step": 4150
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1272068249354085e-06,
+      "loss": 1.1185,
+      "step": 4151
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.124142109954459e-06,
+      "loss": 1.0286,
+      "step": 4152
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1210813187697267e-06,
+      "loss": 1.0792,
+      "step": 4153
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1180244527343065e-06,
+      "loss": 1.0195,
+      "step": 4154
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1149715131995675e-06,
+      "loss": 1.0243,
+      "step": 4155
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1119225015151336e-06,
+      "loss": 1.1018,
+      "step": 4156
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.108877419028902e-06,
+      "loss": 1.0357,
+      "step": 4157
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1058362670870248e-06,
+      "loss": 1.0563,
+      "step": 4158
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1027990470339212e-06,
+      "loss": 1.041,
+      "step": 4159
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0997657602122691e-06,
+      "loss": 1.1005,
+      "step": 4160
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0967364079630117e-06,
+      "loss": 1.1436,
+      "step": 4161
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0937109916253474e-06,
+      "loss": 1.0497,
+      "step": 4162
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.09068951253674e-06,
+      "loss": 1.0238,
+      "step": 4163
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.087671972032911e-06,
+      "loss": 1.086,
+      "step": 4164
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0846583714478355e-06,
+      "loss": 1.1216,
+      "step": 4165
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0816487121137564e-06,
+      "loss": 1.0321,
+      "step": 4166
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0786429953611665e-06,
+      "loss": 1.0886,
+      "step": 4167
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0756412225188205e-06,
+      "loss": 1.0432,
+      "step": 4168
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0726433949137239e-06,
+      "loss": 1.0402,
+      "step": 4169
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0696495138711472e-06,
+      "loss": 1.0392,
+      "step": 4170
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0666595807146053e-06,
+      "loss": 1.0237,
+      "step": 4171
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0636735967658785e-06,
+      "loss": 1.0604,
+      "step": 4172
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.060691563344991e-06,
+      "loss": 1.0017,
+      "step": 4173
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0577134817702305e-06,
+      "loss": 1.1419,
+      "step": 4174
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0547393533581285e-06,
+      "loss": 1.0935,
+      "step": 4175
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0517691794234774e-06,
+      "loss": 1.0429,
+      "step": 4176
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0488029612793138e-06,
+      "loss": 1.1014,
+      "step": 4177
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.045840700236932e-06,
+      "loss": 1.0486,
+      "step": 4178
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.042882397605871e-06,
+      "loss": 1.0659,
+      "step": 4179
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0399280546939261e-06,
+      "loss": 1.0772,
+      "step": 4180
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0369776728071367e-06,
+      "loss": 1.0537,
+      "step": 4181
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.034031253249792e-06,
+      "loss": 1.1452,
+      "step": 4182
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0310887973244355e-06,
+      "loss": 1.0799,
+      "step": 4183
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0281503063318477e-06,
+      "loss": 1.0798,
+      "step": 4184
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0252157815710685e-06,
+      "loss": 1.1527,
+      "step": 4185
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0222852243393732e-06,
+      "loss": 1.0158,
+      "step": 4186
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0193586359322927e-06,
+      "loss": 1.0494,
+      "step": 4187
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0164360176435962e-06,
+      "loss": 1.0513,
+      "step": 4188
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0135173707653023e-06,
+      "loss": 1.058,
+      "step": 4189
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0106026965876703e-06,
+      "loss": 1.0975,
+      "step": 4190
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0076919963992071e-06,
+      "loss": 1.0621,
+      "step": 4191
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0047852714866591e-06,
+      "loss": 1.0736,
+      "step": 4192
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0018825231350203e-06,
+      "loss": 1.0518,
+      "step": 4193
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.989837526275193e-07,
+      "loss": 1.0692,
+      "step": 4194
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.960889612456349e-07,
+      "loss": 1.0999,
+      "step": 4195
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.931981502690803e-07,
+      "loss": 1.085,
+      "step": 4196
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.903113209758098e-07,
+      "loss": 1.0925,
+      "step": 4197
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.874284746420216e-07,
+      "loss": 1.0402,
+      "step": 4198
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.845496125421473e-07,
+      "loss": 1.0921,
+      "step": 4199
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.816747359488632e-07,
+      "loss": 1.0081,
+      "step": 4200
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.788038461330784e-07,
+      "loss": 1.0632,
+      "step": 4201
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.759369443639455e-07,
+      "loss": 1.0389,
+      "step": 4202
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.730740319088473e-07,
+      "loss": 1.1144,
+      "step": 4203
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.702151100334078e-07,
+      "loss": 1.0883,
+      "step": 4204
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.673601800014832e-07,
+      "loss": 1.1475,
+      "step": 4205
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.645092430751702e-07,
+      "loss": 1.0663,
+      "step": 4206
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.616623005147952e-07,
+      "loss": 1.1297,
+      "step": 4207
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.588193535789214e-07,
+      "loss": 1.1199,
+      "step": 4208
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.559804035243437e-07,
+      "loss": 1.106,
+      "step": 4209
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.531454516060945e-07,
+      "loss": 1.0931,
+      "step": 4210
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.503144990774315e-07,
+      "loss": 1.0749,
+      "step": 4211
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.474875471898526e-07,
+      "loss": 1.087,
+      "step": 4212
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.446645971930812e-07,
+      "loss": 1.0163,
+      "step": 4213
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.418456503350714e-07,
+      "loss": 1.0556,
+      "step": 4214
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.390307078620131e-07,
+      "loss": 1.1173,
+      "step": 4215
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.362197710183174e-07,
+      "loss": 1.1242,
+      "step": 4216
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.334128410466359e-07,
+      "loss": 1.1059,
+      "step": 4217
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.306099191878381e-07,
+      "loss": 1.0648,
+      "step": 4218
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.278110066810286e-07,
+      "loss": 1.075,
+      "step": 4219
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.250161047635342e-07,
+      "loss": 1.0828,
+      "step": 4220
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.222252146709143e-07,
+      "loss": 1.1012,
+      "step": 4221
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.194383376369509e-07,
+      "loss": 1.0679,
+      "step": 4222
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.166554748936541e-07,
+      "loss": 1.0332,
+      "step": 4223
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.138766276712552e-07,
+      "loss": 1.131,
+      "step": 4224
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.111017971982195e-07,
+      "loss": 1.067,
+      "step": 4225
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.083309847012234e-07,
+      "loss": 1.1375,
+      "step": 4226
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.055641914051783e-07,
+      "loss": 1.1062,
+      "step": 4227
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.028014185332168e-07,
+      "loss": 1.0873,
+      "step": 4228
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.000426673066876e-07,
+      "loss": 1.0614,
+      "step": 4229
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.972879389451716e-07,
+      "loss": 1.0708,
+      "step": 4230
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.945372346664583e-07,
+      "loss": 1.1558,
+      "step": 4231
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.917905556865714e-07,
+      "loss": 1.0486,
+      "step": 4232
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.890479032197463e-07,
+      "loss": 1.0799,
+      "step": 4233
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.863092784784444e-07,
+      "loss": 1.0935,
+      "step": 4234
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.835746826733404e-07,
+      "loss": 1.1296,
+      "step": 4235
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.808441170133341e-07,
+      "loss": 1.0462,
+      "step": 4236
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.781175827055388e-07,
+      "loss": 1.1091,
+      "step": 4237
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.753950809552891e-07,
+      "loss": 1.0802,
+      "step": 4238
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.726766129661335e-07,
+      "loss": 1.0017,
+      "step": 4239
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.699621799398394e-07,
+      "loss": 1.1052,
+      "step": 4240
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.672517830763938e-07,
