diff --git "a/llava_mae/trainer_state.json" "b/llava_mae/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/llava_mae/trainer_state.json"
@@ -0,0 +1,29305 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "global_step": 4880,
+  "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": 2.9199,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.72108843537415e-07,
+      "loss": 2.6914,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0816326530612243e-07,
+      "loss": 2.6641,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.4421768707483e-07,
+      "loss": 2.7949,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.802721088435376e-07,
+      "loss": 2.7227,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.163265306122449e-07,
+      "loss": 2.5703,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.523809523809525e-07,
+      "loss": 2.498,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.08843537414966e-06,
+      "loss": 2.5508,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2244897959183673e-06,
+      "loss": 2.5889,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3605442176870751e-06,
+      "loss": 2.3682,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4965986394557825e-06,
+      "loss": 1.9717,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6326530612244897e-06,
+      "loss": 2.0117,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7687074829931975e-06,
+      "loss": 1.5811,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.904761904761905e-06,
+      "loss": 1.6689,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0408163265306125e-06,
+      "loss": 1.6504,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.17687074829932e-06,
+      "loss": 1.4814,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3129251700680273e-06,
+      "loss": 1.4688,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4489795918367347e-06,
+      "loss": 1.3223,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5850340136054425e-06,
+      "loss": 1.2852,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7210884353741503e-06,
+      "loss": 1.3232,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 1.3135,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.993197278911565e-06,
+      "loss": 1.2681,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1292517006802725e-06,
+      "loss": 1.2588,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2653061224489794e-06,
+      "loss": 1.2354,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4013605442176872e-06,
+      "loss": 1.2266,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.537414965986395e-06,
+      "loss": 1.1729,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6734693877551024e-06,
+      "loss": 1.2861,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.80952380952381e-06,
+      "loss": 1.1543,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.945578231292517e-06,
+      "loss": 1.1709,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.081632653061225e-06,
+      "loss": 1.2109,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.217687074829933e-06,
+      "loss": 1.0879,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.35374149659864e-06,
+      "loss": 1.1064,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.489795918367348e-06,
+      "loss": 1.0952,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6258503401360546e-06,
+      "loss": 1.1108,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.761904761904762e-06,
+      "loss": 1.0723,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.897959183673469e-06,
+      "loss": 1.0127,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.034013605442177e-06,
+      "loss": 1.0557,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.170068027210885e-06,
+      "loss": 1.0679,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.306122448979593e-06,
+      "loss": 1.1304,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.442176870748301e-06,
+      "loss": 1.1182,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.578231292517007e-06,
+      "loss": 1.0327,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.7142857142857145e-06,
+      "loss": 1.0303,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.850340136054422e-06,
+      "loss": 1.0713,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.98639455782313e-06,
+      "loss": 1.0176,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.122448979591837e-06,
+      "loss": 1.0376,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.258503401360545e-06,
+      "loss": 1.0464,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.394557823129253e-06,
+      "loss": 1.0518,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.530612244897959e-06,
+      "loss": 1.0938,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.0127,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.8027210884353745e-06,
+      "loss": 1.0244,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.938775510204082e-06,
+      "loss": 1.0435,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.07482993197279e-06,
+      "loss": 1.0122,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.210884353741497e-06,
+      "loss": 1.0649,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.346938775510205e-06,
+      "loss": 0.9883,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.482993197278913e-06,
+      "loss": 1.0723,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.61904761904762e-06,
+      "loss": 0.9731,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.755102040816327e-06,
+      "loss": 0.999,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.891156462585034e-06,
+      "loss": 1.0186,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.027210884353741e-06,
+      "loss": 1.0312,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.16326530612245e-06,
+      "loss": 1.0269,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.299319727891157e-06,
+      "loss": 0.9717,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.435374149659866e-06,
+      "loss": 1.0469,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.571428571428571e-06,
+      "loss": 1.0054,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.70748299319728e-06,
+      "loss": 0.9604,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.843537414965987e-06,
+      "loss": 0.9927,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.979591836734695e-06,
+      "loss": 0.9814,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.115646258503402e-06,
+      "loss": 0.9897,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.251700680272109e-06,
+      "loss": 0.9727,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.387755102040818e-06,
+      "loss": 1.0229,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.523809523809525e-06,
+      "loss": 0.9863,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.659863945578232e-06,
+      "loss": 0.9312,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.795918367346939e-06,
+      "loss": 0.959,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.931972789115647e-06,
+      "loss": 1.0122,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0068027210884354e-05,
+      "loss": 0.938,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0204081632653063e-05,
+      "loss": 1.0312,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.034013605442177e-05,
+      "loss": 0.9761,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0476190476190477e-05,
+      "loss": 1.0581,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0612244897959186e-05,
+      "loss": 0.8813,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0748299319727893e-05,
+      "loss": 0.9653,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0884353741496601e-05,
+      "loss": 0.9956,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1020408163265306e-05,
+      "loss": 0.959,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1156462585034013e-05,
+      "loss": 0.9771,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1292517006802722e-05,
+      "loss": 0.9175,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1428571428571429e-05,
+      "loss": 0.936,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1564625850340136e-05,
+      "loss": 1.0308,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1700680272108845e-05,
+      "loss": 1.0342,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1836734693877552e-05,
+      "loss": 0.9316,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.197278911564626e-05,
+      "loss": 0.9214,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2108843537414967e-05,
+      "loss": 0.9482,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2244897959183674e-05,
+      "loss": 0.9712,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2380952380952383e-05,
+      "loss": 0.9844,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.251700680272109e-05,
+      "loss": 0.9468,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2653061224489798e-05,
+      "loss": 0.9829,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2789115646258505e-05,
+      "loss": 0.9829,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2925170068027212e-05,
+      "loss": 0.9312,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3061224489795918e-05,
+      "loss": 1.001,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3197278911564626e-05,
+      "loss": 0.9868,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 1.0327,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3469387755102042e-05,
+      "loss": 1.0005,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3605442176870749e-05,
+      "loss": 1.0151,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3741496598639456e-05,
+      "loss": 0.9917,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3877551020408165e-05,
+      "loss": 0.9194,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4013605442176872e-05,
+      "loss": 0.9165,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.414965986394558e-05,
+      "loss": 0.9653,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4285714285714287e-05,
+      "loss": 0.9702,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4421768707482994e-05,
+      "loss": 0.9541,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4557823129251703e-05,
+      "loss": 0.9644,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.469387755102041e-05,
+      "loss": 0.9248,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4829931972789118e-05,
+      "loss": 0.9028,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4965986394557825e-05,
+      "loss": 0.9585,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.510204081632653e-05,
+      "loss": 0.9131,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.523809523809524e-05,
+      "loss": 0.9468,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5374149659863945e-05,
+      "loss": 0.9839,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5510204081632655e-05,
+      "loss": 0.9854,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5646258503401362e-05,
+      "loss": 0.9595,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.578231292517007e-05,
+      "loss": 0.9492,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5918367346938776e-05,
+      "loss": 0.9121,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6054421768707483e-05,
+      "loss": 0.9004,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6190476190476193e-05,
+      "loss": 0.9414,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.63265306122449e-05,
+      "loss": 0.895,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6462585034013607e-05,
+      "loss": 0.9575,
+      "step": 121
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6598639455782314e-05,
+      "loss": 0.9644,
+      "step": 122
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.673469387755102e-05,
+      "loss": 0.958,
+      "step": 123
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.687074829931973e-05,
+      "loss": 0.9663,
+      "step": 124
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7006802721088435e-05,
+      "loss": 0.9629,
+      "step": 125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7142857142857142e-05,
+      "loss": 0.9312,
+      "step": 126
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7278911564625852e-05,
+      "loss": 0.9385,
+      "step": 127
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.741496598639456e-05,
+      "loss": 0.9663,
+      "step": 128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7551020408163266e-05,
+      "loss": 0.9834,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7687074829931973e-05,
+      "loss": 0.9878,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.782312925170068e-05,
+      "loss": 0.9434,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.795918367346939e-05,
+      "loss": 0.9478,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8095238095238097e-05,
+      "loss": 0.9941,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8231292517006804e-05,
+      "loss": 0.9399,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.836734693877551e-05,
+      "loss": 0.9268,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8503401360544218e-05,
+      "loss": 0.9287,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.863945578231293e-05,
+      "loss": 0.9858,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8775510204081636e-05,
+      "loss": 0.9395,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8911564625850343e-05,
+      "loss": 0.876,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.904761904761905e-05,
+      "loss": 0.9722,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9183673469387756e-05,
+      "loss": 0.9985,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9319727891156463e-05,
+      "loss": 0.9375,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.945578231292517e-05,
+      "loss": 0.876,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9591836734693877e-05,
+      "loss": 0.9448,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9727891156462588e-05,
+      "loss": 0.9829,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9863945578231295e-05,
+      "loss": 0.9834,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.9087,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999779709015e-05,
+      "loss": 0.9136,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999118836157e-05,
+      "loss": 0.9038,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999998017381717e-05,
+      "loss": 0.8945,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999996475346181e-05,
+      "loss": 0.9614,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999944927302274e-05,
+      "loss": 0.9004,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999920695347306e-05,
+      "loss": 1.0029,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999892057607574e-05,
+      "loss": 0.9658,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999859014095705e-05,
+      "loss": 0.9707,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999821564826244e-05,
+      "loss": 0.8994,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999779709815704e-05,
+      "loss": 0.9722,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999733449082514e-05,
+      "loss": 0.9292,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999682782647064e-05,
+      "loss": 0.936,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999627710531675e-05,
+      "loss": 0.9053,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999956823276061e-05,
+      "loss": 0.9624,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999504349360072e-05,
+      "loss": 0.9326,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999436060358208e-05,
+      "loss": 0.9502,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999363365785106e-05,
+      "loss": 0.9189,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999286265672794e-05,
+      "loss": 0.9292,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999920476005524e-05,
+      "loss": 0.9561,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999118848968355e-05,
+      "loss": 0.9453,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999902853244999e-05,
+      "loss": 0.937,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999893381053993e-05,
+      "loss": 0.9414,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998834683279922e-05,
+      "loss": 1.0127,
+      "step": 170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998731150713623e-05,
+      "loss": 0.9497,
+      "step": 171
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998623212886662e-05,
+      "loss": 0.856,
+      "step": 172
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998510869846586e-05,
+      "loss": 0.9575,
+      "step": 173
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999839412164289e-05,
+      "loss": 0.9609,
+      "step": 174
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998272968327023e-05,
+      "loss": 0.9111,
+      "step": 175
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998147409952348e-05,
+      "loss": 0.9175,
+      "step": 176
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998017446574195e-05,
+      "loss": 0.9438,
+      "step": 177
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997883078249816e-05,
+      "loss": 0.9673,
+      "step": 178
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997744305038415e-05,
+      "loss": 0.9448,
+      "step": 179
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999760112700113e-05,
+      "loss": 0.9102,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997453544201048e-05,
+      "loss": 0.9517,
+      "step": 181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999730155670319e-05,
+      "loss": 0.8877,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997145164574516e-05,
+      "loss": 0.9092,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996984367883927e-05,
+      "loss": 0.8901,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996819166702274e-05,
+      "loss": 0.8818,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999664956110234e-05,
+      "loss": 0.9473,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996475551158846e-05,
+      "loss": 0.9219,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996297136948462e-05,
+      "loss": 0.9302,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996114318549787e-05,
+      "loss": 0.9487,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999592709604338e-05,
+      "loss": 0.9053,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999573546951172e-05,
+      "loss": 0.918,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995539439039232e-05,
+      "loss": 0.9595,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999533900471229e-05,
+      "loss": 0.9131,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999513416661919e-05,
+      "loss": 0.9287,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994924924850196e-05,
+      "loss": 0.9785,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994711279497488e-05,
+      "loss": 0.8652,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994493230655193e-05,
+      "loss": 0.957,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994270778419383e-05,
+      "loss": 0.9077,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999404392288806e-05,
+      "loss": 0.9077,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999381266416118e-05,
+      "loss": 0.9741,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993577002340628e-05,
+      "loss": 0.9575,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999333693753023e-05,
+      "loss": 0.939,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999309246983576e-05,
+      "loss": 0.9473,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992843599364925e-05,
+      "loss": 0.9092,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999259032622737e-05,
+      "loss": 0.9087,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999233265053468e-05,
+      "loss": 0.9502,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999207057240039e-05,
+      "loss": 0.9365,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999180409193996e-05,
+      "loss": 0.9204,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99915332092708e-05,
+      "loss": 0.9072,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991257924512252e-05,
+      "loss": 0.8965,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990978237785606e-05,
+      "loss": 0.876,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990694149214088e-05,
+      "loss": 0.9277,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999040565892286e-05,
+      "loss": 0.9429,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999011276703902e-05,
+      "loss": 0.9546,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989815473691623e-05,
+      "loss": 0.9409,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998951377901164e-05,
+      "loss": 0.9966,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989207683131998e-05,
+      "loss": 0.9697,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988897186187555e-05,
+      "loss": 0.9731,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988582288315115e-05,
+      "loss": 0.9248,
+      "step": 219
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998826298965341e-05,
+      "loss": 0.9404,
+      "step": 220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998793929034312e-05,
+      "loss": 0.9058,
+      "step": 221
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998761119052686e-05,
+      "loss": 0.9292,
+      "step": 222
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987278690349186e-05,
+      "loss": 0.9292,
+      "step": 223
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986941789956593e-05,
+      "loss": 0.8599,
+      "step": 224
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998660048949751e-05,
+      "loss": 0.916,
+      "step": 225
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998625478912231e-05,
+      "loss": 0.9004,
+      "step": 226
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985904688983303e-05,
+      "loss": 0.9062,
+      "step": 227
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985550189234733e-05,
+      "loss": 0.8306,
+      "step": 228
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985191290032794e-05,
+      "loss": 0.8999,
+      "step": 229
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984827991535602e-05,
+      "loss": 0.8799,
+      "step": 230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998446029390322e-05,
+      "loss": 0.9473,
+      "step": 231
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984088197297654e-05,
+      "loss": 0.9849,
+      "step": 232
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983711701882844e-05,
+      "loss": 0.8989,
+      "step": 233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983330807824664e-05,
+      "loss": 0.896,
+      "step": 234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982945515290926e-05,
+      "loss": 0.8979,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998255582445139e-05,
+      "loss": 0.9087,
+      "step": 236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998216173547774e-05,
+      "loss": 0.9453,
+      "step": 237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998176324854361e-05,
+      "loss": 0.8652,
+      "step": 238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998136036382456e-05,
+      "loss": 0.9111,
+      "step": 239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99809530814981e-05,
+      "loss": 0.9512,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980541401743666e-05,
+      "loss": 0.9409,
+      "step": 241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998012532474264e-05,
+      "loss": 0.9238,
+      "step": 242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979704850678342e-05,
+      "loss": 0.9458,
+      "step": 243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979279979736016e-05,
+      "loss": 0.9351,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997885071210286e-05,
+      "loss": 0.9771,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978417047967996e-05,
+      "loss": 0.9194,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997797898752249e-05,
+      "loss": 0.8794,
+      "step": 247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997753653095935e-05,
+      "loss": 0.9307,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977089678473505e-05,
+      "loss": 0.9395,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976638430261836e-05,
+      "loss": 0.9175,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976182786523154e-05,
+      "loss": 0.9326,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975722747458203e-05,
+      "loss": 0.9077,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975258313269674e-05,
+      "loss": 0.936,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974789484162187e-05,
+      "loss": 0.8916,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99743162603423e-05,
+      "loss": 0.9009,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973838642018498e-05,
+      "loss": 0.8896,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973356629401225e-05,
+      "loss": 0.8975,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997287022270284e-05,
+      "loss": 0.9893,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997237942213764e-05,
+      "loss": 0.9243,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971884227921872e-05,
+      "loss": 0.8833,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997138464027371e-05,
+      "loss": 0.897,
+      "step": 261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970880659413253e-05,
+      "loss": 0.9375,
+      "step": 262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970372285562556e-05,
+      "loss": 0.9185,
+      "step": 263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969859518945592e-05,
+      "loss": 0.9126,
+      "step": 264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969342359788283e-05,
+      "loss": 0.981,
+      "step": 265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996882080831848e-05,
+      "loss": 0.9019,
+      "step": 266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996829486476596e-05,
+      "loss": 0.9336,
+      "step": 267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967764529362456e-05,
+      "loss": 0.8745,
+      "step": 268
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967229802341616e-05,
+      "loss": 0.918,
+      "step": 269
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966690683939033e-05,
+      "loss": 0.936,
+      "step": 270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996614717439224e-05,
+      "loss": 0.8843,
+      "step": 271
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965599273940682e-05,
+      "loss": 0.9595,
+      "step": 272
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996504698282577e-05,
+      "loss": 0.9419,
+      "step": 273
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964490301290827e-05,
+      "loss": 0.8662,
+      "step": 274
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963929229581118e-05,
+      "loss": 0.8794,
+      "step": 275
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963363767943832e-05,
+      "loss": 0.8906,
+      "step": 276
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962793916628117e-05,
+      "loss": 0.9629,
+      "step": 277
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962219675885026e-05,
+      "loss": 0.9478,
+      "step": 278
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961641045967564e-05,