+      "loss": 1.0786,
+      "step": 4241
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.645454235739903e-07,
+      "loss": 1.0485,
+      "step": 4242
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.618431026290475e-07,
+      "loss": 1.1451,
+      "step": 4243
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.591448214361919e-07,
+      "loss": 1.1016,
+      "step": 4244
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.564505811882684e-07,
+      "loss": 1.0427,
+      "step": 4245
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.53760383076333e-07,
+      "loss": 1.0954,
+      "step": 4246
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.510742282896545e-07,
+      "loss": 1.0674,
+      "step": 4247
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.483921180157151e-07,
+      "loss": 1.0821,
+      "step": 4248
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.457140534402098e-07,
+      "loss": 1.0583,
+      "step": 4249
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.430400357470437e-07,
+      "loss": 1.0791,
+      "step": 4250
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.403700661183356e-07,
+      "loss": 1.0892,
+      "step": 4251
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.377041457344104e-07,
+      "loss": 1.1196,
+      "step": 4252
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.350422757738075e-07,
+      "loss": 1.0387,
+      "step": 4253
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.3238445741327e-07,
+      "loss": 1.0019,
+      "step": 4254
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.297306918277559e-07,
+      "loss": 1.0833,
+      "step": 4255
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.270809801904301e-07,
+      "loss": 1.0949,
+      "step": 4256
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.24435323672661e-07,
+      "loss": 1.0805,
+      "step": 4257
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.21793723444032e-07,
+      "loss": 1.0721,
+      "step": 4258
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.191561806723236e-07,
+      "loss": 1.0737,
+      "step": 4259
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.165226965235328e-07,
+      "loss": 1.073,
+      "step": 4260
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.138932721618542e-07,
+      "loss": 1.1467,
+      "step": 4261
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.112679087496933e-07,
+      "loss": 1.0654,
+      "step": 4262
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.086466074476562e-07,
+      "loss": 1.0771,
+      "step": 4263
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.060293694145571e-07,
+      "loss": 1.0826,
+      "step": 4264
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.034161958074082e-07,
+      "loss": 1.0393,
+      "step": 4265
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.008070877814323e-07,
+      "loss": 1.0991,
+      "step": 4266
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.982020464900486e-07,
+      "loss": 1.1131,
+      "step": 4267
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.956010730848817e-07,
+      "loss": 1.0952,
+      "step": 4268
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.930041687157608e-07,
+      "loss": 1.0674,
+      "step": 4269
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.904113345307073e-07,
+      "loss": 1.1093,
+      "step": 4270
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.878225716759535e-07,
+      "loss": 1.0596,
+      "step": 4271
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.852378812959227e-07,
+      "loss": 1.0622,
+      "step": 4272
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.826572645332475e-07,
+      "loss": 1.086,
+      "step": 4273
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.800807225287498e-07,
+      "loss": 1.1137,
+      "step": 4274
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.775082564214575e-07,
+      "loss": 1.0458,
+      "step": 4275
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.749398673485919e-07,
+      "loss": 1.0691,
+      "step": 4276
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.723755564455771e-07,
+      "loss": 1.0872,
+      "step": 4277
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.698153248460271e-07,
+      "loss": 1.1093,
+      "step": 4278
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.672591736817603e-07,
+      "loss": 0.9902,
+      "step": 4279
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.647071040827869e-07,
+      "loss": 1.0506,
+      "step": 4280
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.621591171773102e-07,
+      "loss": 1.0587,
+      "step": 4281
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.596152140917368e-07,
+      "loss": 1.1722,
+      "step": 4282
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.570753959506582e-07,
+      "loss": 1.1161,
+      "step": 4283
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.545396638768698e-07,
+      "loss": 1.0788,
+      "step": 4284
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.520080189913503e-07,
+      "loss": 1.0335,
+      "step": 4285
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.494804624132834e-07,
+      "loss": 1.0327,
+      "step": 4286
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.46956995260033e-07,
+      "loss": 1.0616,
+      "step": 4287
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.444376186471658e-07,
+      "loss": 1.009,
+      "step": 4288
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.419223336884307e-07,
+      "loss": 1.0723,
+      "step": 4289
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.394111414957783e-07,
+      "loss": 1.1084,
+      "step": 4290
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.369040431793406e-07,
+      "loss": 1.0816,
+      "step": 4291
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.344010398474455e-07,
+      "loss": 1.1599,
+      "step": 4292
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.319021326066055e-07,
+      "loss": 1.0212,
+      "step": 4293
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.294073225615294e-07,
+      "loss": 0.9687,
+      "step": 4294
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.269166108151071e-07,
+      "loss": 1.0086,
+      "step": 4295
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.244299984684233e-07,
+      "loss": 1.1387,
+      "step": 4296
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.219474866207465e-07,
+      "loss": 1.0583,
+      "step": 4297
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.194690763695312e-07,
+      "loss": 1.1118,
+      "step": 4298
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.169947688104251e-07,
+      "loss": 1.1298,
+      "step": 4299
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.145245650372546e-07,
+      "loss": 1.0346,
+      "step": 4300
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.120584661420382e-07,
+      "loss": 0.9563,
+      "step": 4301
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.095964732149741e-07,
+      "loss": 1.0328,
+      "step": 4302
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.071385873444503e-07,
+      "loss": 1.1252,
+      "step": 4303
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.046848096170356e-07,
+      "loss": 1.064,
+      "step": 4304
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.022351411174866e-07,
+      "loss": 1.136,
+      "step": 4305
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.997895829287382e-07,
+      "loss": 1.0736,
+      "step": 4306
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.973481361319124e-07,
+      "loss": 1.1041,
+      "step": 4307
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.949108018063111e-07,
+      "loss": 1.0741,
+      "step": 4308
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.924775810294204e-07,
+      "loss": 1.0952,
+      "step": 4309
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.900484748769054e-07,
+      "loss": 1.0924,
+      "step": 4310
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.876234844226159e-07,
+      "loss": 1.1453,
+      "step": 4311
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.852026107385756e-07,
+      "loss": 1.0934,
+      "step": 4312
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.827858548949984e-07,
+      "loss": 1.0825,
+      "step": 4313
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.803732179602685e-07,
+      "loss": 1.0495,
+      "step": 4314
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.779647010009516e-07,
+      "loss": 1.1144,
+      "step": 4315
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.755603050817972e-07,
+      "loss": 1.0693,
+      "step": 4316
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.731600312657238e-07,
+      "loss": 1.0837,
+      "step": 4317
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.707638806138372e-07,
+      "loss": 1.0177,
+      "step": 4318
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.683718541854134e-07,
+      "loss": 1.0895,
+      "step": 4319
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.659839530379108e-07,
+      "loss": 1.0843,
+      "step": 4320
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.636001782269574e-07,
+      "loss": 1.0679,
+      "step": 4321
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.612205308063646e-07,
+      "loss": 1.1011,
+      "step": 4322
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.588450118281109e-07,
+      "loss": 1.0625,
+      "step": 4323
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.564736223423596e-07,
+      "loss": 1.0377,
+      "step": 4324
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.541063633974376e-07,
+      "loss": 1.0731,
+      "step": 4325
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.517432360398556e-07,
+      "loss": 1.0164,
+      "step": 4326
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.493842413142915e-07,
+      "loss": 1.0589,
+      "step": 4327
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.470293802635985e-07,
+      "loss": 1.0339,
+      "step": 4328
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.446786539288019e-07,
+      "loss": 1.1146,
+      "step": 4329
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.423320633491003e-07,
+      "loss": 1.1371,
+      "step": 4330
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.399896095618629e-07,
+      "loss": 1.1351,
+      "step": 4331
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.37651293602628e-07,
+      "loss": 1.094,
+      "step": 4332
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.353171165051109e-07,
+      "loss": 1.1229,
+      "step": 4333
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.329870793011894e-07,
+      "loss": 1.1021,
+      "step": 4334
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.306611830209186e-07,
+      "loss": 1.1144,
+      "step": 4335
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.28339428692516e-07,
+      "loss": 1.0807,
+      "step": 4336
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.260218173423749e-07,
+      "loss": 1.0541,
+      "step": 4337
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.237083499950514e-07,
+      "loss": 1.0485,
+      "step": 4338
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.213990276732729e-07,
+      "loss": 1.0877,
+      "step": 4339
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.190938513979317e-07,
+      "loss": 1.0415,
+      "step": 4340
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.167928221880925e-07,
+      "loss": 0.977,
+      "step": 4341
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.144959410609785e-07,
+      "loss": 1.0557,
+      "step": 4342
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.122032090319884e-07,
+      "loss": 1.0261,
+      "step": 4343
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.099146271146783e-07,
+      "loss": 1.0225,
+      "step": 4344
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.076301963207754e-07,
+      "loss": 1.0951,
+      "step": 4345
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.053499176601707e-07,
+      "loss": 1.0947,
+      "step": 4346
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.030737921409169e-07,
+      "loss": 1.1085,
+      "step": 4347
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.008018207692335e-07,
+      "loss": 1.1159,
+      "step": 4348
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.985340045495003e-07,
+      "loss": 1.1518,
+      "step": 4349
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.962703444842655e-07,
+      "loss": 1.0741,
+      "step": 4350
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.940108415742352e-07,
+      "loss": 1.017,
+      "step": 4351
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.917554968182803e-07,
+      "loss": 1.0862,
+      "step": 4352
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.895043112134324e-07,
+      "loss": 1.0129,
+      "step": 4353
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.872572857548853e-07,
+      "loss": 1.0527,
+      "step": 4354
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.850144214359921e-07,
+      "loss": 1.0527,
+      "step": 4355
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.827757192482686e-07,
+      "loss": 1.1044,
+      "step": 4356
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.805411801813865e-07,
+      "loss": 1.1064,
+      "step": 4357
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.783108052231845e-07,
+      "loss": 1.0869,
+      "step": 4358
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.760845953596527e-07,
+      "loss": 1.0432,
+      "step": 4359
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.73862551574943e-07,
+      "loss": 1.0238,
+      "step": 4360
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.716446748513682e-07,
+      "loss": 1.074,
+      "step": 4361
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.694309661693942e-07,
+      "loss": 1.1116,
+      "step": 4362
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.672214265076503e-07,
+      "loss": 1.032,
+      "step": 4363
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.650160568429141e-07,
+      "loss": 1.0175,
+      "step": 4364
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.628148581501281e-07,
+      "loss": 1.0253,
+      "step": 4365
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.606178314023836e-07,
+      "loss": 1.0752,
+      "step": 4366
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.584249775709372e-07,
+      "loss": 1.0867,
+      "step": 4367
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.562362976251901e-07,
+      "loss": 1.0965,
+      "step": 4368
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.540517925327071e-07,
+      "loss": 1.1178,
+      "step": 4369
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.518714632592003e-07,
+      "loss": 1.1149,
+      "step": 4370
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.496953107685421e-07,
+      "loss": 1.0589,
+      "step": 4371
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.475233360227516e-07,
+      "loss": 1.1074,
+      "step": 4372
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.453555399820088e-07,
+      "loss": 1.1122,
+      "step": 4373
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.431919236046424e-07,
+      "loss": 1.1267,
+      "step": 4374
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.410324878471296e-07,
+      "loss": 1.1078,
+      "step": 4375
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.38877233664108e-07,
+      "loss": 1.0746,
+      "step": 4376
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.367261620083575e-07,
+      "loss": 1.0053,
+      "step": 4377
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.345792738308176e-07,
+      "loss": 1.0571,
+      "step": 4378
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.324365700805701e-07,
+      "loss": 1.1629,
+      "step": 4379
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.302980517048562e-07,
+      "loss": 1.1191,
+      "step": 4380
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.281637196490552e-07,
+      "loss": 1.1176,
+      "step": 4381
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.26033574856708e-07,
+      "loss": 1.0761,
+      "step": 4382
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.239076182694946e-07,
+      "loss": 1.0989,
+      "step": 4383
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.217858508272511e-07,
+      "loss": 1.0802,
+      "step": 4384
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.196682734679559e-07,
+      "loss": 1.1365,
+      "step": 4385
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.175548871277358e-07,
+      "loss": 1.0804,
+      "step": 4386
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.154456927408713e-07,
+      "loss": 1.1298,
+      "step": 4387
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.133406912397809e-07,
+      "loss": 1.0725,
+      "step": 4388
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.112398835550348e-07,
+      "loss": 1.1049,
+      "step": 4389
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.091432706153465e-07,
+      "loss": 1.0916,
+      "step": 4390
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.070508533475793e-07,
+      "loss": 1.0811,
+      "step": 4391
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.049626326767366e-07,
+      "loss": 1.1615,
+      "step": 4392
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.028786095259697e-07,
+      "loss": 1.103,
+      "step": 4393
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.007987848165718e-07,
+      "loss": 1.1558,
+      "step": 4394
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.987231594679853e-07,
+      "loss": 1.0479,
+      "step": 4395
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.966517343977884e-07,
+      "loss": 1.0645,
+      "step": 4396
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.945845105217118e-07,
+      "loss": 1.0916,
+      "step": 4397
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.925214887536167e-07,
+      "loss": 1.0788,
+      "step": 4398
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.904626700055193e-07,
+      "loss": 1.0376,
+      "step": 4399
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.884080551875714e-07,
+      "loss": 1.1407,
+      "step": 4400
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.863576452080654e-07,
+      "loss": 1.0837,
+      "step": 4401
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.843114409734384e-07,
+      "loss": 1.0868,
+      "step": 4402
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.822694433882635e-07,
+      "loss": 1.0841,
+      "step": 4403
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.80231653355262e-07,
+      "loss": 1.09,
+      "step": 4404
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.781980717752855e-07,
+      "loss": 1.0937,
+      "step": 4405
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.761686995473336e-07,
+      "loss": 1.0321,
+      "step": 4406
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7414353756853773e-07,
+      "loss": 1.0135,
+      "step": 4407
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.72122586734175e-07,
+      "loss": 1.0907,
+      "step": 4408
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7010584793765524e-07,
+      "loss": 1.0767,
+      "step": 4409
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.6809332207053083e-07,
+      "loss": 1.0597,
+      "step": 4410
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.660850100224856e-07,
+      "loss": 1.0261,
+      "step": 4411
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6408091268134836e-07,
+      "loss": 1.0699,
+      "step": 4412
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.620810309330803e-07,
+      "loss": 1.0418,
+      "step": 4413
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.600853656617765e-07,
+      "loss": 1.0344,
+      "step": 4414
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.580939177496735e-07,
+      "loss": 1.1077,
+      "step": 4415
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.561066880771392e-07,
+      "loss": 1.1008,
+      "step": 4416
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5412367752268094e-07,
+      "loss": 1.0341,
+      "step": 4417
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5214488696293634e-07,
+      "loss": 1.1435,
+      "step": 4418
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5017031727268124e-07,
+      "loss": 1.0802,
+      "step": 4419
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4819996932482067e-07,
+      "loss": 1.101,
+      "step": 4420
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4623384399040017e-07,
+      "loss": 1.084,
+      "step": 4421
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4427194213859216e-07,
+      "loss": 1.1361,
+      "step": 4422