+      "loss": 0.874,
+      "step": 279
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996105802713067e-05,
+      "loss": 0.9502,
+      "step": 280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960470619631206e-05,
+      "loss": 0.96,
+      "step": 281
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959878823727972e-05,
+      "loss": 0.9121,
+      "step": 282
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995928263968171e-05,
+      "loss": 0.8921,
+      "step": 283
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958682067755073e-05,
+      "loss": 0.9165,
+      "step": 284
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995807710821268e-05,
+      "loss": 0.957,
+      "step": 285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957467761321053e-05,
+      "loss": 0.917,
+      "step": 286
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956854027348664e-05,
+      "loss": 0.9473,
+      "step": 287
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956235906565912e-05,
+      "loss": 0.9004,
+      "step": 288
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955613399245132e-05,
+      "loss": 0.9634,
+      "step": 289
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954986505660587e-05,
+      "loss": 0.8882,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954355226088472e-05,
+      "loss": 0.9229,
+      "step": 291
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953719560806925e-05,
+      "loss": 0.917,
+      "step": 292
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953079510096004e-05,
+      "loss": 0.8813,
+      "step": 293
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952435074237703e-05,
+      "loss": 0.9316,
+      "step": 294
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995178625351595e-05,
+      "loss": 0.8872,
+      "step": 295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951133048216608e-05,
+      "loss": 0.9102,
+      "step": 296
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950475458627463e-05,
+      "loss": 0.9004,
+      "step": 297
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949813485038237e-05,
+      "loss": 0.9263,
+      "step": 298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949147127740585e-05,
+      "loss": 0.9058,
+      "step": 299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994847638702809e-05,
+      "loss": 0.8691,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994780126319627e-05,
+      "loss": 0.9346,
+      "step": 301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947121756542573e-05,
+      "loss": 0.8638,
+      "step": 302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946437867366373e-05,
+      "loss": 0.8882,
+      "step": 303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945749595968985e-05,
+      "loss": 0.9365,
+      "step": 304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994505694265365e-05,
+      "loss": 0.915,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994435990772553e-05,
+      "loss": 0.9097,
+      "step": 306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943658491491734e-05,
+      "loss": 0.9219,
+      "step": 307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994295269426129e-05,
+      "loss": 0.9463,
+      "step": 308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994224251634516e-05,
+      "loss": 0.9287,
+      "step": 309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941527958056235e-05,
+      "loss": 0.9521,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994080901970934e-05,
+      "loss": 0.9155,
+      "step": 311
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994008570162122e-05,
+      "loss": 0.9355,
+      "step": 312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939358004110562e-05,
+      "loss": 0.9072,
+      "step": 313
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993862592749797e-05,
+      "loss": 0.8799,
+      "step": 314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937889472105988e-05,
+      "loss": 0.9688,
+      "step": 315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993714863825909e-05,
+      "loss": 0.9243,
+      "step": 316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936403426283665e-05,
+      "loss": 0.9238,
+      "step": 317
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935653836508042e-05,
+      "loss": 0.9341,
+      "step": 318
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993489986926248e-05,
+      "loss": 0.8594,
+      "step": 319
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934141524879163e-05,
+      "loss": 0.8931,
+      "step": 320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933378803692197e-05,
+      "loss": 0.9214,
+      "step": 321
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932611706037635e-05,
+      "loss": 0.9014,
+      "step": 322
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931840232253436e-05,
+      "loss": 0.8901,
+      "step": 323
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931064382679502e-05,
+      "loss": 0.9155,
+      "step": 324
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993028415765766e-05,
+      "loss": 0.9282,
+      "step": 325
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992949955753166e-05,
+      "loss": 0.9229,
+      "step": 326
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928710582647183e-05,
+      "loss": 0.9263,
+      "step": 327
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992791723335184e-05,
+      "loss": 0.9624,
+      "step": 328
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992711950999516e-05,
+      "loss": 0.8853,
+      "step": 329
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926317412928612e-05,
+      "loss": 0.9219,
+      "step": 330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925510942505583e-05,
+      "loss": 0.8965,
+      "step": 331
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992470009908139e-05,
+      "loss": 0.8857,
+      "step": 332
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992388488301327e-05,
+      "loss": 0.8843,
+      "step": 333
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923065294660407e-05,
+      "loss": 0.9277,
+      "step": 334
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922241334383884e-05,
+      "loss": 0.9141,
+      "step": 335
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921413002546724e-05,
+      "loss": 0.9429,
+      "step": 336
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920580299513878e-05,
+      "loss": 0.8979,
+      "step": 337
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919743225652223e-05,
+      "loss": 0.8408,
+      "step": 338
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918901781330556e-05,
+      "loss": 0.8848,
+      "step": 339
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918055966919597e-05,
+      "loss": 0.8691,
+      "step": 340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917205782792007e-05,
+      "loss": 0.9019,
+      "step": 341
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916351229322353e-05,
+      "loss": 0.9292,
+      "step": 342
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991549230688714e-05,
+      "loss": 0.9644,
+      "step": 343
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991462901586479e-05,
+      "loss": 0.9058,
+      "step": 344
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991376135663566e-05,
+      "loss": 0.897,
+      "step": 345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912889329582018e-05,
+      "loss": 0.8691,
+      "step": 346
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991201293508807e-05,
+      "loss": 0.8892,
+      "step": 347
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991113217353993e-05,
+      "loss": 0.9126,
+      "step": 348
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910247045325658e-05,
+      "loss": 0.8867,
+      "step": 349
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909357550835214e-05,
+      "loss": 0.9038,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908463690460506e-05,
+      "loss": 0.9399,
+      "step": 351
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990756546459534e-05,
+      "loss": 0.9116,
+      "step": 352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906662873635463e-05,
+      "loss": 0.8804,
+      "step": 353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905755917978542e-05,
+      "loss": 0.814,
+      "step": 354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904844598024166e-05,
+      "loss": 0.8618,
+      "step": 355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990392891417384e-05,
+      "loss": 0.9116,
+      "step": 356
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903008866831005e-05,
+      "loss": 0.8799,
+      "step": 357
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902084456401013e-05,
+      "loss": 0.9248,
+      "step": 358
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901155683291147e-05,
+      "loss": 0.9214,
+      "step": 359
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99002225479106e-05,
+      "loss": 0.8838,
+      "step": 360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899285050670504e-05,
+      "loss": 0.9126,
+      "step": 361
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898343191983893e-05,
+      "loss": 0.9375,
+      "step": 362
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897396972265745e-05,
+      "loss": 0.9131,
+      "step": 363
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896446391932937e-05,
+      "loss": 0.8936,
+      "step": 364
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989549145140428e-05,
+      "loss": 0.9521,
+      "step": 365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9894532151100505e-05,
+      "loss": 0.8755,
+      "step": 366
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893568491444265e-05,
+      "loss": 0.8682,
+      "step": 367
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892600472860127e-05,
+      "loss": 0.9185,
+      "step": 368
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891628095774586e-05,
+      "loss": 0.9219,
+      "step": 369
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989065136061605e-05,
+      "loss": 0.9067,
+      "step": 370
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889670267814852e-05,
+      "loss": 0.9243,
+      "step": 371
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888684817803244e-05,
+      "loss": 0.9307,
+      "step": 372
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98876950110154e-05,
+      "loss": 0.9014,
+      "step": 373
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988670084788741e-05,
+      "loss": 0.8774,
+      "step": 374
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988570232885728e-05,
+      "loss": 0.8901,
+      "step": 375
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884699454364943e-05,
+      "loss": 0.8936,
+      "step": 376
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988369222485225e-05,
+      "loss": 0.9321,
+      "step": 377
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882680640762966e-05,
+      "loss": 0.8574,
+      "step": 378
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881664702542775e-05,
+      "loss": 0.8726,
+      "step": 379
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988064441063928e-05,
+      "loss": 0.9297,
+      "step": 380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879619765502008e-05,
+      "loss": 0.896,
+      "step": 381
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878590767582398e-05,
+      "loss": 0.8838,
+      "step": 382
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98775574173338e-05,
+      "loss": 0.8828,
+      "step": 383
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876519715211504e-05,
+      "loss": 0.8652,
+      "step": 384
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875477661672692e-05,
+      "loss": 0.9092,
+      "step": 385
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874431257176474e-05,
+      "loss": 0.8765,
+      "step": 386
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987338050218388e-05,
+      "loss": 0.9722,
+      "step": 387
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872325397157856e-05,
+      "loss": 0.9355,
+      "step": 388
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987126594256326e-05,
+      "loss": 0.8911,
+      "step": 389
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987020213886687e-05,
+      "loss": 0.9072,
+      "step": 390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869133986537372e-05,
+      "loss": 0.9043,
+      "step": 391
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986806148604538e-05,
+      "loss": 0.9087,
+      "step": 392
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866984637863424e-05,
+      "loss": 0.8989,
+      "step": 393
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865903442465936e-05,
+      "loss": 0.9263,
+      "step": 394
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864817900329273e-05,
+      "loss": 0.9038,
+      "step": 395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863728011931703e-05,
+      "loss": 0.9219,
+      "step": 396
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862633777753417e-05,
+      "loss": 0.8604,
+      "step": 397
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9861535198276507e-05,
+      "loss": 0.9204,
+      "step": 398
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860432273985e-05,
+      "loss": 0.8813,
+      "step": 399
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985932500536481e-05,
+      "loss": 0.9033,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858213392903786e-05,
+      "loss": 0.9419,
+      "step": 401
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857097437091683e-05,
+      "loss": 0.9307,
+      "step": 402
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855977138420174e-05,
+      "loss": 0.8457,
+      "step": 403
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854852497382843e-05,
+      "loss": 0.9414,
+      "step": 404
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853723514475183e-05,
+      "loss": 0.856,
+      "step": 405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852590190194604e-05,
+      "loss": 0.938,
+      "step": 406
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851452525040433e-05,
+      "loss": 0.9443,
+      "step": 407
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98503105195139e-05,
+      "loss": 0.9092,
+      "step": 408
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849164174118147e-05,
+      "loss": 0.9155,
+      "step": 409
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9848013489358245e-05,
+      "loss": 0.8447,
+      "step": 410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9846858465741157e-05,
+      "loss": 0.8496,
+      "step": 411
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845699103775766e-05,
+      "loss": 0.9248,
+      "step": 412
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984453540397287e-05,
+      "loss": 0.8896,
+      "step": 413
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9843367366845172e-05,
+      "loss": 0.9468,
+      "step": 414
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984219499290729e-05,
+      "loss": 0.9019,
+      "step": 415
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841018282675743e-05,
+      "loss": 0.8628,
+      "step": 416
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839837236668975e-05,
+      "loss": 0.9092,
+      "step": 417
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838651855407335e-05,
+      "loss": 0.8877,
+      "step": 418
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983746213941308e-05,
+      "loss": 0.9399,
+      "step": 419
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983626808921037e-05,
+      "loss": 0.9395,
+      "step": 420
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835069705325292e-05,
+      "loss": 0.9258,
+      "step": 421
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833866988285825e-05,
+      "loss": 0.8784,
+      "step": 422
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9832659938621866e-05,
+      "loss": 0.8823,
+      "step": 423
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831448556865225e-05,
+      "loss": 0.9272,
+      "step": 424
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830232843549612e-05,
+      "loss": 0.8569,
+      "step": 425
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982901279921064e-05,
+      "loss": 0.8145,
+      "step": 426
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827788424385852e-05,
+      "loss": 0.9688,
+      "step": 427
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9826559719614678e-05,
+      "loss": 0.8799,
+      "step": 428
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982532668543846e-05,
+      "loss": 0.9111,
+      "step": 429
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824089322400457e-05,
+      "loss": 0.916,
+      "step": 430
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982284763104583e-05,
+      "loss": 0.9434,
+      "step": 431
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821601611921635e-05,
+      "loss": 0.8687,
+      "step": 432
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982035126557686e-05,
+      "loss": 0.8711,
+      "step": 433
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819096592562377e-05,
+      "loss": 0.8623,
+      "step": 434
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981783759343097e-05,
+      "loss": 0.8804,
+      "step": 435
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816574268737335e-05,
+      "loss": 0.8135,
+      "step": 436
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815306619038074e-05,
+      "loss": 0.9219,
+      "step": 437
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9814034644891685e-05,
+      "loss": 0.9141,
+      "step": 438
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812758346858575e-05,
+      "loss": 0.9194,
+      "step": 439
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9811477725501062e-05,
+      "loss": 0.8638,
+      "step": 440
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9810192781383368e-05,
+      "loss": 0.8208,
+      "step": 441
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980890351507161e-05,
+      "loss": 0.8584,
+      "step": 442
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9807609927133815e-05,
+      "loss": 0.8833,
+      "step": 443
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980631201813992e-05,
+      "loss": 0.8984,
+      "step": 444
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9805009788661754e-05,
+      "loss": 0.8989,
+      "step": 445
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980370323927306e-05,
+      "loss": 0.9219,
+      "step": 446
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980239237054948e-05,
+      "loss": 0.9243,
+      "step": 447
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801077183068563e-05,
+      "loss": 0.8848,
+      "step": 448
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9799757677409747e-05,
+      "loss": 0.9546,
+      "step": 449
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9798433854154387e-05,
+      "loss": 0.9292,
+      "step": 450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979710571388574e-05,
+      "loss": 0.8882,
+      "step": 451
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795773257188955e-05,
+      "loss": 0.8506,
+      "step": 452
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979443648465109e-05,
+      "loss": 0.8843,
+      "step": 453
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9793095396861104e-05,
+      "loss": 0.8862,
+      "step": 454
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9791749994409856e-05,
+      "loss": 0.8633,
+      "step": 455
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9790400277890107e-05,
+      "loss": 0.8794,
+      "step": 456
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9789046247896514e-05,
+      "loss": 0.8481,
+      "step": 457
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9787687905025642e-05,
+      "loss": 0.8359,
+      "step": 458
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9786325249875945e-05,
+      "loss": 0.9229,
+      "step": 459
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97849582830478e-05,
+      "loss": 0.8848,
+      "step": 460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978358700514345e-05,
+      "loss": 0.9419,
+      "step": 461
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782211416767066e-05,
+      "loss": 0.8438,
+      "step": 462
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780831518524704e-05,
+      "loss": 0.9185,
+      "step": 463
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9779447311024326e-05,
+      "loss": 0.9399,
+      "step": 464
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977805879487578e-05,
+      "loss": 0.8765,
+      "step": 465
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9776665970690834e-05,
+      "loss": 0.9404,
+      "step": 466
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977526883908313e-05,
+      "loss": 0.8862,
+      "step": 467
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9773867400668224e-05,
+      "loss": 0.8291,
+      "step": 468
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9772461656063563e-05,
+      "loss": 0.9897,
+      "step": 469
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97710516058885e-05,
+      "loss": 0.8848,
+      "step": 470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9769637250764262e-05,
+      "loss": 0.8794,
+      "step": 471
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768218591314003e-05,
+      "loss": 0.8857,
+      "step": 472
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9766795628162752e-05,
+      "loss": 0.9072,
+      "step": 473
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765368361937444e-05,
+      "loss": 0.9082,
+      "step": 474
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9763936793266905e-05,
+      "loss": 0.8975,
+      "step": 475
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9762500922781856e-05,
+      "loss": 0.9292,
+      "step": 476
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976106075111492e-05,
+      "loss": 0.939,
+      "step": 477
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9759616278900607e-05,
+      "loss": 0.8521,
+      "step": 478
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9758167506775326e-05,
+      "loss": 0.9116,
+      "step": 479
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9756714435377383e-05,
+      "loss": 0.9287,
+      "step": 480
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9755257065346975e-05,
+      "loss": 0.8994,
+      "step": 481
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975379539732619e-05,
+      "loss": 0.9424,
+      "step": 482
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9752329431959014e-05,
+      "loss": 0.8091,
+      "step": 483
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9750859169891325e-05,
+      "loss": 0.9136,
+      "step": 484
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974938461177089e-05,
+      "loss": 0.8833,
+      "step": 485
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747905758247375e-05,
+      "loss": 0.9424,
+      "step": 486
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9746422609972344e-05,
+      "loss": 0.8843,
+      "step": 487
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974493516759923e-05,
+      "loss": 0.8711,
+      "step": 488
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743443431783392e-05,
+      "loss": 0.8862,
+      "step": 489
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974194740318205e-05,
+      "loss": 0.8652,
+      "step": 490
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9740447082454324e-05,
+      "loss": 0.8745,
+      "step": 491
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738942470261237e-05,
+      "loss": 0.8877,
+      "step": 492
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973743356726569e-05,
+      "loss": 0.9053,
+      "step": 493
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9735920374132477e-05,
+      "loss": 0.9062,
+      "step": 494
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9734402891528287e-05,
+      "loss": 0.9019,
+      "step": 495
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9732881120121693e-05,
+      "loss": 0.877,
+      "step": 496
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9731355060583163e-05,
+      "loss": 0.9277,
+      "step": 497
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972982471358505e-05,
+      "loss": 0.897,
+      "step": 498
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9728290079801594e-05,
+      "loss": 0.8872,
+      "step": 499
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9726751159908933e-05,
+      "loss": 0.895,
+      "step": 500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972520795458508e-05,
+      "loss": 0.9072,
+      "step": 501
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9723660464509947e-05,
+      "loss": 0.8784,
+      "step": 502
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9722108690365334e-05,
+      "loss": 0.9165,
+      "step": 503
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972055263283492e-05,
+      "loss": 0.8799,
+      "step": 504
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971899229260428e-05,
+      "loss": 0.8813,
+      "step": 505
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9717427670360867e-05,
+      "loss": 0.9092,
+      "step": 506
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715858766794028e-05,
+      "loss": 0.9165,
+      "step": 507
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971428558259499e-05,
+      "loss": 0.9424,
+      "step": 508
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9712708118456877e-05,
+      "loss": 0.9233,
+      "step": 509
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9711126375074686e-05,
+      "loss": 0.9116,
+      "step": 510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9709540353145306e-05,
+      "loss": 0.8271,
+      "step": 511
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9707950053367506e-05,
+      "loss": 0.8857,
+      "step": 512
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9706355476441947e-05,
+      "loss": 0.9033,
+      "step": 513
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9704756623071172e-05,
+      "loss": 0.8359,
+      "step": 514
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9703153493959606e-05,
+      "loss": 0.8657,
+      "step": 515
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9701546089813556e-05,
+      "loss": 0.938,
+      "step": 516
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969993441134122e-05,
+      "loss": 0.8633,
+      "step": 517
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969831845925267e-05,
+      "loss": 0.8779,
+      "step": 518
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969669823425987e-05,
+      "loss": 0.9141,
+      "step": 519
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9695073737076653e-05,
+      "loss": 0.8745,
+      "step": 520
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9693444968418753e-05,
+      "loss": 0.853,
+      "step": 521
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9691811929003775e-05,
+      "loss": 0.8765,
+      "step": 522