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4231426463670736e-07,
+      "loss": 1.0527,
+      "step": 4423
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4036081235018347e-07,
+      "loss": 1.056,
+      "step": 4424
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.3841158614259636e-07,
+      "loss": 1.0093,
+      "step": 4425
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.364665868756468e-07,
+      "loss": 1.0683,
+      "step": 4426
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.345258154091747e-07,
+      "loss": 1.1054,
+      "step": 4427
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.325892726011427e-07,
+      "loss": 1.0756,
+      "step": 4428
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.306569593076515e-07,
+      "loss": 1.0451,
+      "step": 4429
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.287288763829278e-07,
+      "loss": 1.1047,
+      "step": 4430
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.268050246793276e-07,
+      "loss": 1.1302,
+      "step": 4431
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2488540504734056e-07,
+      "loss": 1.1509,
+      "step": 4432
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2297001833558005e-07,
+      "loss": 1.1075,
+      "step": 4433
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.210588653907943e-07,
+      "loss": 1.0511,
+      "step": 4434
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.191519470578531e-07,
+      "loss": 1.1139,
+      "step": 4435
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1724926417976097e-07,
+      "loss": 1.1159,
+      "step": 4436
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1535081759764286e-07,
+      "loss": 1.0826,
+      "step": 4437
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.134566081507585e-07,
+      "loss": 1.0596,
+      "step": 4438
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.115666366764881e-07,
+      "loss": 1.0863,
+      "step": 4439
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0968090401034444e-07,
+      "loss": 1.0749,
+      "step": 4440
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0779941098595844e-07,
+      "loss": 0.9786,
+      "step": 4441
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0592215843509585e-07,
+      "loss": 1.0903,
+      "step": 4442
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0404914718763955e-07,
+      "loss": 0.9429,
+      "step": 4443
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0218037807160495e-07,
+      "loss": 1.0577,
+      "step": 4444
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.003158519131245e-07,
+      "loss": 1.0517,
+      "step": 4445
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9845556953646334e-07,
+      "loss": 1.0385,
+      "step": 4446
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.965995317640026e-07,
+      "loss": 1.0455,
+      "step": 4447
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9474773941625133e-07,
+      "loss": 1.033,
+      "step": 4448
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9290019331184147e-07,
+      "loss": 1.1096,
+      "step": 4449
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.910568942675264e-07,
+      "loss": 1.1297,
+      "step": 4450
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.892178430981841e-07,
+      "loss": 1.0732,
+      "step": 4451
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8738304061681107e-07,
+      "loss": 1.0713,
+      "step": 4452
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.855524876345318e-07,
+      "loss": 1.1293,
+      "step": 4453
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.837261849605844e-07,
+      "loss": 1.04,
+      "step": 4454
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8190413340233436e-07,
+      "loss": 1.1086,
+      "step": 4455
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.800863337652638e-07,
+      "loss": 1.077,
+      "step": 4456
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7827278685297785e-07,
+      "loss": 1.1428,
+      "step": 4457
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.764634934671996e-07,
+      "loss": 1.1456,
+      "step": 4458
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.746584544077736e-07,
+      "loss": 1.1158,
+      "step": 4459
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.728576704726616e-07,
+      "loss": 1.1714,
+      "step": 4460
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.71061142457948e-07,
+      "loss": 1.0569,
+      "step": 4461
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.692688711578296e-07,
+      "loss": 1.1196,
+      "step": 4462
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6748085736462825e-07,
+      "loss": 1.0304,
+      "step": 4463
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6569710186877937e-07,
+      "loss": 1.0559,
+      "step": 4464
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6391760545883556e-07,
+      "loss": 1.0162,
+      "step": 4465
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6214236892146983e-07,
+      "loss": 1.0748,
+      "step": 4466
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.603713930414676e-07,
+      "loss": 1.0591,
+      "step": 4467
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.586046786017361e-07,
+      "loss": 1.082,
+      "step": 4468
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.568422263832938e-07,
+      "loss": 1.0623,
+      "step": 4469
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5508403716527975e-07,
+      "loss": 1.0798,
+      "step": 4470
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5333011172494213e-07,
+      "loss": 1.0158,
+      "step": 4471
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.515804508376508e-07,
+      "loss": 1.161,
+      "step": 4472
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.498350552768859e-07,
+      "loss": 1.1215,
+      "step": 4473
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.480939258142435e-07,
+      "loss": 1.1461,
+      "step": 4474
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4635706321943355e-07,
+      "loss": 1.0661,
+      "step": 4475
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.446244682602817e-07,
+      "loss": 1.0558,
+      "step": 4476
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.428961417027221e-07,
+      "loss": 1.0432,
+      "step": 4477
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4117208431080574e-07,
+      "loss": 1.1051,
+      "step": 4478
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.394522968466984e-07,
+      "loss": 1.0602,
+      "step": 4479
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.377367800706732e-07,
+      "loss": 1.0821,
+      "step": 4480
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3602553474111674e-07,
+      "loss": 1.1085,
+      "step": 4481
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3431856161452835e-07,
+      "loss": 1.0665,
+      "step": 4482
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3261586144551995e-07,
+      "loss": 1.1326,
+      "step": 4483
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.309174349868105e-07,
+      "loss": 1.0268,
+      "step": 4484
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.292232829892361e-07,
+      "loss": 1.0111,
+      "step": 4485
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2753340620173433e-07,
+      "loss": 1.1007,
+      "step": 4486
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2584780537136206e-07,
+      "loss": 1.0153,
+      "step": 4487
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2416648124327767e-07,
+      "loss": 1.0646,
+      "step": 4488
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2248943456075655e-07,
+      "loss": 1.0907,
+      "step": 4489
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.208166660651757e-07,
+      "loss": 1.0419,
+      "step": 4490
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.1914817649602694e-07,
+      "loss": 1.1775,
+      "step": 4491
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.1748396659090797e-07,
+      "loss": 1.054,
+      "step": 4492
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1582403708552366e-07,
+      "loss": 1.0885,
+      "step": 4493
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1416838871368925e-07,
+      "loss": 1.0553,
+      "step": 4494
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.125170222073226e-07,
+      "loss": 1.0517,
+      "step": 4495
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.108699382964553e-07,
+      "loss": 1.0066,
+      "step": 4496
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0922713770922155e-07,
+      "loss": 1.0583,
+      "step": 4497
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.075886211718604e-07,
+      "loss": 1.0691,
+      "step": 4498
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0595438940872023e-07,
+      "loss": 1.0347,
+      "step": 4499
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.043244431422565e-07,
+      "loss": 1.067,
+      "step": 4500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.026987830930239e-07,
+      "loss": 1.136,
+      "step": 4501
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.010774099796898e-07,
+      "loss": 1.06,
+      "step": 4502
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9946032451902195e-07,
+      "loss": 1.0552,
+      "step": 4503
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9784752742589294e-07,
+      "loss": 1.0922,
+      "step": 4504
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9623901941328135e-07,
+      "loss": 1.0631,
+      "step": 4505
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.946348011922673e-07,
+      "loss": 1.0656,
+      "step": 4506
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.930348734720379e-07,
+      "loss": 1.114,
+      "step": 4507
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9143923695987955e-07,
+      "loss": 1.1293,
+      "step": 4508
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.898478923611847e-07,
+      "loss": 1.0217,
+      "step": 4509
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8826084037944513e-07,
+      "loss": 1.0832,
+      "step": 4510
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8667808171625953e-07,
+      "loss": 1.0175,
+      "step": 4511
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8509961707132496e-07,
+      "loss": 1.0914,
+      "step": 4512
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8352544714244315e-07,
+      "loss": 1.0013,
+      "step": 4513
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8195557262551297e-07,
+      "loss": 1.0442,
+      "step": 4514
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8038999421453827e-07,
+      "loss": 1.157,
+      "step": 4515
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7882871260162204e-07,
+      "loss": 1.0705,
+      "step": 4516
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.772717284769677e-07,
+      "loss": 1.0265,
+      "step": 4517
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.757190425288803e-07,
+      "loss": 1.1042,
+      "step": 4518
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7417065544376175e-07,
+      "loss": 1.1674,
+      "step": 4519
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7262656790611907e-07,
+      "loss": 1.0681,
+      "step": 4520
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7108678059855064e-07,
+      "loss": 1.0737,
+      "step": 4521
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6955129420176193e-07,
+      "loss": 1.0687,
+      "step": 4522
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6802010939455114e-07,
+      "loss": 1.1302,
+      "step": 4523
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6649322685381784e-07,
+      "loss": 1.0578,
+      "step": 4524
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.649706472545577e-07,
+      "loss": 1.048,
+      "step": 4525
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.634523712698678e-07,
+      "loss": 1.0957,
+      "step": 4526
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6193839957093683e-07,
+      "loss": 1.0937,
+      "step": 4527
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.604287328270572e-07,
+      "loss": 1.1086,
+      "step": 4528
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.589233717056128e-07,
+      "loss": 1.1473,
+      "step": 4529
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5742231687209016e-07,
+      "loss": 1.1087,
+      "step": 4530
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5592556899006193e-07,
+      "loss": 1.1032,
+      "step": 4531
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5443312872120763e-07,
+      "loss": 1.144,
+      "step": 4532
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.529449967252984e-07,
+      "loss": 1.0077,
+      "step": 4533
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5146117366019794e-07,
+      "loss": 1.0681,
+      "step": 4534
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.499816601818694e-07,
+      "loss": 1.0561,
+      "step": 4535
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4850645694436736e-07,
+      "loss": 1.1067,
+      "step": 4536
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4703556459984456e-07,
+      "loss": 1.0391,
+      "step": 4537
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4556898379854424e-07,
+      "loss": 1.145,
+      "step": 4538
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.441067151888066e-07,
+      "loss": 1.0569,
+      "step": 4539
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4264875941706236e-07,
+      "loss": 1.1197,
+      "step": 4540
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.411951171278393e-07,
+      "loss": 1.1141,
+      "step": 4541
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3974578896375555e-07,
+      "loss": 1.1202,
+      "step": 4542
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3830077556552424e-07,
+      "loss": 1.0838,
+      "step": 4543
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3686007757194873e-07,
+      "loss": 1.0998,
+      "step": 4544
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.354236956199263e-07,
+      "loss": 1.0513,
+      "step": 4545
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.339916303444456e-07,
+      "loss": 1.1201,
+      "step": 4546
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3256388237858806e-07,
+      "loss": 1.0148,
+      "step": 4547
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3114045235352433e-07,
+      "loss": 1.1291,
+      "step": 4548
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.297213408985166e-07,
+      "loss": 1.1005,
+      "step": 4549
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2830654864092083e-07,
+      "loss": 1.1167,
+      "step": 4550
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2689607620618003e-07,
+      "loss": 1.0682,
+      "step": 4551
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.25489924217831e-07,
+      "loss": 1.1151,
+      "step": 4552
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2408809329749648e-07,
+      "loss": 1.0212,
+      "step": 4553
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2269058406489185e-07,
+      "loss": 1.1242,
+      "step": 4554
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2129739713782071e-07,
+      "loss": 1.0539,
+      "step": 4555
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1990853313217818e-07,
+      "loss": 1.0765,
+      "step": 4556
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1852399266194312e-07,
+      "loss": 1.1269,
+      "step": 4557
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1714377633919038e-07,
+      "loss": 1.0473,
+      "step": 4558
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.157678847740774e-07,
+      "loss": 1.0732,
+      "step": 4559
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.14396318574851e-07,
+      "loss": 1.0609,
+      "step": 4560
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1302907834784725e-07,
+      "loss": 1.1106,
+      "step": 4561
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1166616469749047e-07,
+      "loss": 1.0903,
+      "step": 4562
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1030757822628756e-07,
+      "loss": 1.089,
+      "step": 4563
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.089533195348392e-07,
+      "loss": 1.097,
+      "step": 4564
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0760338922182765e-07,
+      "loss": 1.0109,
+      "step": 4565
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.062577878840244e-07,
+      "loss": 1.119,
+      "step": 4566
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0491651611628582e-07,
+      "loss": 1.1109,
+      "step": 4567
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.035795745115543e-07,
+      "loss": 1.1128,
+      "step": 4568
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.022469636608604e-07,
+      "loss": 0.9993,
+      "step": 4569
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0091868415331505e-07,
+      "loss": 1.1202,
+      "step": 4570
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9959473657612193e-07,
+      "loss": 1.0906,
+      "step": 4571
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9827512151456175e-07,
+      "loss": 1.072,
+      "step": 4572
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9695983955200558e-07,
+      "loss": 1.0362,
+      "step": 4573
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9564889126990505e-07,
+      "loss": 1.1357,
+      "step": 4574
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9434227724779987e-07,
+      "loss": 1.1519,
+      "step": 4575
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9303999806331019e-07,
+      "loss": 1.1298,
+      "step": 4576
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.917420542921433e-07,
+      "loss": 1.0555,
+      "step": 4577
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9044844650808468e-07,
+      "loss": 1.0955,
+      "step": 4578
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8915917528300798e-07,
+      "loss": 1.1447,
+      "step": 4579
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8787424118686948e-07,
+      "loss": 1.0618,
+      "step": 4580
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.865936447877026e-07,
+      "loss": 1.1263,
+      "step": 4581
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8531738665163112e-07,
+      "loss": 1.0633,
+      "step": 4582
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8404546734285488e-07,
+      "loss": 1.1315,
+      "step": 4583
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8277788742365966e-07,
+      "loss": 1.0848,
+      "step": 4584
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8151464745440828e-07,
+      "loss": 1.0401,
+      "step": 4585
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8025574799354961e-07,
+      "loss": 1.1319,
+      "step": 4586
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7900118959761181e-07,
+      "loss": 1.1162,
+      "step": 4587
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7775097282120346e-07,
+      "loss": 1.083,
+      "step": 4588
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7650509821701467e-07,
+      "loss": 1.1055,
+      "step": 4589
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.75263566335816e-07,
+      "loss": 1.0894,
+      "step": 4590
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7402637772645613e-07,
+      "loss": 1.0381,
+      "step": 4591
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7279353293586765e-07,
+      "loss": 1.0297,
+      "step": 4592
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.71565032509059e-07,
+      "loss": 1.0663,
+      "step": 4593
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7034087698912016e-07,
+      "loss": 1.0621,
+      "step": 4594
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6912106691722053e-07,
+      "loss": 1.0242,
+      "step": 4595
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6790560283260647e-07,
+      "loss": 1.0814,
+      "step": 4596
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6669448527260602e-07,
+      "loss": 1.0368,
+      "step": 4597
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6548771477262304e-07,
+      "loss": 1.0988,
+      "step": 4598
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6428529186614195e-07,
+      "loss": 1.1154,
+      "step": 4599
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6308721708472196e-07,
+      "loss": 1.0729,
+      "step": 4600
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6189349095800388e-07,
+      "loss": 1.0529,
+      "step": 4601
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6070411401370335e-07,
+      "loss": 1.0804,
+      "step": 4602
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.595190867776164e-07,
+      "loss": 1.1366,
+      "step": 4603
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.583384097736107e-07,