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9690174619551203e-05,
+      "loss": 0.8516,
+      "step": 523
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9688533040782404e-05,
+      "loss": 0.9253,
+      "step": 524
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9686887193420635e-05,
+      "loss": 0.9443,
+      "step": 525
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968523707819102e-05,
+      "loss": 0.9053,
+      "step": 526
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9683582695820575e-05,
+      "loss": 0.8716,
+      "step": 527
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9681924047038187e-05,
+      "loss": 0.8906,
+      "step": 528
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968026113257463e-05,
+      "loss": 0.9082,
+      "step": 529
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9678593953162556e-05,
+      "loss": 0.8906,
+      "step": 530
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9676922509536486e-05,
+      "loss": 0.8979,
+      "step": 531
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9675246802432834e-05,
+      "loss": 0.9102,
+      "step": 532
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9673566832589888e-05,
+      "loss": 0.9004,
+      "step": 533
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9671882600747806e-05,
+      "loss": 0.939,
+      "step": 534
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967019410764864e-05,
+      "loss": 0.9409,
+      "step": 535
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9668501354036298e-05,
+      "loss": 0.8828,
+      "step": 536
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9666804340656584e-05,
+      "loss": 0.9092,
+      "step": 537
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9665103068257168e-05,
+      "loss": 0.8911,
+      "step": 538
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9663397537587604e-05,
+      "loss": 0.8599,
+      "step": 539
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9661687749399315e-05,
+      "loss": 0.9087,
+      "step": 540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9659973704445604e-05,
+      "loss": 0.9209,
+      "step": 541
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9658255403481645e-05,
+      "loss": 0.8145,
+      "step": 542
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9656532847264497e-05,
+      "loss": 0.9058,
+      "step": 543
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965480603655308e-05,
+      "loss": 0.9038,
+      "step": 544
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9653074972108203e-05,
+      "loss": 0.8721,
+      "step": 545
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9651339654692537e-05,
+      "loss": 0.8652,
+      "step": 546
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964960008507063e-05,
+      "loss": 0.8965,
+      "step": 547
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9647856264008913e-05,
+      "loss": 0.8882,
+      "step": 548
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9646108192275673e-05,
+      "loss": 0.8818,
+      "step": 549
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964435587064108e-05,
+      "loss": 0.918,
+      "step": 550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9642599299877183e-05,
+      "loss": 0.8994,
+      "step": 551
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9640838480757887e-05,
+      "loss": 0.8638,
+      "step": 552
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963907341405898e-05,
+      "loss": 0.8525,
+      "step": 553
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9637304100558122e-05,
+      "loss": 0.9302,
+      "step": 554
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9635530541034835e-05,
+      "loss": 0.9116,
+      "step": 555
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963375273627052e-05,
+      "loss": 0.8604,
+      "step": 556
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9631970687048446e-05,
+      "loss": 0.9253,
+      "step": 557
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9630184394153755e-05,
+      "loss": 0.9233,
+      "step": 558
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962839385837345e-05,
+      "loss": 0.875,
+      "step": 559
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9626599080496407e-05,
+      "loss": 0.8984,
+      "step": 560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962480006131338e-05,
+      "loss": 0.9033,
+      "step": 561
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962299680161698e-05,
+      "loss": 0.8706,
+      "step": 562
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9621189302201692e-05,
+      "loss": 0.8433,
+      "step": 563
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9619377563863867e-05,
+      "loss": 0.938,
+      "step": 564
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9617561587401726e-05,
+      "loss": 0.8828,
+      "step": 565
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9615741373615352e-05,
+      "loss": 0.8872,
+      "step": 566
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.96139169233067e-05,
+      "loss": 0.8794,
+      "step": 567
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9612088237279596e-05,
+      "loss": 0.8706,
+      "step": 568
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9610255316339712e-05,
+      "loss": 0.9253,
+      "step": 569
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9608418161294612e-05,
+      "loss": 0.9189,
+      "step": 570
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960657677295371e-05,
+      "loss": 0.8916,
+      "step": 571
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9604731152128287e-05,
+      "loss": 0.8447,
+      "step": 572
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960288129963149e-05,
+      "loss": 0.9248,
+      "step": 573
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9601027216278338e-05,
+      "loss": 0.8291,
+      "step": 574
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9599168902885696e-05,
+      "loss": 0.9307,
+      "step": 575
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959730636027231e-05,
+      "loss": 0.8726,
+      "step": 576
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959543958925878e-05,
+      "loss": 0.8501,
+      "step": 577
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959356859066757e-05,
+      "loss": 0.8335,
+      "step": 578
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9591693365323015e-05,
+      "loss": 0.9268,
+      "step": 579
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9589813914051297e-05,
+      "loss": 0.8838,
+      "step": 580
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9587930237680475e-05,
+      "loss": 0.8774,
+      "step": 581
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9586042337040464e-05,
+      "loss": 0.8828,
+      "step": 582
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958415021296303e-05,
+      "loss": 0.9072,
+      "step": 583
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9582253866281817e-05,
+      "loss": 0.8701,
+      "step": 584
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9580353297832317e-05,
+      "loss": 0.9253,
+      "step": 585
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957844850845189e-05,
+      "loss": 0.9121,
+      "step": 586
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9576539498979745e-05,
+      "loss": 0.896,
+      "step": 587
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9574626270256963e-05,
+      "loss": 0.8833,
+      "step": 588
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9572708823126477e-05,
+      "loss": 0.9131,
+      "step": 589
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957078715843308e-05,
+      "loss": 0.9219,
+      "step": 590
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956886127702342e-05,
+      "loss": 0.8892,
+      "step": 591
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956693117974601e-05,
+      "loss": 0.877,
+      "step": 592
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9564996867451214e-05,
+      "loss": 0.8374,
+      "step": 593
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9563058340991254e-05,
+      "loss": 0.9155,
+      "step": 594
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956111560122021e-05,
+      "loss": 0.873,
+      "step": 595
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955916864899402e-05,
+      "loss": 0.8867,
+      "step": 596
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9557217485170473e-05,
+      "loss": 0.9487,
+      "step": 597
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955526211060922e-05,
+      "loss": 0.8667,
+      "step": 598
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955330252617176e-05,
+      "loss": 0.877,
+      "step": 599
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9551338732721458e-05,
+      "loss": 0.8486,
+      "step": 600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9549370731123516e-05,
+      "loss": 0.9072,
+      "step": 601
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9547398522245007e-05,
+      "loss": 0.8848,
+      "step": 602
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9545422106954848e-05,
+      "loss": 0.9155,
+      "step": 603
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954344148612381e-05,
+      "loss": 0.9141,
+      "step": 604
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9541456660624525e-05,
+      "loss": 0.9248,
+      "step": 605
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9539467631331465e-05,
+      "loss": 0.9399,
+      "step": 606
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9537474399120965e-05,
+      "loss": 0.8872,
+      "step": 607
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95354769648712e-05,
+      "loss": 0.8809,
+      "step": 608
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9533475329462213e-05,
+      "loss": 0.8955,
+      "step": 609
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9531469493775884e-05,
+      "loss": 0.8608,
+      "step": 610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952945945869595e-05,
+      "loss": 0.9053,
+      "step": 611
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9527445225107993e-05,
+      "loss": 0.8262,
+      "step": 612
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9525426793899447e-05,
+      "loss": 0.8784,
+      "step": 613
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95234041659596e-05,
+      "loss": 0.8838,
+      "step": 614
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9521377342179586e-05,
+      "loss": 0.8975,
+      "step": 615
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9519346323452383e-05,
+      "loss": 0.9077,
+      "step": 616
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9517311110672824e-05,
+      "loss": 0.8838,
+      "step": 617
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951527170473759e-05,
+      "loss": 0.8955,
+      "step": 618
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95132281065452e-05,
+      "loss": 0.9155,
+      "step": 619
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951118031699603e-05,
+      "loss": 0.8584,
+      "step": 620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9509128336992294e-05,
+      "loss": 0.9087,
+      "step": 621
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950707216743807e-05,
+      "loss": 0.9009,
+      "step": 622
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9505011809239252e-05,
+      "loss": 0.8975,
+      "step": 623
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9502947263303615e-05,
+      "loss": 0.834,
+      "step": 624
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9500878530540748e-05,
+      "loss": 0.877,
+      "step": 625
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94988056118621e-05,
+      "loss": 0.9043,
+      "step": 626
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9496728508180963e-05,
+      "loss": 0.9062,
+      "step": 627
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9494647220412474e-05,
+      "loss": 0.9185,
+      "step": 628
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9492561749473604e-05,
+      "loss": 0.9204,
+      "step": 629
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9490472096283183e-05,
+      "loss": 0.8862,
+      "step": 630
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9488378261761866e-05,
+      "loss": 0.8984,
+      "step": 631
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948628024683216e-05,
+      "loss": 0.9346,
+      "step": 632
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948417805241842e-05,
+      "loss": 0.8682,
+      "step": 633
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9482071679446823e-05,
+      "loss": 0.9297,
+      "step": 634
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9479961128845407e-05,
+      "loss": 0.8872,
+      "step": 635
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9477846401544044e-05,
+      "loss": 0.8389,
+      "step": 636
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9475727498474442e-05,
+      "loss": 0.9551,
+      "step": 637
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9473604420570148e-05,
+      "loss": 0.8765,
+      "step": 638
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947147716876656e-05,
+      "loss": 0.877,
+      "step": 639
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94693457440009e-05,
+      "loss": 0.8755,
+      "step": 640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9467210147212235e-05,
+      "loss": 0.8813,
+      "step": 641
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9465070379341475e-05,
+      "loss": 0.8608,
+      "step": 642
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946292644133136e-05,
+      "loss": 0.8779,
+      "step": 643
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946077833412647e-05,
+      "loss": 0.834,
+      "step": 644
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945862605867323e-05,
+      "loss": 0.8438,
+      "step": 645
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9456469615919876e-05,
+      "loss": 0.8887,
+      "step": 646
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9454309006816517e-05,
+      "loss": 0.9106,
+      "step": 647
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945214423231507e-05,
+      "loss": 0.8926,
+      "step": 648
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9449975293369296e-05,
+      "loss": 0.9438,
+      "step": 649
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9447802190934788e-05,
+      "loss": 0.8604,
+      "step": 650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944562492596898e-05,
+      "loss": 0.8481,
+      "step": 651
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9443443499431133e-05,
+      "loss": 0.8726,
+      "step": 652
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9441257912282347e-05,
+      "loss": 0.8457,
+      "step": 653
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943906816548555e-05,
+      "loss": 0.9058,
+      "step": 654
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9436874260005504e-05,
+      "loss": 0.8828,
+      "step": 655
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9434676196808806e-05,
+      "loss": 0.9092,
+      "step": 656
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9432473976863884e-05,
+      "loss": 0.8887,
+      "step": 657
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9430267601140996e-05,
+      "loss": 0.854,
+      "step": 658
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9428057070612224e-05,
+      "loss": 0.8447,
+      "step": 659
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.94258423862515e-05,
+      "loss": 0.8818,
+      "step": 660
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942362354903457e-05,
+      "loss": 0.9131,
+      "step": 661
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942140055993901e-05,
+      "loss": 0.8857,
+      "step": 662
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941917341994423e-05,
+      "loss": 0.8823,
+      "step": 663
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941694213003147e-05,
+      "loss": 0.8462,
+      "step": 664
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941470669118379e-05,
+      "loss": 0.9043,
+      "step": 665
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9412467104386095e-05,
+      "loss": 0.877,
+      "step": 666
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9410223370625096e-05,
+      "loss": 0.9243,
+      "step": 667
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940797549088935e-05,
+      "loss": 0.8931,
+      "step": 668
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9405723466169223e-05,
+      "loss": 0.8618,
+      "step": 669
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9403467297456925e-05,
+      "loss": 0.8081,
+      "step": 670
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940120698574648e-05,
+      "loss": 0.8613,
+      "step": 671
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9398942532033744e-05,
+      "loss": 0.8774,
+      "step": 672
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9396673937316386e-05,
+      "loss": 0.8511,
+      "step": 673
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9394401202593914e-05,
+      "loss": 0.9136,
+      "step": 674
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9392124328867653e-05,
+      "loss": 0.8921,
+      "step": 675
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9389843317140756e-05,
+      "loss": 0.9097,
+      "step": 676
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938755816841819e-05,
+      "loss": 0.9375,
+      "step": 677
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9385268883706748e-05,
+      "loss": 0.9082,
+      "step": 678
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9382975464015056e-05,
+      "loss": 0.8955,
+      "step": 679
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938067791035355e-05,
+      "loss": 0.8096,
+      "step": 680
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9378376223734488e-05,
+      "loss": 0.7925,
+      "step": 681
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9376070405171953e-05,
+      "loss": 0.875,
+      "step": 682
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937376045568185e-05,
+      "loss": 0.8247,
+      "step": 683
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93714463762819e-05,
+      "loss": 0.8965,
+      "step": 684
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936912816799164e-05,
+      "loss": 0.894,
+      "step": 685
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936680583183244e-05,
+      "loss": 0.8818,
+      "step": 686
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936447936882747e-05,
+      "loss": 0.8521,
+      "step": 687
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9362148780001728e-05,
+      "loss": 0.8579,
+      "step": 688
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9359814066382035e-05,
+      "loss": 0.9038,
+      "step": 689
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9357475228997025e-05,
+      "loss": 0.8628,
+      "step": 690
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9355132268877143e-05,
+      "loss": 0.9023,
+      "step": 691
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9352785187054654e-05,
+      "loss": 0.7905,
+      "step": 692
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9350433984563644e-05,
+      "loss": 0.9575,
+      "step": 693
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9348078662440008e-05,
+      "loss": 0.8398,
+      "step": 694
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9345719221721457e-05,
+      "loss": 0.8936,
+      "step": 695
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9343355663447523e-05,
+      "loss": 0.9287,
+      "step": 696
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934098798865954e-05,
+      "loss": 0.9102,
+      "step": 697
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933861619840067e-05,
+      "loss": 0.8643,
+      "step": 698
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933624029371588e-05,
+      "loss": 0.8667,
+      "step": 699
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9333860275651947e-05,
+      "loss": 0.8916,
+      "step": 700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9331476145257464e-05,
+      "loss": 0.9409,
+      "step": 701
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932908790358284e-05,
+      "loss": 0.9409,
+      "step": 702
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932669555168029e-05,
+      "loss": 0.9023,
+      "step": 703
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9324299090603836e-05,
+      "loss": 0.8135,
+      "step": 704
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932189852140932e-05,
+      "loss": 0.8477,
+      "step": 705
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9319493845154392e-05,
+      "loss": 0.8389,
+      "step": 706
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9317085062898502e-05,
+      "loss": 0.9326,
+      "step": 707
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9314672175702924e-05,
+      "loss": 0.9048,
+      "step": 708
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9312255184630728e-05,
+      "loss": 0.8994,
+      "step": 709
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9309834090746796e-05,
+      "loss": 0.9238,
+      "step": 710
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9307408895117825e-05,
+      "loss": 0.9146,
+      "step": 711
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.93049795988123e-05,
+      "loss": 0.8994,
+      "step": 712
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9302546202900535e-05,
+      "loss": 0.8511,
+      "step": 713
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9300108708454637e-05,
+      "loss": 0.8979,
+      "step": 714
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9297667116548525e-05,
+      "loss": 0.875,
+      "step": 715
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9295221428257912e-05,
+      "loss": 0.876,
+      "step": 716
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9292771644660337e-05,
+      "loss": 0.8335,
+      "step": 717
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929031776683512e-05,
+      "loss": 0.9121,
+      "step": 718
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9287859795863402e-05,
+      "loss": 0.8735,
+      "step": 719
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9285397732828117e-05,
+      "loss": 0.8921,
+      "step": 720
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9282931578814004e-05,
+      "loss": 0.8472,
+      "step": 721
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9280461334907614e-05,
+      "loss": 0.874,
+      "step": 722
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9277987002197283e-05,
+      "loss": 0.8428,
+      "step": 723
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927550858177316e-05,
+      "loss": 0.8828,
+      "step": 724
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9273026074727192e-05,
+      "loss": 0.8579,
+      "step": 725
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927053948215313e-05,
+      "loss": 0.8833,
+      "step": 726
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9268048805146518e-05,
+      "loss": 0.8916,
+      "step": 727
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9265554044804707e-05,
+      "loss": 0.8711,
+      "step": 728
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926305520222684e-05,
+      "loss": 0.8159,
+      "step": 729
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9260552278513868e-05,
+      "loss": 0.8726,
+      "step": 730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9258045274768524e-05,
+      "loss": 0.9092,
+      "step": 731
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9255534192095355e-05,
+      "loss": 0.9282,
+      "step": 732
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92530190316007e-05,
+      "loss": 0.9097,
+      "step": 733
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925049979439269e-05,
+      "loss": 0.8594,
+      "step": 734
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924797648158126e-05,
+      "loss": 0.873,
+      "step": 735
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924544909427813e-05,
+      "loss": 0.8979,
+      "step": 736
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9242917633596824e-05,
+      "loss": 0.8867,
+      "step": 737
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9240382100652658e-05,
+      "loss": 0.873,
+      "step": 738
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9237842496562744e-05,
+      "loss": 0.8013,
+      "step": 739
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9235298822445983e-05,
+      "loss": 0.9131,
+      "step": 740
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9232751079423073e-05,
+      "loss": 0.876,
+      "step": 741
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9230199268616504e-05,
+      "loss": 0.8442,
+      "step": 742
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9227643391150558e-05,
+      "loss": 0.8936,
+      "step": 743
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922508344815131e-05,
+      "loss": 0.8545,
+      "step": 744
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9222519440746618e-05,
+      "loss": 0.8521,
+      "step": 745
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9219951370066145e-05,
+      "loss": 0.8594,
+      "step": 746
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9217379237241334e-05,
+      "loss": 0.9688,
+      "step": 747
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921480304340542e-05,
+      "loss": 0.9048,
+      "step": 748
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9212222789693427e-05,
+      "loss": 0.856,
+      "step": 749
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9209638477242168e-05,
+      "loss": 0.8994,
+      "step": 750
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920705010719025e-05,
+      "loss": 0.8911,
+      "step": 751
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9204457680678053e-05,
+      "loss": 0.8555,
+      "step": 752
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9201861198847757e-05,
+      "loss": 0.8652,
+      "step": 753
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919926066284333e-05,
+      "loss": 0.8433,
+      "step": 754
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9196656073810517e-05,
+      "loss": 0.8242,
+      "step": 755
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9194047432896853e-05,
+      "loss": 0.8433,
+      "step": 756
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9191434741251655e-05,
+      "loss": 0.8589,
+      "step": 757
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918881800002604e-05,
+      "loss": 0.9072,
+      "step": 758
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918619721037288e-05,
+      "loss": 0.8696,
+      "step": 759
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9183572373446856e-05,
+      "loss": 0.9302,
+      "step": 760
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9180943490404428e-05,
+      "loss": 0.8901,
+      "step": 761
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917831056240383e-05,
+      "loss": 0.8467,
+      "step": 762
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917567359060508e-05,
+      "loss": 0.8872,
+      "step": 763
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9173032576169986e-05,