+      "loss": 1.0769,
+      "step": 4604
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.571620835236376e-07,
+      "loss": 1.1093,
+      "step": 4605
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5599010854772002e-07,
+      "loss": 1.1756,
+      "step": 4606
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5482248536395904e-07,
+      "loss": 1.0898,
+      "step": 4607
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5365921448853183e-07,
+      "loss": 1.0788,
+      "step": 4608
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5250029643569142e-07,
+      "loss": 1.0924,
+      "step": 4609
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.513457317177669e-07,
+      "loss": 0.9625,
+      "step": 4610
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5019552084516332e-07,
+      "loss": 1.0212,
+      "step": 4611
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4904966432635947e-07,
+      "loss": 1.0912,
+      "step": 4612
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4790816266791018e-07,
+      "loss": 1.0455,
+      "step": 4613
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4677101637444734e-07,
+      "loss": 1.0902,
+      "step": 4614
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4563822594866993e-07,
+      "loss": 0.9559,
+      "step": 4615
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4450979189136072e-07,
+      "loss": 1.0698,
+      "step": 4616
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4338571470137063e-07,
+      "loss": 1.1282,
+      "step": 4617
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4226599487562665e-07,
+      "loss": 1.0411,
+      "step": 4618
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4115063290912833e-07,
+      "loss": 1.0905,
+      "step": 4619
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.400396292949513e-07,
+      "loss": 1.0822,
+      "step": 4620
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.3893298452424043e-07,
+      "loss": 1.0613,
+      "step": 4621
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.3783069908621772e-07,
+      "loss": 1.0523,
+      "step": 4622
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.367327734681756e-07,
+      "loss": 1.032,
+      "step": 4623
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3563920815547915e-07,
+      "loss": 1.051,
+      "step": 4624
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3455000363156723e-07,
+      "loss": 1.0183,
+      "step": 4625
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.334651603779491e-07,
+      "loss": 1.0491,
+      "step": 4626
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.323846788742078e-07,
+      "loss": 1.0337,
+      "step": 4627
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3130855959799683e-07,
+      "loss": 1.0758,
+      "step": 4628
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3023680302504338e-07,
+      "loss": 1.0868,
+      "step": 4629
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2916940962914292e-07,
+      "loss": 1.0554,
+      "step": 4630
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2810637988216358e-07,
+      "loss": 1.0294,
+      "step": 4631
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2704771425404382e-07,
+      "loss": 1.0732,
+      "step": 4632
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2599341321279379e-07,
+      "loss": 1.0887,
+      "step": 4633
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2494347722449506e-07,
+      "loss": 1.0899,
+      "step": 4634
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.238979067532975e-07,
+      "loss": 1.0543,
+      "step": 4635
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.228567022614202e-07,
+      "loss": 1.1219,
+      "step": 4636
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2181986420915615e-07,
+      "loss": 1.0259,
+      "step": 4637
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2078739305486532e-07,
+      "loss": 1.0595,
+      "step": 4638
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1975928925497592e-07,
+      "loss": 1.1573,
+      "step": 4639
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1873555326398889e-07,
+      "loss": 0.9989,
+      "step": 4640
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1771618553447217e-07,
+      "loss": 1.0441,
+      "step": 4641
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1670118651706197e-07,
+      "loss": 1.095,
+      "step": 4642
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1569055666046381e-07,
+      "loss": 1.1098,
+      "step": 4643
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1468429641145474e-07,
+      "loss": 1.0957,
+      "step": 4644
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1368240621487447e-07,
+      "loss": 1.0539,
+      "step": 4645
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.126848865136343e-07,
+      "loss": 1.1184,
+      "step": 4646
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1169173774871478e-07,
+      "loss": 1.0632,
+      "step": 4647
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1070296035916028e-07,
+      "loss": 1.0426,
+      "step": 4648
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.097185547820856e-07,
+      "loss": 1.0383,
+      "step": 4649
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.087385214526715e-07,
+      "loss": 1.0243,
+      "step": 4650
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0776286080416698e-07,
+      "loss": 1.111,
+      "step": 4651
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0679157326788592e-07,
+      "loss": 1.0583,
+      "step": 4652
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0582465927321372e-07,
+      "loss": 0.9831,
+      "step": 4653
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0486211924759515e-07,
+      "loss": 1.1081,
+      "step": 4654
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.039039536165476e-07,
+      "loss": 1.0914,
+      "step": 4655
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0295016280365111e-07,
+      "loss": 1.138,
+      "step": 4656
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0200074723055397e-07,
+      "loss": 1.0804,
+      "step": 4657
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0105570731696712e-07,
+      "loss": 1.1037,
+      "step": 4658
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0011504348067303e-07,
+      "loss": 1.0883,
+      "step": 4659
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.917875613751127e-08,
+      "loss": 1.0372,
+      "step": 4660
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.82468457013952e-08,
+      "loss": 1.1238,
+      "step": 4661
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.731931258429638e-08,
+      "loss": 1.0869,
+      "step": 4662
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.639615719625683e-08,
+      "loss": 1.1722,
+      "step": 4663
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.547737994537787e-08,
+      "loss": 1.0382,
+      "step": 4664
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.456298123782903e-08,
+      "loss": 1.0356,
+      "step": 4665
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.365296147784475e-08,
+      "loss": 1.0879,
+      "step": 4666
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.274732106771989e-08,
+      "loss": 1.1027,
+      "step": 4667
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.184606040781641e-08,
+      "loss": 1.081,
+      "step": 4668
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.094917989656005e-08,
+      "loss": 1.1136,
+      "step": 4669
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.005667993043921e-08,
+      "loss": 1.1213,
+      "step": 4670
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.916856090400383e-08,
+      "loss": 1.14,
+      "step": 4671
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.82848232098732e-08,
+      "loss": 1.1363,
+      "step": 4672
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.740546723872145e-08,
+      "loss": 1.0355,
+      "step": 4673
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.65304933792932e-08,
+      "loss": 1.1118,
+      "step": 4674
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.565990201839125e-08,
+      "loss": 1.0963,
+      "step": 4675
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.479369354088329e-08,
+      "loss": 1.0826,
+      "step": 4676
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.393186832969746e-08,
+      "loss": 1.0427,
+      "step": 4677
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.307442676582678e-08,
+      "loss": 1.0249,
+      "step": 4678
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.222136922832246e-08,
+      "loss": 1.1386,
+      "step": 4679
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.137269609430176e-08,
+      "loss": 1.078,
+      "step": 4680
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.052840773894233e-08,
+      "loss": 1.0713,
+      "step": 4681
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.968850453548227e-08,
+      "loss": 1.044,
+      "step": 4682
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.885298685522235e-08,
+      "loss": 1.1025,
+      "step": 4683
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.802185506752269e-08,
+      "loss": 1.0962,
+      "step": 4684
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.719510953980824e-08,
+      "loss": 1.0879,
+      "step": 4685
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.637275063756112e-08,
+      "loss": 1.0517,
+      "step": 4686
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.555477872432715e-08,
+      "loss": 1.0971,
+      "step": 4687
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.474119416170933e-08,
+      "loss": 1.0925,
+      "step": 4688
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.39319973093744e-08,
+      "loss": 1.0313,
+      "step": 4689
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.31271885250484e-08,
+      "loss": 1.1236,
+      "step": 4690
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.232676816451789e-08,
+      "loss": 1.0606,
+      "step": 4691
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.153073658162646e-08,
+      "loss": 1.0956,
+      "step": 4692
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.073909412828262e-08,
+      "loss": 1.0001,
+      "step": 4693
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.995184115444976e-08,
+      "loss": 1.0904,
+      "step": 4694