+      "loss": 0.8584,
+      "step": 764
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9170387520262127e-05,
+      "loss": 0.8418,
+      "step": 765
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9167738424046865e-05,
+      "loss": 0.8413,
+      "step": 766
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9165085288691356e-05,
+      "loss": 0.8516,
+      "step": 767
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916242811536451e-05,
+      "loss": 0.8682,
+      "step": 768
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9159766905237032e-05,
+      "loss": 0.8994,
+      "step": 769
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915710165948141e-05,
+      "loss": 0.8906,
+      "step": 770
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91544323792719e-05,
+      "loss": 0.9971,
+      "step": 771
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9151759065784536e-05,
+      "loss": 0.9136,
+      "step": 772
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914908172019713e-05,
+      "loss": 0.894,
+      "step": 773
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914640034368928e-05,
+      "loss": 0.8828,
+      "step": 774
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9143714937442347e-05,
+      "loss": 0.8608,
+      "step": 775
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9141025502639475e-05,
+      "loss": 0.8667,
+      "step": 776
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9138332040465576e-05,
+      "loss": 0.9004,
+      "step": 777
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9135634552107347e-05,
+      "loss": 0.8242,
+      "step": 778
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9132933038753246e-05,
+      "loss": 0.897,
+      "step": 779
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913022750159352e-05,
+      "loss": 0.9194,
+      "step": 780
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9127517941820172e-05,
+      "loss": 0.8467,
+      "step": 781
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9124804360626986e-05,
+      "loss": 0.8462,
+      "step": 782
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9122086759209518e-05,
+      "loss": 0.875,
+      "step": 783
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9119365138765098e-05,
+      "loss": 0.8779,
+      "step": 784
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9116639500492818e-05,
+      "loss": 0.8564,
+      "step": 785
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9113909845593548e-05,
+      "loss": 0.9126,
+      "step": 786
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9111176175269923e-05,
+      "loss": 0.8257,
+      "step": 787
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9108438490726348e-05,
+      "loss": 0.8164,
+      "step": 788
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9105696793169002e-05,
+      "loss": 0.8584,
+      "step": 789
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9102951083805822e-05,
+      "loss": 0.8267,
+      "step": 790
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9100201363846523e-05,
+      "loss": 0.8633,
+      "step": 791
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909744763450258e-05,
+      "loss": 0.8359,
+      "step": 792
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9094689896987232e-05,
+      "loss": 0.917,
+      "step": 793
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909192815251549e-05,
+      "loss": 0.9019,
+      "step": 794
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908916240230414e-05,
+      "loss": 0.8242,
+      "step": 795
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9086392647571707e-05,
+      "loss": 0.8423,
+      "step": 796
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.90836188895385e-05,
+      "loss": 0.8887,
+      "step": 797
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908084112942659e-05,
+      "loss": 0.9185,
+      "step": 798
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9078059368459807e-05,
+      "loss": 0.8931,
+      "step": 799
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907527360786374e-05,
+      "loss": 0.8306,
+      "step": 800
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907248384886575e-05,
+      "loss": 0.8301,
+      "step": 801
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9069690092694956e-05,
+      "loss": 0.9155,
+      "step": 802
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906689234058223e-05,
+      "loss": 0.8579,
+      "step": 803
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9064090593760215e-05,
+      "loss": 0.9014,
+      "step": 804
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9061284853463313e-05,
+      "loss": 0.8267,
+      "step": 805
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9058475120927674e-05,
+      "loss": 0.8838,
+      "step": 806
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9055661397391224e-05,
+      "loss": 0.8574,
+      "step": 807
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9052843684093633e-05,
+      "loss": 0.8882,
+      "step": 808
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905002198227634e-05,
+      "loss": 0.8711,
+      "step": 809
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9047196293182533e-05,
+      "loss": 0.8774,
+      "step": 810
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9044366618057157e-05,
+      "loss": 0.856,
+      "step": 811
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9041532958146924e-05,
+      "loss": 0.8149,
+      "step": 812
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9038695314700283e-05,
+      "loss": 0.8848,
+      "step": 813
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9035853688967455e-05,
+      "loss": 0.8389,
+      "step": 814
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903300808220041e-05,
+      "loss": 0.8799,
+      "step": 815
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9030158495652867e-05,
+      "loss": 0.9004,
+      "step": 816
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9027304930580307e-05,
+      "loss": 0.8618,
+      "step": 817
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9024447388239956e-05,
+      "loss": 0.9346,
+      "step": 818
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9021585869890792e-05,
+      "loss": 0.8467,
+      "step": 819
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9018720376793556e-05,
+      "loss": 0.854,
+      "step": 820
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9015850910210728e-05,
+      "loss": 0.8511,
+      "step": 821
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9012977471406547e-05,
+      "loss": 0.8862,
+      "step": 822
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9010100061646993e-05,
+      "loss": 0.8901,
+      "step": 823
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900721868219981e-05,
+      "loss": 0.8359,
+      "step": 824
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900433333433447e-05,
+      "loss": 0.7871,
+      "step": 825
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9001444019322212e-05,
+      "loss": 0.8628,
+      "step": 826
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8998550738436013e-05,
+      "loss": 0.9155,
+      "step": 827
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8995653492950603e-05,
+      "loss": 0.8672,
+      "step": 828
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8992752284142458e-05,
+      "loss": 0.8896,
+      "step": 829
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8989847113289794e-05,
+      "loss": 0.8496,
+      "step": 830
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8986937981672575e-05,
+      "loss": 0.9189,
+      "step": 831
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898402489057252e-05,
+      "loss": 0.874,
+      "step": 832
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8981107841273082e-05,
+      "loss": 0.8921,
+      "step": 833
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8978186835059455e-05,
+      "loss": 0.8643,
+      "step": 834
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8975261873218582e-05,
+      "loss": 0.8647,
+      "step": 835
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8972332957039158e-05,
+      "loss": 0.9019,
+      "step": 836
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89694000878116e-05,
+      "loss": 0.9399,
+      "step": 837
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896646326682808e-05,
+      "loss": 0.9111,
+      "step": 838
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8963522495382512e-05,
+      "loss": 0.9263,
+      "step": 839
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8960577774770544e-05,
+      "loss": 0.8809,
+      "step": 840
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895762910628957e-05,
+      "loss": 0.897,
+      "step": 841
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8954676491238714e-05,
+      "loss": 0.8657,
+      "step": 842
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8951719930918847e-05,
+      "loss": 0.8545,
+      "step": 843
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894875942663258e-05,
+      "loss": 0.8779,
+      "step": 844
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894579497968426e-05,
+      "loss": 0.8984,
+      "step": 845
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8942826591379958e-05,
+      "loss": 0.8843,
+      "step": 846
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8939854263027504e-05,
+      "loss": 0.8247,
+      "step": 847
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8936877995936442e-05,
+      "loss": 0.9268,
+      "step": 848
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8933897791418072e-05,
+      "loss": 0.8467,
+      "step": 849
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893091365078541e-05,
+      "loss": 0.8892,
+      "step": 850
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8927925575353215e-05,
+      "loss": 0.8369,
+      "step": 851
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8924933566437985e-05,
+      "loss": 0.8745,
+      "step": 852
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8921937625357944e-05,
+      "loss": 0.8276,
+      "step": 853
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8918937753433046e-05,
+      "loss": 0.8486,
+      "step": 854
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8915933951984978e-05,
+      "loss": 0.9443,
+      "step": 855
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8912926222337168e-05,
+      "loss": 0.8667,
+      "step": 856
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8909914565814765e-05,
+      "loss": 0.8516,
+      "step": 857
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890689898374465e-05,
+      "loss": 0.8877,
+      "step": 858
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890387947745543e-05,
+      "loss": 0.8701,
+      "step": 859
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8900856048277453e-05,
+      "loss": 0.8584,
+      "step": 860
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8897828697542783e-05,
+      "loss": 0.8423,
+      "step": 861
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8894797426585216e-05,
+      "loss": 0.854,
+      "step": 862
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8891762236740278e-05,
+      "loss": 0.8867,
+      "step": 863
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888872312934521e-05,
+      "loss": 0.9023,
+      "step": 864
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8885680105739002e-05,
+      "loss": 0.8628,
+      "step": 865
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8882633167262344e-05,
+      "loss": 0.8794,
+      "step": 866
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887958231525767e-05,
+      "loss": 0.8506,
+      "step": 867
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887652755106912e-05,
+      "loss": 0.8477,
+      "step": 868
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8873468876042576e-05,
+      "loss": 0.9307,
+      "step": 869
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8870406291525638e-05,
+      "loss": 0.853,
+      "step": 870
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886733979886762e-05,
+      "loss": 0.8394,
+      "step": 871
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8864269399419562e-05,
+      "loss": 0.8848,
+      "step": 872
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886119509453423e-05,
+      "loss": 0.8462,
+      "step": 873
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8858116885566108e-05,
+      "loss": 0.8638,
+      "step": 874
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88550347738714e-05,
+      "loss": 0.8359,
+      "step": 875
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8851948760808022e-05,
+      "loss": 0.9106,
+      "step": 876
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8848858847735622e-05,
+      "loss": 0.8447,
+      "step": 877
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884576503601556e-05,
+      "loss": 0.8477,
+      "step": 878
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8842667327010908e-05,
+      "loss": 0.9409,
+      "step": 879
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883956572208647e-05,
+      "loss": 0.8916,
+      "step": 880
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8836460222608748e-05,
+      "loss": 0.8823,
+      "step": 881
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8833350829945977e-05,
+      "loss": 0.8359,
+      "step": 882
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8830237545468092e-05,
+      "loss": 0.9429,
+      "step": 883
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8827120370546753e-05,
+      "loss": 0.8696,
+      "step": 884
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8823999306555334e-05,
+      "loss": 0.9326,
+      "step": 885
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8820874354868916e-05,
+      "loss": 0.894,
+      "step": 886
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8817745516864296e-05,
+      "loss": 0.8745,
+      "step": 887
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8814612793919987e-05,
+      "loss": 0.856,
+      "step": 888
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8811476187416206e-05,
+      "loss": 0.8506,
+      "step": 889
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880833569873489e-05,
+      "loss": 0.8145,
+      "step": 890
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8805191329259677e-05,
+      "loss": 0.8237,
+      "step": 891
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8802043080375924e-05,
+      "loss": 0.8696,
+      "step": 892
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879889095347069e-05,
+      "loss": 0.8555,
+      "step": 893
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8795734949932746e-05,
+      "loss": 0.8657,
+      "step": 894
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8792575071152565e-05,
+      "loss": 0.9009,
+      "step": 895
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8789411318522342e-05,
+      "loss": 0.9111,
+      "step": 896
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8786243693435968e-05,
+      "loss": 0.8882,
+      "step": 897
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878307219728903e-05,
+      "loss": 0.8677,
+      "step": 898
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877989683147885e-05,
+      "loss": 0.8813,
+      "step": 899
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8776717597404422e-05,
+      "loss": 0.8638,
+      "step": 900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8773534496466463e-05,
+      "loss": 0.8633,
+      "step": 901
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8770347530067395e-05,
+      "loss": 0.8438,
+      "step": 902
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8767156699611334e-05,
+      "loss": 0.9229,
+      "step": 903
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87639620065041e-05,
+      "loss": 0.8911,
+      "step": 904
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876076345215322e-05,
+      "loss": 0.8516,
+      "step": 905
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875756103796792e-05,
+      "loss": 0.8564,
+      "step": 906
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8754354765359132e-05,
+      "loss": 0.9229,
+      "step": 907
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875114463573947e-05,
+      "loss": 0.8481,
+      "step": 908
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8747930650523264e-05,
+      "loss": 0.9126,
+      "step": 909
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8744712811126536e-05,
+      "loss": 0.8442,
+      "step": 910
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8741491118967013e-05,
+      "loss": 0.8906,
+      "step": 911
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8738265575464112e-05,
+      "loss": 0.8384,
+      "step": 912
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873503618203895e-05,
+      "loss": 0.8887,
+      "step": 913
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8731802940114334e-05,
+      "loss": 0.8389,
+      "step": 914
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8728565851114782e-05,
+      "loss": 0.8184,
+      "step": 915
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872532491646649e-05,
+      "loss": 0.8359,
+      "step": 916
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8722080137597355e-05,
+      "loss": 0.939,
+      "step": 917
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8718831515936973e-05,
+      "loss": 0.8594,
+      "step": 918
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8715579052916625e-05,
+      "loss": 0.9189,
+      "step": 919
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8712322749969288e-05,
+      "loss": 0.8555,
+      "step": 920
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8709062608529628e-05,
+      "loss": 0.8394,
+      "step": 921
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870579863003401e-05,
+      "loss": 0.8843,
+      "step": 922
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870253081592048e-05,
+      "loss": 0.9106,
+      "step": 923
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869925916762878e-05,
+      "loss": 0.896,
+      "step": 924
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8695983686600335e-05,
+      "loss": 0.9106,
+      "step": 925
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8692704374278268e-05,
+      "loss": 0.8296,
+      "step": 926
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8689421232107382e-05,
+      "loss": 0.7861,
+      "step": 927
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8686134261534176e-05,
+      "loss": 0.8965,
+      "step": 928
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868284346400682e-05,
+      "loss": 0.8101,
+      "step": 929
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8679548840975185e-05,
+      "loss": 0.9243,
+      "step": 930
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8676250393890828e-05,
+      "loss": 0.8384,
+      "step": 931
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8672948124206973e-05,
+      "loss": 0.9312,
+      "step": 932
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866964203337855e-05,
+      "loss": 0.8574,
+      "step": 933
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866633212286216e-05,
+      "loss": 0.8374,
+      "step": 934
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8663018394116092e-05,
+      "loss": 0.8657,
+      "step": 935
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8659700848600314e-05,
+      "loss": 0.9233,
+      "step": 936
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8656379487776478e-05,
+      "loss": 0.8506,
+      "step": 937
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865305431310791e-05,
+      "loss": 0.8828,
+      "step": 938
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8649725326059628e-05,
+      "loss": 0.8135,
+      "step": 939
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864639252809832e-05,
+      "loss": 0.8105,
+      "step": 940
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864305592069236e-05,
+      "loss": 0.8472,
+      "step": 941
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8639715505311794e-05,
+      "loss": 0.9175,
+      "step": 942
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8636371283428355e-05,
+      "loss": 0.8779,
+      "step": 943
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8633023256515436e-05,
+      "loss": 0.8491,
+      "step": 944
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8629671426048126e-05,
+      "loss": 0.7832,
+      "step": 945
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8626315793503175e-05,
+      "loss": 0.8521,
+      "step": 946
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8622956360359025e-05,
+      "loss": 0.8442,
+      "step": 947
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8619593128095767e-05,
+      "loss": 0.8535,
+      "step": 948
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8616226098195192e-05,
+      "loss": 0.7466,
+      "step": 949
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861285527214075e-05,
+      "loss": 0.8325,
+      "step": 950
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.860948065141756e-05,
+      "loss": 0.8735,
+      "step": 951
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8606102237512426e-05,
+      "loss": 0.9126,
+      "step": 952
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860272003191381e-05,
+      "loss": 0.8872,
+      "step": 953
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859933403611186e-05,
+      "loss": 0.8857,
+      "step": 954
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8595944251598377e-05,
+      "loss": 0.8618,
+      "step": 955
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8592550679866843e-05,
+      "loss": 0.8325,
+      "step": 956
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8589153322412405e-05,
+      "loss": 0.832,
+      "step": 957
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858575218073187e-05,
+      "loss": 0.8955,
+      "step": 958
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858234725632373e-05,
+      "loss": 0.8569,
+      "step": 959
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8578938550688123e-05,
+      "loss": 0.8462,
+      "step": 960
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8575526065326874e-05,
+      "loss": 0.8438,
+      "step": 961
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8572109801743453e-05,
+      "loss": 0.8774,
+      "step": 962
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856868976144301e-05,
+      "loss": 0.7886,
+      "step": 963
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8565265945932352e-05,
+      "loss": 0.8325,
+      "step": 964
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856183835671995e-05,
+      "loss": 0.8193,
+      "step": 965
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8558406995315932e-05,
+      "loss": 0.8794,
+      "step": 966
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855497186323211e-05,
+      "loss": 0.8408,
+      "step": 967
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8551532961981923e-05,
+      "loss": 0.8442,
+      "step": 968
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85480902930805e-05,
+      "loss": 0.8931,
+      "step": 969
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8544643858044616e-05,
+      "loss": 0.9141,
+      "step": 970
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8541193658392705e-05,
+      "loss": 0.8491,
+      "step": 971
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853773969564487e-05,
+      "loss": 0.8384,
+      "step": 972
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8534281971322856e-05,
+      "loss": 0.853,
+      "step": 973
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853082048695008e-05,
+      "loss": 0.8711,
+      "step": 974
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8527355244051607e-05,
+      "loss": 0.8242,
+      "step": 975
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8523886244154163e-05,
+      "loss": 0.8384,
+      "step": 976
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8520413488786127e-05,
+      "loss": 0.8877,
+      "step": 977
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8516936979477527e-05,
+      "loss": 0.876,
+      "step": 978
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8513456717760057e-05,
+      "loss": 0.8438,
+      "step": 979
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8509972705167052e-05,
+      "loss": 0.8535,
+      "step": 980
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850648494323351e-05,
+      "loss": 0.8267,
+      "step": 981
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8502993433496073e-05,
+      "loss": 0.895,
+      "step": 982
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8499498177493036e-05,
+      "loss": 0.8774,
+      "step": 983
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849599917676435e-05,
+      "loss": 0.8545,
+      "step": 984
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849249643285161e-05,
+      "loss": 0.8062,
+      "step": 985
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848898994729806e-05,
+      "loss": 0.8599,
+      "step": 986
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8485479721648595e-05,
+      "loss": 0.9399,
+      "step": 987
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8481965757449755e-05,
+      "loss": 0.8486,
+      "step": 988
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8478448056249737e-05,
+      "loss": 0.8081,
+      "step": 989
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847492661959837e-05,
+      "loss": 0.8501,
+      "step": 990
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8471401449047138e-05,
+      "loss": 0.8716,
+      "step": 991
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8467872546149165e-05,
+      "loss": 0.8633,
+      "step": 992
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846433991245922e-05,
+      "loss": 0.8643,
+      "step": 993
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846080354953373e-05,
+      "loss": 0.8628,
+      "step": 994
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8457263458930737e-05,
+      "loss": 0.8779,
+      "step": 995
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845371964220995e-05,
+      "loss": 0.8555,
+      "step": 996
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8450172100932712e-05,
+      "loss": 0.8608,
+      "step": 997
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8446620836662e-05,
+      "loss": 0.8911,
+      "step": 998
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8443065850962442e-05,
+      "loss": 0.8896,
+      "step": 999
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8439507145400294e-05,
+      "loss": 0.8364,
+      "step": 1000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8435944721543466e-05,
+      "loss": 0.8853,
+      "step": 1001
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.843237858096149e-05,
+      "loss": 0.8467,
+      "step": 1002
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8428808725225548e-05,
+      "loss": 0.8525,
+      "step": 1003
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8425235155908455e-05,
+      "loss": 0.9131,
+      "step": 1004
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8421657874584658e-05,
+      "loss": 0.9146,
+      "step": 1005
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8418076882830244e-05,
+      "loss": 0.8726,
+      "step": 1006
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8414492182222933e-05,
+      "loss": 0.8574,
+      "step": 1007
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841090377434208e-05,
+      "loss": 0.8564,
+      "step": 1008
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840731166076867e-05,
+      "loss": 0.8154,
+      "step": 1009
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8403715843085325e-05,
+      "loss": 0.9321,
+      "step": 1010
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8400116322876302e-05,
+      "loss": 0.8911,
+      "step": 1011
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839651310172748e-05,
+      "loss": 0.8999,
+      "step": 1012
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8392906181226376e-05,
+      "loss": 0.8364,
+      "step": 1013
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8389295562962132e-05,
+      "loss": 0.8838,
+      "step": 1014
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838568124852552e-05,
+      "loss": 0.8193,
+      "step": 1015
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8382063239508944e-05,
+      "loss": 0.8354,
+      "step": 1016
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837844153750643e-05,
+      "loss": 0.9014,
+      "step": 1017
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837481614411364e-05,
+      "loss": 0.8906,
+      "step": 1018
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8371187060927855e-05,
+      "loss": 0.8232,
+      "step": 1019
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8367554289547985e-05,
+      "loss": 0.8003,
+      "step": 1020
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836391783157456e-05,
+      "loss": 0.8828,
+      "step": 1021
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836027768860974e-05,
+      "loss": 0.8018,
+      "step": 1022
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8356633862257304e-05,
+      "loss": 0.8789,
+      "step": 1023
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8352986354122663e-05,
+      "loss": 0.9019,
+      "step": 1024
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8349335165812835e-05,
+      "loss": 0.8501,
+      "step": 1025
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8345680298936472e-05,
+      "loss": 0.8638,
+      "step": 1026
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834202175510384e-05,
+      "loss": 0.8999,
+      "step": 1027
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8338359535926834e-05,
+      "loss": 0.8369,
+      "step": 1028
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833469364301895e-05,
+      "loss": 0.8818,
+      "step": 1029
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8331024077995326e-05,
+      "loss": 0.8701,
+      "step": 1030
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83273508424727e-05,
+      "loss": 0.8877,
+      "step": 1031
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8323673938069434e-05,
+      "loss": 0.8159,
+      "step": 1032
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831999336640551e-05,
+      "loss": 0.873,
+      "step": 1033
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8316309129102513e-05,
+      "loss": 0.8979,
+      "step": 1034
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831262122778366e-05,
+      "loss": 0.8657,
+      "step": 1035
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830892966407377e-05,
+      "loss": 0.8887,
+      "step": 1036
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8305234439599276e-05,
+      "loss": 0.8418,
+      "step": 1037
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8301535555988235e-05,
+      "loss": 0.8838,
+      "step": 1038
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8297833014870307e-05,
+      "loss": 0.8193,
+      "step": 1039
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8294126817876758e-05,
+      "loss": 0.832,
+      "step": 1040
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8290416966640475e-05,
+      "loss": 0.9058,
+      "step": 1041
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8286703462795957e-05,
+      "loss": 0.8369,
+      "step": 1042
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.82829863079793e-05,
+      "loss": 0.8604,
+      "step": 1043
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8279265503828215e-05,
+      "loss": 0.8628,
+      "step": 1044
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827554105198203e-05,
+      "loss": 0.8828,
+      "step": 1045
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8271812954081662e-05,
+      "loss": 0.8403,
+      "step": 1046
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8268081211769645e-05,
+      "loss": 0.8525,
+      "step": 1047
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8264345826690123e-05,
+      "loss": 0.8047,
+      "step": 1048
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826060680048883e-05,
+      "loss": 0.9097,
+      "step": 1049
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8256864134813126e-05,
+      "loss": 0.9126,
+      "step": 1050
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8253117831311954e-05,
+      "loss": 0.8066,
+      "step": 1051
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8249367891635864e-05,
+      "loss": 0.9146,
+      "step": 1052
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824561431743702e-05,
+      "loss": 0.8843,
+      "step": 1053
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824185711036918e-05,
+      "loss": 0.957,
+      "step": 1054
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8238096272087694e-05,
+      "loss": 0.8701,
+      "step": 1055
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823433180424952e-05,
+      "loss": 0.894,
+      "step": 1056
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823056370851322e-05,
+      "loss": 0.876,
+      "step": 1057
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822679198653895e-05,
+      "loss": 0.8828,
+      "step": 1058
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8223016639988458e-05,
+      "loss": 0.8857,
+      "step": 1059
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8219237670525092e-05,
+      "loss": 0.8594,
+      "step": 1060
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8215455079813806e-05,
+      "loss": 0.9087,
+      "step": 1061
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8211668869521136e-05,
+      "loss": 0.7915,
+      "step": 1062
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8207879041315216e-05,
+      "loss": 0.8379,
+      "step": 1063
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8204085596865778e-05,
+      "loss": 0.8633,
+      "step": 1064
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8200288537844146e-05,
+      "loss": 0.8281,
+      "step": 1065
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8196487865923234e-05,
+      "loss": 0.8359,
+      "step": 1066
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8192683582777553e-05,
+      "loss": 0.8242,
+      "step": 1067
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8188875690083197e-05,
+      "loss": 0.8662,
+      "step": 1068
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8185064189517857e-05,
+      "loss": 0.8267,
+      "step": 1069
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8181249082760813e-05,
+      "loss": 0.8252,
+      "step": 1070
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817743037149293e-05,
+      "loss": 0.7998,
+      "step": 1071
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8173608057396667e-05,
+      "loss": 0.8389,
+      "step": 1072
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816978214215606e-05,
+      "loss": 0.895,
+      "step": 1073
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8165952627456746e-05,
+      "loss": 0.9121,
+      "step": 1074
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8162119514985934e-05,
+      "loss": 0.8525,
+      "step": 1075
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8158282806432428e-05,
+      "loss": 0.8931,
+      "step": 1076
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8154442503486608e-05,
+      "loss": 0.9355,
+      "step": 1077
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8150598607840448e-05,
+      "loss": 0.8618,
+      "step": 1078
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8146751121187497e-05,
+      "loss": 0.895,
+      "step": 1079
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8142900045222888e-05,
+      "loss": 0.8691,
+      "step": 1080
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8139045381643332e-05,
+      "loss": 0.8608,
+      "step": 1081
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813518713214713e-05,
+      "loss": 0.9019,
+      "step": 1082
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813132529843415e-05,
+      "loss": 0.8447,
+      "step": 1083
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812745988220586e-05,
+      "loss": 0.9038,
+      "step": 1084
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812359088516528e-05,
+      "loss": 0.8315,
+      "step": 1085
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811971830901702e-05,
+      "loss": 0.9263,
+      "step": 1086
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8115842155467275e-05,
+      "loss": 0.8574,
+      "step": 1087
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8111962426223802e-05,
+      "loss": 0.9253,
+      "step": 1088
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8108079122995946e-05,
+      "loss": 0.8193,
+      "step": 1089
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8104192247494618e-05,
+      "loss": 0.9551,
+      "step": 1090
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81003018014323e-05,
+      "loss": 0.8799,
+      "step": 1091
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809640778652306e-05,
+      "loss": 0.8291,
+      "step": 1092
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8092510204482526e-05,
+      "loss": 0.8545,
+      "step": 1093
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8088609057027902e-05,
+      "loss": 0.8413,
+      "step": 1094
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8084704345877966e-05,
+      "loss": 0.8921,
+      "step": 1095
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8080796072753063e-05,
+      "loss": 0.8369,
+      "step": 1096
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807688423937511e-05,
+      "loss": 0.8086,
+      "step": 1097
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.807296884746758e-05,
+      "loss": 0.8472,
+      "step": 1098
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8069049898755534e-05,
+      "loss": 0.8535,
+      "step": 1099
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806512739496559e-05,
+      "loss": 0.8823,
+      "step": 1100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8061201337825927e-05,
+      "loss": 0.8447,
+      "step": 1101
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8057271729066297e-05,
+      "loss": 0.8516,
+      "step": 1102
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8053338570418017e-05,
+      "loss": 0.8394,
+      "step": 1103
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8049401863613966e-05,
+      "loss": 0.8364,
+      "step": 1104
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8045461610388583e-05,
+      "loss": 0.8569,
+      "step": 1105
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8041517812477875e-05,
+      "loss": 0.8574,
+      "step": 1106
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8037570471619406e-05,
+      "loss": 0.8921,
+      "step": 1107
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8033619589552304e-05,
+      "loss": 0.8594,
+      "step": 1108
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8029665168017258e-05,
+      "loss": 0.8633,
+      "step": 1109
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802570720875651e-05,
+      "loss": 0.9019,
+      "step": 1110
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8021745713513877e-05,
+      "loss": 0.8018,
+      "step": 1111
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801778068403471e-05,
+      "loss": 0.8086,
+      "step": 1112
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801381212206593e-05,
+      "loss": 0.8418,
+      "step": 1113
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800984002935602e-05,
+      "loss": 0.8193,
+      "step": 1114
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8005864407655014e-05,
+      "loss": 0.8672,
+      "step": 1115
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8001885258714494e-05,
+      "loss": 0.8506,
+      "step": 1116
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79979025842876e-05,
+      "loss": 0.8857,
+      "step": 1117
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7993916386129033e-05,
+      "loss": 0.8403,
+      "step": 1118
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7989926665995033e-05,
+      "loss": 0.8477,
+      "step": 1119
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7985933425643404e-05,
+      "loss": 0.8145,
+      "step": 1120
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7981936666833494e-05,
+      "loss": 0.8584,
+      "step": 1121
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7977936391326204e-05,
+      "loss": 0.8228,
+      "step": 1122
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7973932600883982e-05,
+      "loss": 0.8213,
+      "step": 1123
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796992529727082e-05,
+      "loss": 0.8394,
+      "step": 1124
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7965914482252274e-05,
+      "loss": 0.8496,
+      "step": 1125
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7961900157595435e-05,
+      "loss": 0.9253,
+      "step": 1126
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7957882325068938e-05,
+      "loss": 0.854,
+      "step": 1127
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7953860986442964e-05,
+      "loss": 0.834,
+      "step": 1128
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7949836143489256e-05,
+      "loss": 0.8594,
+      "step": 1129
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794580779798107e-05,
+      "loss": 0.7791,
+      "step": 1130
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7941775951693236e-05,
+      "loss": 0.8735,
+      "step": 1131
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7937740606402104e-05,
+      "loss": 0.8789,
+      "step": 1132
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793370176388558e-05,
+      "loss": 0.9082,
+      "step": 1133
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7929659425923104e-05,
+      "loss": 0.856,
+      "step": 1134
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7925613594295655e-05,
+      "loss": 0.8232,
+      "step": 1135
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7921564270785757e-05,
+      "loss": 0.8481,
+      "step": 1136
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7917511457177464e-05,
+      "loss": 0.8291,
+      "step": 1137
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791345515525638e-05,
+      "loss": 0.8042,
+      "step": 1138
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7909395366809632e-05,
+      "loss": 0.8662,
+      "step": 1139
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7905332093625893e-05,
+      "loss": 0.7881,
+      "step": 1140
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790126533749536e-05,
+      "loss": 0.8867,
+      "step": 1141
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7897195100209783e-05,
+      "loss": 0.8911,
+      "step": 1142
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789312138356243e-05,
+      "loss": 0.8242,
+      "step": 1143
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788904418934811e-05,
+      "loss": 0.8657,
+      "step": 1144
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7884963519363162e-05,
+      "loss": 0.8789,
+      "step": 1145
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788087937540545e-05,
+      "loss": 0.8511,
+      "step": 1146
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7876791759274378e-05,
+      "loss": 0.8921,
+      "step": 1147
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7872700672770876e-05,
+      "loss": 0.9116,
+      "step": 1148
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7868606117697396e-05,
+      "loss": 0.877,
+      "step": 1149
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786450809585794e-05,
+      "loss": 0.8633,
+      "step": 1150
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7860406609058006e-05,
+      "loss": 0.8555,
+      "step": 1151
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7856301659104647e-05,
+      "loss": 0.8999,
+      "step": 1152
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7852193247806427e-05,
+      "loss": 0.8159,
+      "step": 1153
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7848081376973437e-05,
+      "loss": 0.9053,
+      "step": 1154
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7843966048417292e-05,
+      "loss": 0.936,
+      "step": 1155
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783984726395113e-05,
+      "loss": 0.8428,
+      "step": 1156
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7835725025389617e-05,
+      "loss": 0.9351,
+      "step": 1157
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783159933454894e-05,
+      "loss": 0.9087,
+      "step": 1158
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7827470193246796e-05,
+      "loss": 0.8438,
+      "step": 1159
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7823337603302412e-05,
+      "loss": 0.8359,
+      "step": 1160
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781920156653654e-05,
+      "loss": 0.8008,
+      "step": 1161
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7815062084771434e-05,
+      "loss": 0.8711,
+      "step": 1162
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781091915983088e-05,
+      "loss": 0.835,
+      "step": 1163
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780677279354017e-05,
+      "loss": 0.8755,
+      "step": 1164
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780262298772613e-05,
+      "loss": 0.8389,
+      "step": 1165
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779846974421708e-05,
+      "loss": 0.8804,
+      "step": 1166
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7794313064842868e-05,
+      "loss": 0.8403,
+      "step": 1167
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779015295143485e-05,
+      "loss": 0.8394,
+      "step": 1168
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7785989405825898e-05,
+      "loss": 0.8867,
+      "step": 1169
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7781822429850396e-05,
+      "loss": 0.873,
+      "step": 1170
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7777652025344235e-05,
+      "loss": 0.8535,
+      "step": 1171
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7773478194144825e-05,
+      "loss": 0.8726,
+      "step": 1172
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7769300938091076e-05,
+      "loss": 0.8794,
+      "step": 1173
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7765120259023416e-05,
+      "loss": 0.8589,
+      "step": 1174
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7760936158783773e-05,
+      "loss": 0.8423,
+      "step": 1175
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775674863921559e-05,
+      "loss": 0.8579,
+      "step": 1176
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7752557702163807e-05,
+      "loss": 0.8477,
+      "step": 1177
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774836334947488e-05,
+      "loss": 0.814,
+      "step": 1178
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7744165582996768e-05,
+      "loss": 0.8906,
+      "step": 1179
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7739964404578925e-05,
+      "loss": 0.8169,
+      "step": 1180
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7735759816072312e-05,
+      "loss": 0.7998,
+      "step": 1181
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7731551819329407e-05,
+      "loss": 0.8799,
+      "step": 1182
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7727340416204165e-05,
+      "loss": 0.8794,
+      "step": 1183
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7723125608552063e-05,
+      "loss": 0.874,
+      "step": 1184
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7718907398230067e-05,
+      "loss": 0.853,
+      "step": 1185
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7714685787096636e-05,
+      "loss": 0.9165,
+      "step": 1186
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7710460777011745e-05,
+      "loss": 0.832,
+      "step": 1187
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770623236983686e-05,
+      "loss": 0.8638,
+      "step": 1188
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770200056743493e-05,
+      "loss": 0.8735,
+      "step": 1189
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769776537167042e-05,
+      "loss": 0.8638,
+      "step": 1190
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7693526784409278e-05,
+      "loss": 0.9038,
+      "step": 1191
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7689284807518948e-05,
+      "loss": 0.8755,
+      "step": 1192
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768503944286837e-05,
+      "loss": 0.8662,
+      "step": 1193
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7680790692327974e-05,
+      "loss": 0.895,
+      "step": 1194
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7676538557769685e-05,
+      "loss": 0.8247,
+      "step": 1195
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7672283041066912e-05,
+      "loss": 0.8423,
+      "step": 1196
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7668024144094565e-05,
+      "loss": 0.7959,
+      "step": 1197
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7663761868729034e-05,
+      "loss": 0.8477,
+      "step": 1198
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7659496216848202e-05,
+      "loss": 0.8281,
+      "step": 1199
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7655227190331434e-05,
+      "loss": 0.8965,
+      "step": 1200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7650954791059592e-05,
+      "loss": 0.8218,
+      "step": 1201
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7646679020915013e-05,
+      "loss": 0.8921,
+      "step": 1202
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764239988178153e-05,
+      "loss": 0.9448,
+      "step": 1203
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7638117375544454e-05,
+      "loss": 0.8911,
+      "step": 1204
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763383150409057e-05,
+      "loss": 0.8589,
+      "step": 1205
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7629542269308165e-05,
+      "loss": 0.8599,
+      "step": 1206
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7625249673087e-05,
+      "loss": 0.8311,
+      "step": 1207
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762095371731831e-05,
+      "loss": 0.8623,
+      "step": 1208
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7616654403894818e-05,
+      "loss": 0.8525,
+      "step": 1209
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7612351734710724e-05,
+      "loss": 0.8931,
+      "step": 1210
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7608045711661708e-05,
+      "loss": 0.8604,
+      "step": 1211
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760373633664492e-05,
+      "loss": 0.8965,
+      "step": 1212
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7599423611559e-05,
+      "loss": 0.8779,
+      "step": 1213
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7595107538304054e-05,
+      "loss": 0.8745,
+      "step": 1214
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7590788118781668e-05,
+      "loss": 0.8447,
+      "step": 1215
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7586465354894895e-05,
+      "loss": 0.9414,
+      "step": 1216
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.758213924854827e-05,
+      "loss": 0.8384,
+      "step": 1217
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75778098016478e-05,
+      "loss": 0.8374,
+      "step": 1218
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7573477016100962e-05,
+      "loss": 0.854,
+      "step": 1219
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7569140893816695e-05,
+      "loss": 0.8481,
+      "step": 1220
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7564801436705425e-05,
+      "loss": 0.8433,
+      "step": 1221
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7560458646679034e-05,
+      "loss": 0.7866,
+      "step": 1222
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7556112525650875e-05,
+      "loss": 0.8193,
+      "step": 1223
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7551763075535778e-05,
+      "loss": 0.8779,
+      "step": 1224
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7547410298250026e-05,
+      "loss": 0.8853,
+      "step": 1225
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754305419571138e-05,
+      "loss": 0.8306,
+      "step": 1226
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7538694769839052e-05,
+      "loss": 0.9165,
+      "step": 1227
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7534332022553733e-05,
+      "loss": 0.834,
+      "step": 1228
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752996595577757e-05,
+      "loss": 0.9302,
+      "step": 1229
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752559657143417e-05,
+      "loss": 0.8687,
+      "step": 1230
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7521223871448608e-05,
+      "loss": 0.8716,
+      "step": 1231
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7516847857747413e-05,
+      "loss": 0.8208,
+      "step": 1232
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751246853225858e-05,
+      "loss": 0.8564,
+      "step": 1233
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7508085896911565e-05,
+      "loss": 0.8882,
+      "step": 1234
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7503699953637272e-05,
+      "loss": 0.8354,
+      "step": 1235
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7499310704368073e-05,
+      "loss": 0.8921,
+      "step": 1236
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749491815103779e-05,
+      "loss": 0.7949,
+      "step": 1237
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7490522295581703e-05,
+      "loss": 0.8311,
+      "step": 1238
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7486123139936543e-05,
+      "loss": 0.8462,
+      "step": 1239
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7481720686040506e-05,
+      "loss": 0.8418,
+      "step": 1240
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7477314935833226e-05,
+      "loss": 0.8345,
+      "step": 1241
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7472905891255806e-05,
+      "loss": 0.8584,
+      "step": 1242
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746849355425078e-05,
+      "loss": 0.8467,
+      "step": 1243
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7464077926762154e-05,
+      "loss": 0.854,