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.916897800815504e-08,
+      "loss": 1.1199,
+      "step": 4695
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.839050503547939e-08,
+      "loss": 1.1189,
+      "step": 4696
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.761642258056977e-08,
+      "loss": 1.1108,
+      "step": 4697
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.684673098562799e-08,
+      "loss": 1.0535,
+      "step": 4698
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.608143059091409e-08,
+      "loss": 1.1267,
+      "step": 4699
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.532052173474968e-08,
+      "loss": 1.1124,
+      "step": 4700
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.456400475351233e-08,
+      "loss": 1.0484,
+      "step": 4701
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.381187998164229e-08,
+      "loss": 1.0987,
+      "step": 4702
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.30641477516314e-08,
+      "loss": 1.0926,
+      "step": 4703
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.232080839403631e-08,
+      "loss": 1.0933,
+      "step": 4704
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.158186223746754e-08,
+      "loss": 1.0778,
+      "step": 4705
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.084730960859598e-08,
+      "loss": 1.0744,
+      "step": 4706
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.011715083214742e-08,
+      "loss": 1.0586,
+      "step": 4707
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.939138623090923e-08,
+      "loss": 1.0568,
+      "step": 4708
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.867001612572143e-08,
+      "loss": 1.0455,
+      "step": 4709
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.795304083548559e-08,
+      "loss": 1.1012,
+      "step": 4710
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.724046067715705e-08,
+      "loss": 1.0639,
+      "step": 4711
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.6532275965751614e-08,
+      "loss": 1.044,
+      "step": 4712
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.5828487014338853e-08,
+      "loss": 1.0738,
+      "step": 4713
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.5129094134046544e-08,
+      "loss": 1.016,
+      "step": 4714
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.44340976340596e-08,
+      "loss": 1.0606,
+      "step": 4715
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.3743497821617806e-08,
+      "loss": 1.0011,
+      "step": 4716
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.305729500201917e-08,
+      "loss": 1.0906,
+      "step": 4717
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.2375489478616593e-08,
+      "loss": 1.07,
+      "step": 4718
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.169808155281786e-08,
+      "loss": 1.0137,
+      "step": 4719
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.1025071524090083e-08,
+      "loss": 1.0227,
+      "step": 4720
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.0356459689954176e-08,
+      "loss": 1.0755,
+      "step": 4721
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.9692246345985905e-08,
+      "loss": 1.0815,
+      "step": 4722
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.903243178581818e-08,
+      "loss": 1.0768,
+      "step": 4723
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.837701630113767e-08,
+      "loss": 1.087,
+      "step": 4724
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.772600018168816e-08,
+      "loss": 1.059,
+      "step": 4725
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.707938371526721e-08,
+      "loss": 1.1728,
+      "step": 4726
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.643716718772839e-08,
+      "loss": 1.1393,
+      "step": 4727
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.5799350882980156e-08,
+      "loss": 1.0394,
+      "step": 4728
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.516593508298362e-08,
+      "loss": 1.1084,
+      "step": 4729
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.453692006775701e-08,
+      "loss": 1.0844,
+      "step": 4730
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.391230611537234e-08,
+      "loss": 1.0466,
+      "step": 4731
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.329209350195651e-08,
+      "loss": 1.1575,
+      "step": 4732
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.267628250168798e-08,
+      "loss": 1.0888,
+      "step": 4733
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.206487338680454e-08,
+      "loss": 1.0716,
+      "step": 4734
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.1457866427592194e-08,
+      "loss": 1.0296,
+      "step": 4735
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.085526189239519e-08,
+      "loss": 1.0326,
+      "step": 4736
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.025706004760932e-08,
+      "loss": 1.0115,
+      "step": 4737
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.966326115768637e-08,
+      "loss": 1.107,
+      "step": 4738
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.907386548512748e-08,
+      "loss": 1.0306,
+      "step": 4739
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.8488873290492005e-08,
+      "loss": 1.0818,
+      "step": 4740
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.790828483238862e-08,
+      "loss": 1.066,
+      "step": 4741
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.7332100367482027e-08,
+      "loss": 1.0778,
+      "step": 4742
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.676032015048958e-08,
+      "loss": 1.1209,
+      "step": 4743
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.61929444341802e-08,
+      "loss": 0.9891,
+      "step": 4744
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5629973469377686e-08,
+      "loss": 1.0679,
+      "step": 4745
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.50714075049563e-08,
+      "loss": 1.0421,
+      "step": 4746
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.451724678784518e-08,
+      "loss": 1.1687,
+      "step": 4747
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.396749156302392e-08,
+      "loss": 1.0165,
+      "step": 4748
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.342214207352701e-08,
+      "loss": 1.0552,
+      "step": 4749
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.2881198560438256e-08,
+      "loss": 1.1295,
+      "step": 4750
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.234466126289637e-08,
+      "loss": 1.1439,
+      "step": 4751
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.181253041809052e-08,
+      "loss": 1.0395,
+      "step": 4752
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1284806261264735e-08,
+      "loss": 1.069,
+      "step": 4753
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.076148902570908e-08,
+      "loss": 1.1097,
+      "step": 4754
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.024257894277183e-08,
+      "loss": 1.1036,
+      "step": 4755
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.972807624184948e-08,
+      "loss": 1.0429,
+      "step": 4756
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9217981150390095e-08,
+      "loss": 1.016,
+      "step": 4757
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8712293893896627e-08,
+      "loss": 1.0894,
+      "step": 4758
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8211014695916917e-08,
+      "loss": 1.0459,
+      "step": 4759
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7714143778058146e-08,
+      "loss": 1.1306,
+      "step": 4760
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7221681359971276e-08,
+      "loss": 1.0666,
+      "step": 4761
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6733627659363272e-08,
+      "loss": 1.0969,
+      "step": 4762
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6249982891990433e-08,
+      "loss": 1.0939,
+      "step": 4763
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5770747271659512e-08,
+      "loss": 1.0899,
+      "step": 4764
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.529592101022993e-08,
+      "loss": 1.0825,
+      "step": 4765
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.4825504317609328e-08,
+      "loss": 1.0571,
+      "step": 4766
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.4359497401758026e-08,
+      "loss": 1.1587,
+      "step": 4767
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.3897900468684566e-08,
+      "loss": 1.0767,
+      "step": 4768
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.344071372245238e-08,
+      "loss": 1.1351,
+      "step": 4769
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.29879373651698e-08,
+      "loss": 1.0431,
+      "step": 4770
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2539571596998934e-08,
+      "loss": 1.1194,
+      "step": 4771
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2095616616150117e-08,
+      "loss": 1.0795,
+      "step": 4772
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.165607261888747e-08,
+      "loss": 1.1166,
+      "step": 4773
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1220939799520003e-08,
+      "loss": 1.0524,
+      "step": 4774
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0790218350409397e-08,
+      "loss": 1.011,
+      "step": 4775
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0363908461967786e-08,
+      "loss": 1.1623,
+      "step": 4776
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9942010322655527e-08,
+      "loss": 1.0762,
+      "step": 4777
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9524524118984533e-08,
+      "loss": 1.1097,
+      "step": 4778
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9111450035513844e-08,
+      "loss": 1.1154,
+      "step": 4779
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8702788254852945e-08,
+      "loss": 1.0087,
+      "step": 4780
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.829853895766176e-08,
+      "loss": 1.1017,
+      "step": 4781
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.7898702322648453e-08,
+      "loss": 1.068,
+      "step": 4782
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.750327852657163e-08,
+      "loss": 1.1563,
+      "step": 4783
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.711226774423813e-08,
+      "loss": 0.9503,
+      "step": 4784
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6725670148503015e-08,