+      "step": 1244
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7459659010735376e-05,
+      "loss": 0.8599,
+      "step": 1245
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745523680811733e-05,
+      "loss": 0.8569,
+      "step": 1246
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745081132085636e-05,
+      "loss": 0.8662,
+      "step": 1247
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744638255090227e-05,
+      "loss": 0.8647,
+      "step": 1248
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744195050020628e-05,
+      "loss": 0.8711,
+      "step": 1249
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743751517072108e-05,
+      "loss": 0.8579,
+      "step": 1250
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743307656440079e-05,
+      "loss": 0.8433,
+      "step": 1251
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7428634683200987e-05,
+      "loss": 0.8359,
+      "step": 1252
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742418952907868e-05,
+      "loss": 0.8955,
+      "step": 1253
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7419741103992326e-05,
+      "loss": 0.8281,
+      "step": 1254
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7415289409901817e-05,
+      "loss": 0.8877,
+      "step": 1255
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7410834448768493e-05,
+      "loss": 0.8242,
+      "step": 1256
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740637622255513e-05,
+      "loss": 0.8193,
+      "step": 1257
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7401914733225934e-05,
+      "loss": 0.8535,
+      "step": 1258
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739744998274657e-05,
+      "loss": 0.8442,
+      "step": 1259
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7392981973084114e-05,
+      "loss": 0.8809,
+      "step": 1260
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7388510706207098e-05,
+      "loss": 0.8086,
+      "step": 1261
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738403618408548e-05,
+      "loss": 0.8354,
+      "step": 1262
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737955840869065e-05,
+      "loss": 0.8398,
+      "step": 1263
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7375077381995443e-05,
+      "loss": 0.8643,
+      "step": 1264
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7370593105974115e-05,
+      "loss": 0.8511,
+      "step": 1265
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7366105582602353e-05,
+      "loss": 0.8594,
+      "step": 1266
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736161481385728e-05,
+      "loss": 0.8789,
+      "step": 1267
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7357120801717452e-05,
+      "loss": 0.8618,
+      "step": 1268
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7352623548162847e-05,
+      "loss": 0.854,
+      "step": 1269
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7348123055174876e-05,
+      "loss": 0.915,
+      "step": 1270
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7343619324736372e-05,
+      "loss": 0.8467,
+      "step": 1271
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7339112358831597e-05,
+      "loss": 0.8589,
+      "step": 1272
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7334602159446245e-05,
+      "loss": 0.8389,
+      "step": 1273
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733008872856742e-05,
+      "loss": 0.9106,
+      "step": 1274
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7325572068183664e-05,
+      "loss": 0.8774,
+      "step": 1275
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7321052180284937e-05,
+      "loss": 0.8442,
+      "step": 1276
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7316529066862614e-05,
+      "loss": 0.8394,
+      "step": 1277
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7312002729909502e-05,
+      "loss": 0.7969,
+      "step": 1278
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7307473171419823e-05,
+      "loss": 0.7764,
+      "step": 1279
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730294039338922e-05,
+      "loss": 0.8667,
+      "step": 1280
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729840439781475e-05,
+      "loss": 0.8896,
+      "step": 1281
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7293865186694893e-05,
+      "loss": 0.8413,
+      "step": 1282
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728932276202954e-05,
+      "loss": 0.7661,
+      "step": 1283
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7284777125820007e-05,
+      "loss": 0.8394,
+      "step": 1284
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7280228280069016e-05,
+      "loss": 0.834,
+      "step": 1285
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7275676226780707e-05,
+      "loss": 0.8496,
+      "step": 1286
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7271120967960632e-05,
+      "loss": 0.8433,
+      "step": 1287
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726656250561576e-05,
+      "loss": 0.8892,
+      "step": 1288
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7262000841754463e-05,
+      "loss": 0.8149,
+      "step": 1289
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7257435978386525e-05,
+      "loss": 0.8618,
+      "step": 1290
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725286791752315e-05,
+      "loss": 0.8647,
+      "step": 1291
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.724829666117694e-05,
+      "loss": 0.8364,
+      "step": 1292
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7243722211361902e-05,
+      "loss": 0.8706,
+      "step": 1293
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7239144570093468e-05,
+      "loss": 0.8833,
+      "step": 1294
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7234563739388454e-05,
+      "loss": 0.9248,
+      "step": 1295
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.72299797212651e-05,
+      "loss": 0.8247,
+      "step": 1296
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7225392517743035e-05,
+      "loss": 0.8467,
+      "step": 1297
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.72208021308433e-05,
+      "loss": 0.8521,
+      "step": 1298
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7216208562588335e-05,
+      "loss": 0.8257,
+      "step": 1299
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7211611815001986e-05,
+      "loss": 0.8271,
+      "step": 1300
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.72070118901095e-05,
+      "loss": 0.8047,
+      "step": 1301
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7202408789937514e-05,
+      "loss": 0.8398,
+      "step": 1302
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7197802516514076e-05,
+      "loss": 0.8716,
+      "step": 1303
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7193193071868622e-05,
+      "loss": 0.9448,
+      "step": 1304
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7188580458032e-05,
+      "loss": 0.853,
+      "step": 1305
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7183964677036435e-05,
+      "loss": 0.8315,
+      "step": 1306
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717934573091556e-05,
+      "loss": 0.8428,
+      "step": 1307
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71747236217044e-05,
+      "loss": 0.8345,
+      "step": 1308
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717009835143937e-05,
+      "loss": 0.8784,
+      "step": 1309
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7165469922158284e-05,
+      "loss": 0.9043,
+      "step": 1310
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7160838335900342e-05,
+      "loss": 0.8486,
+      "step": 1311
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7156203594706144e-05,
+      "loss": 0.8428,
+      "step": 1312
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7151565700617662e-05,
+      "loss": 0.8628,
+      "step": 1313
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714692465567828e-05,
+      "loss": 0.7949,
+      "step": 1314
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714228046193275e-05,
+      "loss": 0.7798,
+      "step": 1315
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7137633121427225e-05,
+      "loss": 0.8682,
+      "step": 1316
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.713298263620924e-05,
+      "loss": 0.8506,
+      "step": 1317
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7128329008327712e-05,
+      "loss": 0.8721,
+      "step": 1318
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7123672239832947e-05,
+      "loss": 0.855,
+      "step": 1319
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7119012332776632e-05,
+      "loss": 0.8706,
+      "step": 1320
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7114349289211838e-05,
+      "loss": 0.8574,
+      "step": 1321
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7109683111193017e-05,
+      "loss": 0.8232,
+      "step": 1322
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710501380077601e-05,
+      "loss": 0.7822,
+      "step": 1323
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7100341360018024e-05,
+      "loss": 0.8472,
+      "step": 1324
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709566579097765e-05,
+      "loss": 0.8135,
+      "step": 1325
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7090987095714863e-05,
+      "loss": 0.8804,
+      "step": 1326
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7086305276291017e-05,
+      "loss": 0.8721,
+      "step": 1327
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7081620334768828e-05,
+      "loss": 0.8525,
+      "step": 1328
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70769322732124e-05,
+      "loss": 0.8389,
+      "step": 1329
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7072241093687215e-05,
+      "loss": 0.8618,
+      "step": 1330
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7067546798260114e-05,
+      "loss": 0.8496,
+      "step": 1331
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7062849388999318e-05,
+      "loss": 0.8506,
+      "step": 1332
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7058148867974427e-05,
+      "loss": 0.9023,
+      "step": 1333
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7053445237256403e-05,
+      "loss": 0.8423,
+      "step": 1334
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704873849891758e-05,
+      "loss": 0.8901,
+      "step": 1335
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7044028655031667e-05,
+      "loss": 0.8076,
+      "step": 1336
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7039315707673726e-05,
+      "loss": 0.8364,
+      "step": 1337
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7034599658920206e-05,
+      "loss": 0.9009,
+      "step": 1338
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702988051084891e-05,
+      "loss": 0.7891,
+      "step": 1339
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702515826553901e-05,
+      "loss": 0.8721,
+      "step": 1340
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702043292507104e-05,
+      "loss": 0.8359,
+      "step": 1341
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7015704491526902e-05,
+      "loss": 0.7842,
+      "step": 1342
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.701097296698986e-05,
+      "loss": 0.8862,
+      "step": 1343
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7006238353544536e-05,
+      "loss": 0.8882,
+      "step": 1344
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700150065327691e-05,
+      "loss": 0.8022,
+      "step": 1345
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699675986827434e-05,
+      "loss": 0.832,
+      "step": 1346
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6992016000625523e-05,
+      "loss": 0.8457,
+      "step": 1347
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698726905242052e-05,
+      "loss": 0.8242,
+      "step": 1348
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698251902575075e-05,
+      "loss": 0.8555,
+      "step": 1349
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6977765922708995e-05,
+      "loss": 0.7959,
+      "step": 1350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6973009745389383e-05,
+      "loss": 0.8682,
+      "step": 1351
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69682504958874e-05,
+      "loss": 0.8379,
+      "step": 1352
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6963488176299888e-05,
+      "loss": 0.8159,
+      "step": 1353
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6958722788725037e-05,
+      "loss": 0.8052,
+      "step": 1354
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6953954335262384e-05,
+      "loss": 0.894,
+      "step": 1355
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6949182818012837e-05,
+      "loss": 0.8091,
+      "step": 1356
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6944408239078632e-05,
+      "loss": 0.8638,
+      "step": 1357
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6939630600563366e-05,
+      "loss": 0.8521,
+      "step": 1358
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6934849904571977e-05,
+      "loss": 0.7515,
+      "step": 1359
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6930066153210756e-05,
+      "loss": 0.8467,
+      "step": 1360
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6925279348587337e-05,
+      "loss": 0.8726,
+      "step": 1361
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6920489492810702e-05,
+      "loss": 0.832,
+      "step": 1362
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6915696587991167e-05,
+      "loss": 0.7842,
+      "step": 1363
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6910900636240407e-05,
+      "loss": 0.8501,
+      "step": 1364
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690610163967143e-05,
+      "loss": 0.8813,
+      "step": 1365
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690129960039859e-05,
+      "loss": 0.8447,
+      "step": 1366
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6896494520537576e-05,
+      "loss": 0.8501,
+      "step": 1367
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689168640220542e-05,
+      "loss": 0.8179,
+      "step": 1368
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688687524752049e-05,
+      "loss": 0.8599,
+      "step": 1369
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.68820610586025e-05,
+      "loss": 0.7998,
+      "step": 1370
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6877243837572486e-05,
+      "loss": 0.8364,
+      "step": 1371
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6872423586552837e-05,
+      "loss": 0.8164,
+      "step": 1372
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6867600307667263e-05,
+      "loss": 0.8379,
+      "step": 1373
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6862774003040817e-05,
+      "loss": 0.8311,
+      "step": 1374
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685794467479988e-05,
+      "loss": 0.8984,
+      "step": 1375
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6853112325072172e-05,
+      "loss": 0.8872,
+      "step": 1376
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684827695598673e-05,
+      "loss": 0.8438,
+      "step": 1377
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684343856967394e-05,
+      "loss": 0.8501,
+      "step": 1378
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6838597168265497e-05,
+      "loss": 0.8276,
+      "step": 1379
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6833752753894444e-05,
+      "loss": 0.8677,
+      "step": 1380
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6828905328695144e-05,
+      "loss": 0.8491,
+      "step": 1381
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682405489480328e-05,
+      "loss": 0.8481,
+      "step": 1382
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6819201454355865e-05,
+      "loss": 0.9009,
+      "step": 1383
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681434500949124e-05,
+      "loss": 0.8462,
+      "step": 1384
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6809485562349064e-05,
+      "loss": 0.874,
+      "step": 1385
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6804623115070325e-05,
+      "loss": 0.7759,
+      "step": 1386
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.679975766979733e-05,
+      "loss": 0.8418,
+      "step": 1387
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6794889228673706e-05,
+      "loss": 0.8008,
+      "step": 1388
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6790017793844398e-05,
+      "loss": 0.812,
+      "step": 1389
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6785143367455673e-05,
+      "loss": 0.874,
+      "step": 1390
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6780265951655116e-05,
+      "loss": 0.7886,
+      "step": 1391
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6775385548591633e-05,
+      "loss": 0.8853,
+      "step": 1392
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6770502160415433e-05,
+      "loss": 0.8486,
+      "step": 1393
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6765615789278055e-05,
+      "loss": 0.8013,
+      "step": 1394
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6760726437332345e-05,
+      "loss": 0.8398,
+      "step": 1395
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675583410673246e-05,
+      "loss": 0.8921,
+      "step": 1396
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6750938799633874e-05,
+      "loss": 0.8989,
+      "step": 1397
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6746040518193376e-05,
+      "loss": 0.8794,
+      "step": 1398
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6741139264569055e-05,
+      "loss": 0.8213,
+      "step": 1399
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6736235040920317e-05,
+      "loss": 0.8623,
+      "step": 1400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6731327849407873e-05,
+      "loss": 0.8179,
+      "step": 1401
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6726417692193745e-05,
+      "loss": 0.876,
+      "step": 1402
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672150457144126e-05,
+      "loss": 0.855,
+      "step": 1403
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6716588489315046e-05,
+      "loss": 0.8555,
+      "step": 1404
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6711669447981048e-05,
+      "loss": 0.8643,
+      "step": 1405
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6706747449606497e-05,
+      "loss": 0.854,
+      "step": 1406
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.670182249635995e-05,
+      "loss": 0.8501,
+      "step": 1407
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6696894590411237e-05,
+      "loss": 0.853,
+      "step": 1408
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6691963733931515e-05,
+      "loss": 0.8706,
+      "step": 1409
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668702992909323e-05,
+      "loss": 0.8262,
+      "step": 1410
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6682093178070122e-05,
+      "loss": 0.875,
+      "step": 1411
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667715348303724e-05,
+      "loss": 0.8682,
+      "step": 1412
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6672210846170922e-05,
+      "loss": 0.8486,
+      "step": 1413
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6667265269648805e-05,
+      "loss": 0.8608,
+      "step": 1414
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666231675564982e-05,
+      "loss": 0.8555,
+      "step": 1415
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665736530635419e-05,
+      "loss": 0.8667,
+      "step": 1416
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6652410923943448e-05,
+      "loss": 0.8257,
+      "step": 1417
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6647453610600387e-05,
+      "loss": 0.8384,
+      "step": 1418
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664249336850912e-05,
+      "loss": 0.9023,
+      "step": 1419
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6637530199855042e-05,
+      "loss": 0.8589,
+      "step": 1420
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.663256410682483e-05,
+      "loss": 0.8647,
+      "step": 1421
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.662759509160646e-05,
+      "loss": 0.8335,
+      "step": 1422
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6622623156389186e-05,
+      "loss": 0.8281,
+      "step": 1423
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6617648303363555e-05,
+      "loss": 0.8569,
+      "step": 1424
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.66126705347214e-05,
+      "loss": 0.8862,
+      "step": 1425
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660768985265583e-05,
+      "loss": 0.8984,
+      "step": 1426
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660270625936125e-05,
+      "loss": 0.834,
+      "step": 1427
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6597719757033336e-05,
+      "loss": 0.8579,
+      "step": 1428
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6592730347869055e-05,
+      "loss": 0.8311,
+      "step": 1429
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6587738034066653e-05,
+      "loss": 0.7793,
+      "step": 1430
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6582742817825646e-05,
+      "loss": 0.8843,
+      "step": 1431
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657774470134684e-05,
+      "loss": 0.8301,
+      "step": 1432
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6572743686832317e-05,
+      "loss": 0.8408,
+      "step": 1433
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.656773977648543e-05,
+      "loss": 0.8374,
+      "step": 1434
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6562732972510816e-05,
+      "loss": 0.8447,
+      "step": 1435
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655772327711438e-05,
+      "loss": 0.8115,
+      "step": 1436
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.65527106925033e-05,
+      "loss": 0.7944,
+      "step": 1437
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654769522088604e-05,
+      "loss": 0.8784,
+      "step": 1438
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6542676864472317e-05,
+      "loss": 0.8789,
+      "step": 1439
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6537655625473133e-05,
+      "loss": 0.9048,
+      "step": 1440
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6532631506100753e-05,
+      "loss": 0.7993,
+      "step": 1441
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6527604508568714e-05,
+      "loss": 0.8652,
+      "step": 1442
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6522574635091818e-05,
+      "loss": 0.8379,
+      "step": 1443
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6517541887886147e-05,
+      "loss": 0.8848,
+      "step": 1444
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651250626916903e-05,
+      "loss": 0.9106,
+      "step": 1445
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6507467781159067e-05,
+      "loss": 0.8188,
+      "step": 1446
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6502426426076134e-05,
+      "loss": 0.8193,
+      "step": 1447
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.649738220614135e-05,
+      "loss": 0.8486,
+      "step": 1448
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6492335123577117e-05,
+      "loss": 0.8979,
+      "step": 1449
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6487285180607085e-05,
+      "loss": 0.9268,
+      "step": 1450
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6482232379456168e-05,
+      "loss": 0.9146,
+      "step": 1451
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647717672235054e-05,
+      "loss": 0.8374,
+      "step": 1452
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6472118211517632e-05,
+      "loss": 0.8447,
+      "step": 1453
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646705684918613e-05,
+      "loss": 0.8877,
+      "step": 1454
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6461992637585983e-05,
+      "loss": 0.8584,
+      "step": 1455
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6456925578948388e-05,
+      "loss": 0.854,
+      "step": 1456
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6451855675505802e-05,
+      "loss": 0.8804,
+      "step": 1457
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6446782929491934e-05,
+      "loss": 0.835,
+      "step": 1458
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644170734314174e-05,
+      "loss": 0.8276,
+      "step": 1459
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6436628918691434e-05,
+      "loss": 0.8989,
+      "step": 1460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6431547658378478e-05,
+      "loss": 0.8477,
+      "step": 1461
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6426463564441585e-05,
+      "loss": 0.8677,
+      "step": 1462
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6421376639120717e-05,
+      "loss": 0.8579,
+      "step": 1463
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641628688465708e-05,
+      "loss": 0.8813,
+      "step": 1464
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641119430329312e-05,
+      "loss": 0.8423,
+      "step": 1465
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640609889727255e-05,
+      "loss": 0.8179,
+      "step": 1466
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6401000668840306e-05,
+      "loss": 0.8413,
+      "step": 1467
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6395899620242577e-05,
+      "loss": 0.7822,
+      "step": 1468
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639079575372679e-05,
+      "loss": 0.8237,
+      "step": 1469
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6385689071541627e-05,
+      "loss": 0.791,
+      "step": 1470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.638057957593699e-05,
+      "loss": 0.8672,
+      "step": 1471
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637546726916403e-05,
+      "loss": 0.8457,
+      "step": 1472
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6370352153475147e-05,
+      "loss": 0.9185,
+      "step": 1473
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6365234231123955e-05,
+      "loss": 0.8345,
+      "step": 1474
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6360113504365333e-05,
+      "loss": 0.8901,
+      "step": 1475
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6354989975455368e-05,
+      "loss": 0.8989,
+      "step": 1476
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63498636466514e-05,
+      "loss": 0.8618,
+      "step": 1477
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6344734520212e-05,
+      "loss": 0.8638,
+      "step": 1478
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6339602598396967e-05,
+      "loss": 0.8569,
+      "step": 1479
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633446788346733e-05,
+      "loss": 0.8994,
+      "step": 1480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6329330377685356e-05,
+      "loss": 0.8398,
+      "step": 1481
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.632419008331453e-05,
+      "loss": 0.8374,
+      "step": 1482
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6319047002619584e-05,
+      "loss": 0.8975,
+      "step": 1483
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6313901137866456e-05,
+      "loss": 0.8657,
+      "step": 1484
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630875249132233e-05,
+      "loss": 0.793,
+      "step": 1485
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63036010652556e-05,
+      "loss": 0.8618,
+      "step": 1486
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6298446861935898e-05,
+      "loss": 0.8877,
+      "step": 1487
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6293289883634066e-05,
+      "loss": 0.8872,
+      "step": 1488
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.628813013262218e-05,
+      "loss": 0.8145,
+      "step": 1489
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6282967611173536e-05,
+      "loss": 0.8398,
+      "step": 1490
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6277802321562642e-05,
+      "loss": 0.8452,
+      "step": 1491
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.627263426606523e-05,
+      "loss": 0.8037,
+      "step": 1492
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626746344695826e-05,
+      "loss": 0.814,
+      "step": 1493
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6262289866519897e-05,
+      "loss": 0.8618,
+      "step": 1494
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6257113527029527e-05,
+      "loss": 0.8052,
+      "step": 1495
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6251934430767752e-05,
+      "loss": 0.8315,
+      "step": 1496
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.624675258001639e-05,
+      "loss": 0.8672,
+      "step": 1497
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.624156797705847e-05,
+      "loss": 0.8857,
+      "step": 1498
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623638062417823e-05,
+      "loss": 0.8345,
+      "step": 1499
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623119052366114e-05,
+      "loss": 0.8911,
+      "step": 1500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6225997677793842e-05,
+      "loss": 0.8047,
+      "step": 1501
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6220802088864227e-05,
+      "loss": 0.873,
+      "step": 1502
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6215603759161372e-05,
+      "loss": 0.8018,
+      "step": 1503
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6210402690975565e-05,
+      "loss": 0.7891,
+      "step": 1504
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620519888659831e-05,
+      "loss": 0.8599,
+      "step": 1505
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6199992348322305e-05,
+      "loss": 0.895,
+      "step": 1506
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6194783078441458e-05,
+      "loss": 0.8652,
+      "step": 1507
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6189571079250875e-05,
+      "loss": 0.8052,
+      "step": 1508
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6184356353046873e-05,
+      "loss": 0.8555,
+      "step": 1509
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6179138902126968e-05,
+      "loss": 0.8271,
+      "step": 1510
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617391872878987e-05,
+      "loss": 0.8506,
+      "step": 1511
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6168695835335496e-05,
+      "loss": 0.8442,
+      "step": 1512
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6163470224064958e-05,
+      "loss": 0.8374,
+      "step": 1513
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6158241897280564e-05,
+      "loss": 0.8047,
+      "step": 1514
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.615301085728583e-05,
+      "loss": 0.8647,
+      "step": 1515
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6147777106385446e-05,
+      "loss": 0.8745,
+      "step": 1516
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6142540646885316e-05,
+      "loss": 0.8262,
+      "step": 1517
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6137301481092523e-05,
+      "loss": 0.8398,
+      "step": 1518
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6132059611315352e-05,
+      "loss": 0.8418,
+      "step": 1519
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6126815039863282e-05,
+      "loss": 0.8369,
+      "step": 1520
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612156776904697e-05,
+      "loss": 0.8433,
+      "step": 1521
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6116317801178273e-05,
+      "loss": 0.8438,
+      "step": 1522
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6111065138570223e-05,
+      "loss": 0.917,
+      "step": 1523
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6105809783537063e-05,
+      "loss": 0.8481,
+      "step": 1524
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6100551738394194e-05,
+      "loss": 0.8638,
+      "step": 1525
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6095291005458223e-05,
+      "loss": 0.8247,
+      "step": 1526
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6090027587046935e-05,
+      "loss": 0.8877,
+      "step": 1527
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6084761485479297e-05,
+      "loss": 0.834,
+      "step": 1528
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6079492703075455e-05,
+      "loss": 0.8438,
+      "step": 1529
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.607422124215674e-05,
+      "loss": 0.7935,
+      "step": 1530
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.606894710504566e-05,
+      "loss": 0.874,
+      "step": 1531
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6063670294065914e-05,
+      "loss": 0.8047,
+      "step": 1532
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6058390811542363e-05,
+      "loss": 0.8711,
+      "step": 1533
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6053108659801053e-05,
+      "loss": 0.8242,
+      "step": 1534
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6047823841169204e-05,
+      "loss": 0.834,
+      "step": 1535
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6042536357975215e-05,
+      "loss": 0.876,
+      "step": 1536
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6037246212548653e-05,
+      "loss": 0.8477,
+      "step": 1537
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6031953407220257e-05,
+      "loss": 0.8735,
+      "step": 1538
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6026657944321948e-05,
+      "loss": 0.8804,
+      "step": 1539
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6021359826186812e-05,
+      "loss": 0.8105,
+      "step": 1540
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.60160590551491e-05,
+      "loss": 0.8618,
+      "step": 1541
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6010755633544235e-05,
+      "loss": 0.8462,
+      "step": 1542
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6005449563708814e-05,
+      "loss": 0.8301,
+      "step": 1543
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6000140847980593e-05,
+      "loss": 0.8403,
+      "step": 1544
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.59948294886985e-05,
+      "loss": 0.8281,
+      "step": 1545
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5989515488202616e-05,
+      "loss": 0.8374,
+      "step": 1546
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.59841988488342e-05,
+      "loss": 0.8511,
+      "step": 1547
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5978879572935667e-05,
+      "loss": 0.813,
+      "step": 1548
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5973557662850594e-05,
+      "loss": 0.7991,
+      "step": 1549
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596823312092372e-05,
+      "loss": 0.8696,
+      "step": 1550
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596290594950093e-05,
+      "loss": 0.8257,
+      "step": 1551
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.59575761509293e-05,
+      "loss": 0.8911,
+      "step": 1552
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.595224372755703e-05,
+      "loss": 0.835,
+      "step": 1553
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.594690868173349e-05,
+      "loss": 0.8384,
+      "step": 1554
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5941571015809207e-05,
+      "loss": 0.833,
+      "step": 1555
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5936230732135866e-05,
+      "loss": 0.855,
+      "step": 1556
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.593088783306629e-05,
+      "loss": 0.7488,
+      "step": 1557
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.592554232095447e-05,
+      "loss": 0.8638,
+      "step": 1558
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5920194198155542e-05,
+      "loss": 0.8149,
+      "step": 1559
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5914843467025788e-05,
+      "loss": 0.8379,
+      "step": 1560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.590949012992265e-05,
+      "loss": 0.8726,
+      "step": 1561
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5904134189204706e-05,
+      "loss": 0.7871,
+      "step": 1562
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.589877564723169e-05,
+      "loss": 0.8208,
+      "step": 1563
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5893414506364483e-05,
+      "loss": 0.8042,
+      "step": 1564
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.58880507689651e-05,
+      "loss": 0.8013,
+      "step": 1565
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5882684437396707e-05,
+      "loss": 0.7769,
+      "step": 1566
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587731551402362e-05,
+      "loss": 0.8442,
+      "step": 1567
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587194400121128e-05,
+      "loss": 0.877,
+      "step": 1568
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586656990132629e-05,
+      "loss": 0.8481,
+      "step": 1569
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5861193216736372e-05,
+      "loss": 0.7886,
+      "step": 1570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5855813949810397e-05,
+      "loss": 0.8306,
+      "step": 1571
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5850432102918383e-05,
+      "loss": 0.8501,
+      "step": 1572
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5845047678431463e-05,
+      "loss": 0.8184,
+      "step": 1573
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.583966067872192e-05,
+      "loss": 0.8979,
+      "step": 1574
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.583427110616318e-05,
+      "loss": 0.8276,
+      "step": 1575
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5828878963129773e-05,
+      "loss": 0.7798,
+      "step": 1576
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5823484251997392e-05,
+      "loss": 0.8574,
+      "step": 1577
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5818086975142847e-05,
+      "loss": 0.7554,
+      "step": 1578
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.581268713494408e-05,
+      "loss": 0.8667,
+      "step": 1579
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.580728473378017e-05,
+      "loss": 0.7788,
+      "step": 1580
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5801879774031302e-05,
+      "loss": 0.9233,
+      "step": 1581
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.579647225807882e-05,
+      "loss": 0.8901,
+      "step": 1582
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.579106218830517e-05,
+      "loss": 0.7876,
+      "step": 1583
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5785649567093933e-05,
+      "loss": 0.8267,
+      "step": 1584
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.578023439682981e-05,
+      "loss": 0.8306,
+      "step": 1585
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.577481667989863e-05,
+      "loss": 0.8311,
+      "step": 1586
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.576939641868734e-05,
+      "loss": 0.8271,
+      "step": 1587
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.576397361558401e-05,
+      "loss": 0.8354,
+      "step": 1588
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.575854827297783e-05,
+      "loss": 0.8457,
+      "step": 1589
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5753120393259106e-05,
+      "loss": 0.8101,
+      "step": 1590
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5747689978819268e-05,
+      "loss": 0.8096,
+      "step": 1591
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5742257032050855e-05,
+      "loss": 0.8818,
+      "step": 1592
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5736821555347525e-05,
+      "loss": 0.8569,
+      "step": 1593
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5731383551104054e-05,
+      "loss": 0.9214,
+      "step": 1594
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5725943021716324e-05,
+      "loss": 0.8174,
+      "step": 1595
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5720499969581342e-05,
+      "loss": 0.8301,
+      "step": 1596
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571505439709721e-05,
+      "loss": 0.813,
+      "step": 1597
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5709606306663155e-05,
+      "loss": 0.8701,
+      "step": 1598
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5704155700679506e-05,
+      "loss": 0.8345,
+      "step": 1599
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.56987025815477e-05,
+      "loss": 0.8164,
+      "step": 1600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5693246951670285e-05,
+      "loss": 0.8325,
+      "step": 1601
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.568778881345091e-05,
+      "loss": 0.7773,
+      "step": 1602
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5682328169294338e-05,
+      "loss": 0.8687,
+      "step": 1603
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567686502160642e-05,
+      "loss": 0.8066,
+      "step": 1604
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5671399372794133e-05,
+      "loss": 0.8477,
+      "step": 1605
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.566593122526554e-05,
+      "loss": 0.8359,
+      "step": 1606
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.56604605814298e-05,
+      "loss": 0.8125,
+      "step": 1607
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5654987443697186e-05,
+      "loss": 0.8232,
+      "step": 1608
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5649511814479065e-05,
+      "loss": 0.8135,
+      "step": 1609
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.56440336961879e-05,
+      "loss": 0.8794,
+      "step": 1610
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563855309123725e-05,
+      "loss": 0.7944,
+      "step": 1611
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5633070002041772e-05,
+      "loss": 0.8618,
+      "step": 1612
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5627584431017213e-05,
+      "loss": 0.8501,
+      "step": 1613
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562209638058042e-05,
+      "loss": 0.8892,
+      "step": 1614
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.561660585314933e-05,
+      "loss": 0.7988,
+      "step": 1615
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.561111285114296e-05,
+      "loss": 0.8218,
+      "step": 1616
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5605617376981447e-05,
+      "loss": 0.8315,
+      "step": 1617
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.560011943308598e-05,
+      "loss": 0.7808,
+      "step": 1618
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5594619021878864e-05,
+      "loss": 0.8237,
+      "step": 1619
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5589116145783477e-05,
+      "loss": 0.8311,
+      "step": 1620
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5583610807224284e-05,
+      "loss": 0.8223,
+      "step": 1621
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5578103008626848e-05,
+      "loss": 0.8916,
+      "step": 1622
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5572592752417797e-05,
+      "loss": 0.8018,
+      "step": 1623
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5567080041024852e-05,
+      "loss": 0.8838,
+      "step": 1624
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5561564876876817e-05,
+      "loss": 0.8872,
+      "step": 1625
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.555604726240357e-05,
+      "loss": 0.8237,
+      "step": 1626
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5550527200036077e-05,
+      "loss": 0.8599,
+      "step": 1627
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5545004692206372e-05,
+      "loss": 0.8501,
+      "step": 1628
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5539479741347578e-05,
+      "loss": 0.8257,
+      "step": 1629
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5533952349893887e-05,
+      "loss": 0.8237,
+      "step": 1630
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5528422520280568e-05,
+      "loss": 0.8726,
+      "step": 1631
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.552289025494396e-05,
+      "loss": 0.8247,
+      "step": 1632
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5517355556321487e-05,
+      "loss": 0.8652,
+      "step": 1633
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5511818426851634e-05,
+      "loss": 0.8086,
+      "step": 1634
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.550627886897396e-05,
+      "loss": 0.8369,
+      "step": 1635
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.550073688512909e-05,
+      "loss": 0.8384,
+      "step": 1636
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5495192477758734e-05,
+      "loss": 0.8208,
+      "step": 1637
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5489645649305643e-05,
+      "loss": 0.8115,
+      "step": 1638
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548409640221366e-05,
+      "loss": 0.8721,
+      "step": 1639
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5478544738927682e-05,
+      "loss": 0.8652,
+      "step": 1640
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547299066189367e-05,
+      "loss": 0.834,
+      "step": 1641
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.546743417355865e-05,
+      "loss": 0.8506,
+      "step": 1642
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.546187527637071e-05,
+      "loss": 0.8784,
+      "step": 1643
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5456313972779002e-05,
+      "loss": 0.8052,
+      "step": 1644
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5450750265233732e-05,
+      "loss": 0.894,
+      "step": 1645
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5445184156186176e-05,
+      "loss": 0.8389,
+      "step": 1646
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5439615648088653e-05,
+      "loss": 0.8433,
+      "step": 1647
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5434044743394555e-05,
+      "loss": 0.8628,
+      "step": 1648
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5428471444558314e-05,
+      "loss": 0.8672,
+      "step": 1649
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5422895754035435e-05,
+      "loss": 0.8477,
+      "step": 1650
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541731767428246e-05,
+      "loss": 0.854,
+      "step": 1651
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541173720775699e-05,
+      "loss": 0.8008,
+      "step": 1652
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.540615435691768e-05,
+      "loss": 0.8608,
+      "step": 1653
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5400569124224237e-05,
+      "loss": 0.8481,
+      "step": 1654
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5394981512137406e-05,
+      "loss": 0.8506,
+      "step": 1655
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5389391523118994e-05,
+      "loss": 0.8203,
+      "step": 1656
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5383799159631843e-05,
+      "loss": 0.8296,
+      "step": 1657
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5378204424139858e-05,
+      "loss": 0.8579,
+      "step": 1658
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5372607319107972e-05,
+      "loss": 0.855,
+      "step": 1659
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5367007847002166e-05,
+      "loss": 0.7993,
+      "step": 1660
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5361406010289473e-05,
+      "loss": 0.8066,
+      "step": 1661
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5355801811437955e-05,
+      "loss": 0.7969,
+      "step": 1662
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5350195252916724e-05,
+      "loss": 0.834,
+      "step": 1663
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5344586337195933e-05,
+      "loss": 0.8042,
+      "step": 1664
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5338975066746757e-05,
+      "loss": 0.7891,
+      "step": 1665
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533336144404143e-05,
+      "loss": 0.8569,
+      "step": 1666
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5327745471553215e-05,
+      "loss": 0.8257,
+      "step": 1667
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5322127151756406e-05,
+      "loss": 0.8076,
+      "step": 1668
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.531650648712633e-05,
+      "loss": 0.8604,
+      "step": 1669
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.531088348013935e-05,
+      "loss": 0.751,
+      "step": 1670
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530525813327286e-05,
+      "loss": 0.8257,
+      "step": 1671
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5299630449005295e-05,
+      "loss": 0.8721,
+      "step": 1672
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5294000429816108e-05,
+      "loss": 0.8174,
+      "step": 1673
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528836807818578e-05,
+      "loss": 0.8521,
+      "step": 1674
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5282733396595824e-05,
+      "loss": 0.8027,
+      "step": 1675
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5277096387528784e-05,
+      "loss": 0.8735,
+      "step": 1676
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5271457053468223e-05,
+      "loss": 0.918,
+      "step": 1677
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5265815396898724e-05,
+      "loss": 0.8452,
+      "step": 1678
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5260171420305904e-05,
+      "loss": 0.8472,
+      "step": 1679
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.52545251261764e-05,
+      "loss": 0.7834,
+      "step": 1680
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5248876516997864e-05,
+      "loss": 0.8193,
+      "step": 1681
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.524322559525897e-05,
+      "loss": 0.8721,
+      "step": 1682
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5237572363449418e-05,
+      "loss": 0.8315,
+      "step": 1683
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5231916824059912e-05,
+      "loss": 0.8838,
+      "step": 1684
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5226258979582182e-05,
+      "loss": 0.8833,
+      "step": 1685
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5220598832508979e-05,
+      "loss": 0.7939,
+      "step": 1686
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5214936385334059e-05,
+      "loss": 0.8276,
+      "step": 1687
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5209271640552187e-05,
+      "loss": 0.8164,
+      "step": 1688
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5203604600659161e-05,
+      "loss": 0.8687,
+      "step": 1689
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5197935268151764e-05,
+      "loss": 0.8037,
+      "step": 1690
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5192263645527808e-05,
+      "loss": 0.8389,
+      "step": 1691
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5186589735286104e-05,
+      "loss": 0.7559,
+      "step": 1692
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5180913539926478e-05,
+      "loss": 0.7524,
+      "step": 1693
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.517523506194976e-05,
+      "loss": 0.8047,
+      "step": 1694
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5169554303857778e-05,
+      "loss": 0.8306,
+      "step": 1695
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5163871268153383e-05,
+      "loss": 0.8193,
+      "step": 1696
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5158185957340407e-05,
+      "loss": 0.8364,
+      "step": 1697
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5152498373923701e-05,
+      "loss": 0.8179,
+      "step": 1698
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5146808520409115e-05,
+      "loss": 0.7764,
+      "step": 1699
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5141116399303486e-05,
+      "loss": 0.8418,
+      "step": 1700
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.513542201311467e-05,
+      "loss": 0.8589,
+      "step": 1701
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5129725364351504e-05,
+      "loss": 0.7959,
+      "step": 1702
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.512402645552383e-05,
+      "loss": 0.8594,
+      "step": 1703
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.511832528914249e-05,
+      "loss": 0.8281,
+      "step": 1704
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5112621867719309e-05,
+      "loss": 0.8022,
+      "step": 1705
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5106916193767114e-05,
+      "loss": 0.854,
+      "step": 1706
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5101208269799718e-05,
+      "loss": 0.8765,
+      "step": 1707
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5095498098331937e-05,
+      "loss": 0.8145,
+      "step": 1708
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5089785681879562e-05,
+      "loss": 0.8325,
+      "step": 1709
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5084071022959388e-05,
+      "loss": 0.8018,
+      "step": 1710
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5078354124089187e-05,
+      "loss": 0.8584,
+      "step": 1711
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5072634987787717e-05,
+      "loss": 0.7998,
+      "step": 1712
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5066913616574734e-05,
+      "loss": 0.854,
+      "step": 1713
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5061190012970968e-05,
+      "loss": 0.8994,
+      "step": 1714
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5055464179498137e-05,
+      "loss": 0.8145,
+      "step": 1715
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5049736118678934e-05,
+      "loss": 0.8125,
+      "step": 1716
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5044005833037048e-05,
+      "loss": 0.8555,
+      "step": 1717
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5038273325097135e-05,
+      "loss": 0.8506,
+      "step": 1718
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5032538597384832e-05,
+      "loss": 0.8584,
+      "step": 1719
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5026801652426761e-05,
+      "loss": 0.8198,
+      "step": 1720
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5021062492750513e-05,
+      "loss": 0.9243,
+      "step": 1721
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5015321120884662e-05,
+      "loss": 0.8379,