+      "loss": 1.0165,
+      "step": 4785
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6343485910272905e-08,
+      "loss": 1.0805,
+      "step": 4786
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.596571519850043e-08,
+      "loss": 1.0673,
+      "step": 4787
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5592358180189782e-08,
+      "loss": 1.036,
+      "step": 4788
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5223415020393374e-08,
+      "loss": 1.1039,
+      "step": 4789
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4858885882210739e-08,
+      "loss": 1.0344,
+      "step": 4790
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.449877092679075e-08,
+      "loss": 1.0552,
+      "step": 4791
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.414307031333273e-08,
+      "loss": 1.0539,
+      "step": 4792
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3791784199080893e-08,
+      "loss": 1.1244,
+      "step": 4793
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3444912739333238e-08,
+      "loss": 1.1037,
+      "step": 4794
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3102456087430437e-08,
+      "loss": 1.1308,
+      "step": 4795
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2764414394764723e-08,
+      "loss": 1.1397,
+      "step": 4796
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2430787810776556e-08,
+      "loss": 1.1122,
+      "step": 4797
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2101576482953514e-08,
+      "loss": 1.0582,
+      "step": 4798
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.1776780556831402e-08,
+      "loss": 1.1956,
+      "step": 4799
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.1456400175994253e-08,
+      "loss": 1.0817,
+      "step": 4800
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.114043548207544e-08,
+      "loss": 1.1282,
+      "step": 4801
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0828886614754342e-08,
+      "loss": 1.0658,
+      "step": 4802
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0521753711759675e-08,
+      "loss": 1.0699,
+      "step": 4803
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0219036908866164e-08,
+      "loss": 1.0787,
+      "step": 4804
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.920736339897874e-09,
+      "loss": 0.9625,
+      "step": 4805
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.626852136725984e-09,
+      "loss": 1.0378,
+      "step": 4806
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.337384429269903e-09,
+      "loss": 1.1167,
+      "step": 4807
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.052333345495045e-09,
+      "loss": 1.1095,
+      "step": 4808
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.771699011416169e-09,
+      "loss": 1.0874,
+      "step": 4809
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.495481551095142e-09,
+      "loss": 1.0577,
+      "step": 4810
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.223681086639846e-09,
+      "loss": 1.0654,
+      "step": 4811
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.956297738207496e-09,
+      "loss": 1.0541,
+      "step": 4812
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.693331624001322e-09,
+      "loss": 1.0726,
+      "step": 4813
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.434782860271661e-09,
+      "loss": 1.0061,
+      "step": 4814
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.180651561315977e-09,
+      "loss": 1.0812,
+      "step": 4815
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.930937839481067e-09,
+      "loss": 1.0455,
+      "step": 4816
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.685641805158627e-09,
+      "loss": 0.9811,
+      "step": 4817
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.444763566786361e-09,
+      "loss": 1.097,
+      "step": 4818
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.208303230852419e-09,
+      "loss": 1.0496,
+      "step": 4819
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.976260901889852e-09,
+      "loss": 1.1274,
+      "step": 4820
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.7486366824777155e-09,
+      "loss": 1.08,
+      "step": 4821
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.525430673244403e-09,
+      "loss": 1.0537,
+      "step": 4822
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.306642972862097e-09,
+      "loss": 1.039,
+      "step": 4823
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.092273678052317e-09,
+      "loss": 1.0036,
+      "step": 4824
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.8823228835825905e-09,
+      "loss": 1.1053,
+      "step": 4825
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.676790682267563e-09,
+      "loss": 1.0815,
+      "step": 4826
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.475677164966774e-09,
+      "loss": 1.0128,
+      "step": 4827
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.278982420586886e-09,
+      "loss": 1.0347,
+      "step": 4828
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.086706536083895e-09,
+      "loss": 1.0964,
+      "step": 4829
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.898849596456477e-09,
+      "loss": 1.1696,
+      "step": 4830
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.7154116847515353e-09,
+      "loss": 1.1213,
+      "step": 4831
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.536392882064199e-09,
+      "loss": 1.0626,
+      "step": 4832
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.3617932675322763e-09,
+      "loss": 1.1226,
+      "step": 4833
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.191612918341802e-09,
+      "loss": 1.0879,
+      "step": 4834
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.02585190972704e-09,
+      "loss": 1.0957,
+      "step": 4835
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.8645103149660404e-09,
+      "loss": 1.028,
+      "step": 4836
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.7075882053828605e-09,
+      "loss": 1.0927,
+      "step": 4837
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.555085650350897e-09,
+      "loss": 1.0743,
+      "step": 4838
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.407002717285112e-09,
+      "loss": 1.0849,
+      "step": 4839
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.263339471652026e-09,
+      "loss": 1.1328,
+      "step": 4840
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.124095976959728e-09,
+      "loss": 1.0723,
+      "step": 4841
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9892722947645328e-09,
+      "loss": 1.1271,
+      "step": 4842
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8588684846698734e-09,
+      "loss": 1.0647,
+      "step": 4843
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.7328846043229707e-09,
+      "loss": 1.0619,
+      "step": 4844
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6113207094181626e-09,
+      "loss": 1.0776,
+      "step": 4845
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4941768536957945e-09,
+      "loss": 1.0811,
+      "step": 4846
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3814530889433298e-09,
+      "loss": 1.0135,
+      "step": 4847
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.2731494649909081e-09,
+      "loss": 1.0737,
+      "step": 4848
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.1692660297191183e-09,
+      "loss": 1.0991,
+      "step": 4849
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0698028290512252e-09,
+      "loss": 1.0321,
+      "step": 4850
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.74759906957612e-10,
+      "loss": 1.112,
+      "step": 4851
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.841373054546687e-10,
+      "loss": 1.0044,
+      "step": 4852
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.979350646036832e-10,
+      "loss": 1.1094,
+      "step": 4853
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.161532225130608e-10,
+      "loss": 1.0727,
+      "step": 4854
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.387918153372141e-10,
+      "loss": 1.0539,
+      "step": 4855
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.658508772743432e-10,
+      "loss": 1.113,
+      "step": 4856
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.973304405697654e-10,
+      "loss": 1.0805,
+      "step": 4857
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.332305355159161e-10,
+      "loss": 1.0595,
+      "step": 4858
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.7355119044901746e-10,
+      "loss": 1.0465,
+      "step": 4859
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.182924317512992e-10,
+      "loss": 1.0607,
+      "step": 4860
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.674542838521088e-10,
+      "loss": 1.1593,
+      "step": 4861
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.2103676922680117e-10,
+      "loss": 1.0662,
+      "step": 4862
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.7903990839229778e-10,
+      "loss": 1.1902,
+      "step": 4863
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4146371991818897e-10,
+      "loss": 1.0695,
+      "step": 4864
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0830822041230093e-10,
+      "loss": 1.0877,
+      "step": 4865
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.957342453401851e-11,
+      "loss": 1.091,
+      "step": 4866
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.525934498651353e-11,
+      "loss": 1.0617,
+      "step": 4867
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.536599251674488e-11,
+      "loss": 1.0094,
+      "step": 4868
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9893375921009595e-11,
+      "loss": 1.0495,
+      "step": 4869
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.84150203828149e-12,
+      "loss": 1.0775,
+      "step": 4870
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.2103757535418825e-12,
+      "loss": 1.0943,
+      "step": 4871
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 1.1064,
+      "step": 4872
+    },
+    {
+      "epoch": 3.0,
+      "step": 4872,
+      "total_flos": 2.4551183342748303e+18,
+      "train_loss": 1.368304116214046,
+      "train_runtime": 26096.5257,
+      "train_samples_per_second": 47.815,
+      "train_steps_per_second": 0.187
+    }
+  ],
+  "max_steps": 4872,
+  "num_train_epochs": 3,
+  "total_flos": 2.4551183342748303e+18,
+  "trial_name": null,
+  "trial_params": null
+}