+      "step": 1722
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5009577539358753e-05,
+      "loss": 0.855,
+      "step": 1723
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5003831750703302e-05,
+      "loss": 0.8521,
+      "step": 1724
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4998083757449799e-05,
+      "loss": 0.7954,
+      "step": 1725
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4992333562130709e-05,
+      "loss": 0.8228,
+      "step": 1726
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4986581167279464e-05,
+      "loss": 0.811,
+      "step": 1727
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4980826575430462e-05,
+      "loss": 0.8198,
+      "step": 1728
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4975069789119079e-05,
+      "loss": 0.8403,
+      "step": 1729
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4969310810881644e-05,
+      "loss": 0.8379,
+      "step": 1730
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4963549643255464e-05,
+      "loss": 0.8174,
+      "step": 1731
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4957786288778801e-05,
+      "loss": 0.7798,
+      "step": 1732
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.495202074999089e-05,
+      "loss": 0.7935,
+      "step": 1733
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4946253029431923e-05,
+      "loss": 0.8462,
+      "step": 1734
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4940483129643048e-05,
+      "loss": 0.8379,
+      "step": 1735
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4934711053166386e-05,
+      "loss": 0.8223,
+      "step": 1736
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4928936802545004e-05,
+      "loss": 0.8647,
+      "step": 1737
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.492316038032294e-05,
+      "loss": 0.814,
+      "step": 1738
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4917381789045174e-05,
+      "loss": 0.8511,
+      "step": 1739
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4911601031257653e-05,
+      "loss": 0.8174,
+      "step": 1740
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4905818109507274e-05,
+      "loss": 0.7998,
+      "step": 1741
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4900033026341883e-05,
+      "loss": 0.8472,
+      "step": 1742
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4894245784310293e-05,
+      "loss": 0.8218,
+      "step": 1743
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4888456385962253e-05,
+      "loss": 0.8208,
+      "step": 1744
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4882664833848467e-05,
+      "loss": 0.8452,
+      "step": 1745
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4876871130520591e-05,
+      "loss": 0.8281,
+      "step": 1746
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4871075278531222e-05,
+      "loss": 0.8535,
+      "step": 1747
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4865277280433912e-05,
+      "loss": 0.853,
+      "step": 1748
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4859477138783154e-05,
+      "loss": 0.8213,
+      "step": 1749
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4853674856134384e-05,
+      "loss": 0.8774,
+      "step": 1750
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4847870435043982e-05,
+      "loss": 0.8921,
+      "step": 1751
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4842063878069276e-05,
+      "loss": 0.8154,
+      "step": 1752
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.483625518776853e-05,
+      "loss": 0.8413,
+      "step": 1753
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4830444366700942e-05,
+      "loss": 0.8457,
+      "step": 1754
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4824631417426661e-05,
+      "loss": 0.8335,
+      "step": 1755
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4818816342506763e-05,
+      "loss": 0.8394,
+      "step": 1756
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4812999144503268e-05,
+      "loss": 0.8657,
+      "step": 1757
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4807179825979127e-05,
+      "loss": 0.8247,
+      "step": 1758
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4801358389498233e-05,
+      "loss": 0.8755,
+      "step": 1759
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4795534837625398e-05,
+      "loss": 0.792,
+      "step": 1760
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4789709172926377e-05,
+      "loss": 0.8691,
+      "step": 1761
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4783881397967851e-05,
+      "loss": 0.8306,
+      "step": 1762
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4778051515317438e-05,
+      "loss": 0.8159,
+      "step": 1763
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4772219527543672e-05,
+      "loss": 0.856,
+      "step": 1764
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4766385437216026e-05,
+      "loss": 0.8193,
+      "step": 1765
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4760549246904896e-05,
+      "loss": 0.8081,
+      "step": 1766
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4754710959181599e-05,
+      "loss": 0.8623,
+      "step": 1767
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4748870576618382e-05,
+      "loss": 0.8193,
+      "step": 1768
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4743028101788408e-05,
+      "loss": 0.8145,
+      "step": 1769
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4737183537265772e-05,
+      "loss": 0.8628,
+      "step": 1770
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.473133688562548e-05,
+      "loss": 0.8506,
+      "step": 1771
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4725488149443462e-05,
+      "loss": 0.8301,
+      "step": 1772
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4719637331296566e-05,
+      "loss": 0.835,
+      "step": 1773
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4713784433762554e-05,
+      "loss": 0.8418,
+      "step": 1774
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4707929459420114e-05,
+      "loss": 0.8262,
+      "step": 1775
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4702072410848838e-05,
+      "loss": 0.8882,
+      "step": 1776
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4696213290629231e-05,
+      "loss": 0.8521,
+      "step": 1777
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4690352101342726e-05,
+      "loss": 0.8364,
+      "step": 1778
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.468448884557165e-05,
+      "loss": 0.8623,
+      "step": 1779
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.467862352589925e-05,
+      "loss": 0.8447,
+      "step": 1780
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4672756144909681e-05,
+      "loss": 0.8618,
+      "step": 1781
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4666886705188002e-05,
+      "loss": 0.791,
+      "step": 1782
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4661015209320186e-05,
+      "loss": 0.8364,
+      "step": 1783
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4655141659893107e-05,
+      "loss": 0.7998,
+      "step": 1784
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4649266059494543e-05,
+      "loss": 0.8442,
+      "step": 1785
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.464338841071318e-05,
+      "loss": 0.772,
+      "step": 1786
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4637508716138605e-05,
+      "loss": 0.874,
+      "step": 1787
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4631626978361304e-05,
+      "loss": 0.8511,
+      "step": 1788
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.462574319997266e-05,
+      "loss": 0.8354,
+      "step": 1789
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4619857383564968e-05,
+      "loss": 0.8447,
+      "step": 1790
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4613969531731409e-05,
+      "loss": 0.8096,
+      "step": 1791
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4608079647066062e-05,
+      "loss": 0.8774,
+      "step": 1792
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4602187732163909e-05,
+      "loss": 0.7671,
+      "step": 1793
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4596293789620814e-05,
+      "loss": 0.792,
+      "step": 1794
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4590397822033553e-05,
+      "loss": 0.8555,
+      "step": 1795
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.458449983199977e-05,
+      "loss": 0.8179,
+      "step": 1796
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4578599822118022e-05,
+      "loss": 0.8354,
+      "step": 1797
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4572697794987745e-05,
+      "loss": 0.8364,
+      "step": 1798
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4566793753209263e-05,
+      "loss": 0.8174,
+      "step": 1799
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4560887699383794e-05,
+      "loss": 0.835,
+      "step": 1800
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.455497963611344e-05,
+      "loss": 0.8413,
+      "step": 1801
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4549069566001181e-05,
+      "loss": 0.8242,
+      "step": 1802
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.454315749165089e-05,
+      "loss": 0.7964,
+      "step": 1803
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4537243415667324e-05,
+      "loss": 0.8462,
+      "step": 1804
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4531327340656114e-05,
+      "loss": 0.7495,
+      "step": 1805
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4525409269223779e-05,
+      "loss": 0.855,
+      "step": 1806
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4519489203977708e-05,
+      "loss": 0.8198,
+      "step": 1807
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4513567147526186e-05,
+      "loss": 0.8809,
+      "step": 1808
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4507643102478355e-05,
+      "loss": 0.8232,
+      "step": 1809
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4501717071444248e-05,
+      "loss": 0.7837,
+      "step": 1810
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4495789057034766e-05,
+      "loss": 0.7759,
+      "step": 1811
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.448985906186168e-05,
+      "loss": 0.7979,
+      "step": 1812
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4483927088537646e-05,
+      "loss": 0.8193,
+      "step": 1813
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.447799313967618e-05,
+      "loss": 0.8418,
+      "step": 1814
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4472057217891678e-05,
+      "loss": 0.7871,
+      "step": 1815
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4466119325799394e-05,
+      "loss": 0.8901,
+      "step": 1816
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.446017946601546e-05,
+      "loss": 0.8535,
+      "step": 1817
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4454237641156872e-05,
+      "loss": 0.877,
+      "step": 1818
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4448293853841486e-05,
+      "loss": 0.8247,
+      "step": 1819
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4442348106688033e-05,
+      "loss": 0.8882,
+      "step": 1820
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4436400402316099e-05,
+      "loss": 0.8589,
+      "step": 1821
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4430450743346137e-05,
+      "loss": 0.8359,
+      "step": 1822
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4424499132399459e-05,
+      "loss": 0.7993,
+      "step": 1823
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.441854557209823e-05,
+      "loss": 0.8657,
+      "step": 1824
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4412590065065497e-05,
+      "loss": 0.8462,
+      "step": 1825
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4406632613925136e-05,
+      "loss": 0.8774,
+      "step": 1826
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4400673221301898e-05,
+      "loss": 0.8013,
+      "step": 1827
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.439471188982138e-05,
+      "loss": 0.8833,
+      "step": 1828
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4388748622110043e-05,
+      "loss": 0.7666,
+      "step": 1829
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4382783420795193e-05,
+      "loss": 0.8545,
+      "step": 1830
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4376816288504985e-05,
+      "loss": 0.7749,
+      "step": 1831
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4370847227868436e-05,
+      "loss": 0.8398,
+      "step": 1832
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4364876241515403e-05,
+      "loss": 0.8213,
+      "step": 1833
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4358903332076601e-05,
+      "loss": 0.8496,
+      "step": 1834
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4352928502183579e-05,
+      "loss": 0.8032,
+      "step": 1835
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4346951754468746e-05,
+      "loss": 0.8838,
+      "step": 1836
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.434097309156534e-05,
+      "loss": 0.873,
+      "step": 1837
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.433499251610746e-05,
+      "loss": 0.8774,
+      "step": 1838
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4329010030730035e-05,
+      "loss": 0.8433,
+      "step": 1839
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4323025638068843e-05,
+      "loss": 0.8027,
+      "step": 1840
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4317039340760501e-05,
+      "loss": 0.8613,
+      "step": 1841
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4311051141442461e-05,
+      "loss": 0.8311,
+      "step": 1842
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4305061042753017e-05,
+      "loss": 0.8926,
+      "step": 1843
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4299069047331295e-05,
+      "loss": 0.8174,
+      "step": 1844
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4293075157817265e-05,
+      "loss": 0.8066,
+      "step": 1845
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4287079376851724e-05,
+      "loss": 0.9082,
+      "step": 1846
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4281081707076305e-05,
+      "loss": 0.8345,
+      "step": 1847
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4275082151133476e-05,
+      "loss": 0.8379,
+      "step": 1848
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4269080711666525e-05,
+      "loss": 0.8296,
+      "step": 1849
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.426307739131959e-05,
+      "loss": 0.9092,
+      "step": 1850
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4257072192737618e-05,
+      "loss": 0.8813,
+      "step": 1851
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4251065118566392e-05,
+      "loss": 0.7515,
+      "step": 1852
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4245056171452523e-05,
+      "loss": 0.8667,
+      "step": 1853
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.423904535404344e-05,
+      "loss": 0.874,
+      "step": 1854
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4233032668987406e-05,
+      "loss": 0.8096,
+      "step": 1855
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4227018118933497e-05,
+      "loss": 0.8672,
+      "step": 1856
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.422100170653162e-05,
+      "loss": 0.771,
+      "step": 1857
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4214983434432496e-05,
+      "loss": 0.8418,
+      "step": 1858
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4208963305287667e-05,
+      "loss": 0.8193,
+      "step": 1859
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4202941321749493e-05,
+      "loss": 0.8784,
+      "step": 1860
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4196917486471152e-05,
+      "loss": 0.7534,
+      "step": 1861
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4190891802106635e-05,
+      "loss": 0.8154,
+      "step": 1862
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4184864271310752e-05,
+      "loss": 0.874,
+      "step": 1863
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4178834896739124e-05,
+      "loss": 0.8379,
+      "step": 1864
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4172803681048185e-05,
+      "loss": 0.769,
+      "step": 1865
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4166770626895179e-05,
+      "loss": 0.8071,
+      "step": 1866
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4160735736938165e-05,
+      "loss": 0.8467,
+      "step": 1867
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4154699013836e-05,
+      "loss": 0.8687,
+      "step": 1868
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.414866046024836e-05,
+      "loss": 0.8662,
+      "step": 1869
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.414262007883572e-05,
+      "loss": 0.79,
+      "step": 1870
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4136577872259365e-05,
+      "loss": 0.8008,
+      "step": 1871
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4130533843181385e-05,
+      "loss": 0.8223,
+      "step": 1872
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4124487994264664e-05,
+      "loss": 0.8228,
+      "step": 1873
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4118440328172895e-05,
+      "loss": 0.792,
+      "step": 1874
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4112390847570576e-05,
+      "loss": 0.7881,
+      "step": 1875
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4106339555122994e-05,
+      "loss": 0.8315,
+      "step": 1876
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4100286453496244e-05,
+      "loss": 0.9048,
+      "step": 1877
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.409423154535721e-05,
+      "loss": 0.8101,
+      "step": 1878
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4088174833373576e-05,
+      "loss": 0.8921,
+      "step": 1879
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4082116320213817e-05,
+      "loss": 0.7949,
+      "step": 1880
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.407605600854721e-05,
+      "loss": 0.8252,
+      "step": 1881
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4069993901043817e-05,
+      "loss": 0.8057,
+      "step": 1882
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4063930000374495e-05,
+      "loss": 0.8301,
+      "step": 1883
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4057864309210883e-05,
+      "loss": 0.8589,
+      "step": 1884
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4051796830225424e-05,
+      "loss": 0.854,
+      "step": 1885
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4045727566091332e-05,
+      "loss": 0.8047,
+      "step": 1886
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.403965651948262e-05,
+      "loss": 0.8193,
+      "step": 1887
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4033583693074081e-05,
+      "loss": 0.8125,
+      "step": 1888
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4027509089541292e-05,
+      "loss": 0.8867,
+      "step": 1889
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4021432711560614e-05,
+      "loss": 0.853,
+      "step": 1890
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4015354561809187e-05,
+      "loss": 0.833,
+      "step": 1891
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.400927464296494e-05,
+      "loss": 0.8384,
+      "step": 1892
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4003192957706575e-05,
+      "loss": 0.8057,
+      "step": 1893
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3997109508713563e-05,
+      "loss": 0.8564,
+      "step": 1894
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3991024298666176e-05,
+      "loss": 0.8218,
+      "step": 1895
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3984937330245436e-05,
+      "loss": 0.8076,
+      "step": 1896
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3978848606133159e-05,
+      "loss": 0.8633,
+      "step": 1897
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3972758129011925e-05,
+      "loss": 0.8203,
+      "step": 1898
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3966665901565087e-05,
+      "loss": 0.8579,
+      "step": 1899
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3960571926476774e-05,
+      "loss": 0.811,
+      "step": 1900
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3954476206431874e-05,
+      "loss": 0.8901,
+      "step": 1901
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3948378744116062e-05,
+      "loss": 0.854,
+      "step": 1902
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3942279542215764e-05,
+      "loss": 0.8433,
+      "step": 1903
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3936178603418176e-05,
+      "loss": 0.7959,
+      "step": 1904
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3930075930411265e-05,
+      "loss": 0.875,
+      "step": 1905
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.392397152588376e-05,
+      "loss": 0.8271,
+      "step": 1906
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.391786539252515e-05,
+      "loss": 0.8369,
+      "step": 1907
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3911757533025684e-05,
+      "loss": 0.8013,
+      "step": 1908
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3905647950076382e-05,
+      "loss": 0.8125,
+      "step": 1909
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3899536646369009e-05,
+      "loss": 0.875,
+      "step": 1910
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3893423624596098e-05,
+      "loss": 0.8311,
+      "step": 1911
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3887308887450937e-05,
+      "loss": 0.8516,
+      "step": 1912
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3881192437627572e-05,
+      "loss": 0.8472,
+      "step": 1913
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.387507427782079e-05,
+      "loss": 0.8789,
+      "step": 1914
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3868954410726154e-05,
+      "loss": 0.8242,
+      "step": 1915
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.386283283903996e-05,
+      "loss": 0.7524,
+      "step": 1916
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3856709565459263e-05,
+      "loss": 0.8462,
+      "step": 1917
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3850584592681871e-05,
+      "loss": 0.8374,
+      "step": 1918
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3844457923406327e-05,
+      "loss": 0.8286,
+      "step": 1919
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3838329560331943e-05,
+      "loss": 0.8403,
+      "step": 1920
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.383219950615876e-05,
+      "loss": 0.8525,
+      "step": 1921
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3826067763587565e-05,
+      "loss": 0.8286,
+      "step": 1922
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3819934335319897e-05,
+      "loss": 0.7964,
+      "step": 1923
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3813799224058038e-05,
+      "loss": 0.8633,
+      "step": 1924
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3807662432505002e-05,
+      "loss": 0.8062,
+      "step": 1925
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.380152396336455e-05,
+      "loss": 0.8262,
+      "step": 1926
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3795383819341184e-05,
+      "loss": 0.8115,
+      "step": 1927
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3789242003140136e-05,
+      "loss": 0.8735,
+      "step": 1928
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3783098517467379e-05,
+      "loss": 0.7866,
+      "step": 1929
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3776953365029628e-05,
+      "loss": 0.8223,
+      "step": 1930
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3770806548534319e-05,
+      "loss": 0.8608,
+      "step": 1931
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3764658070689633e-05,
+      "loss": 0.8315,
+      "step": 1932
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3758507934204479e-05,
+      "loss": 0.8931,
+      "step": 1933
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3752356141788496e-05,
+      "loss": 0.8091,
+      "step": 1934
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3746202696152052e-05,
+      "loss": 0.8096,
+      "step": 1935
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3740047600006241e-05,
+      "loss": 0.8462,
+      "step": 1936
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3733890856062893e-05,
+      "loss": 0.7681,
+      "step": 1937
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3727732467034556e-05,
+      "loss": 0.8472,
+      "step": 1938
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3721572435634502e-05,
+      "loss": 0.9077,
+      "step": 1939
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3715410764576736e-05,
+      "loss": 0.7905,
+      "step": 1940
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3709247456575974e-05,
+      "loss": 0.8594,
+      "step": 1941
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.370308251434766e-05,
+      "loss": 0.8159,
+      "step": 1942
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3696915940607957e-05,
+      "loss": 0.8022,
+      "step": 1943
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3690747738073749e-05,
+      "loss": 0.7781,
+      "step": 1944
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3684577909462626e-05,
+      "loss": 0.7578,
+      "step": 1945
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3678406457492912e-05,
+      "loss": 0.8296,
+      "step": 1946
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3672233384883636e-05,
+      "loss": 0.8296,
+      "step": 1947
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3666058694354538e-05,
+      "loss": 0.7993,
+      "step": 1948
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3659882388626077e-05,
+      "loss": 0.8999,
+      "step": 1949
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.365370447041943e-05,
+      "loss": 0.8667,
+      "step": 1950
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3647524942456465e-05,
+      "loss": 0.7949,
+      "step": 1951
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3641343807459775e-05,
+      "loss": 0.8062,
+      "step": 1952
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3635161068152657e-05,
+      "loss": 0.8545,
+      "step": 1953
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3628976727259113e-05,
+      "loss": 0.854,
+      "step": 1954
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3622790787503855e-05,
+      "loss": 0.8491,
+      "step": 1955
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3616603251612293e-05,
+      "loss": 0.8843,
+      "step": 1956
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3610414122310549e-05,
+      "loss": 0.8413,
+      "step": 1957
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3604223402325436e-05,
+      "loss": 0.853,
+      "step": 1958
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3598031094384475e-05,
+      "loss": 0.8262,
+      "step": 1959
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3591837201215889e-05,
+      "loss": 0.8667,
+      "step": 1960
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.358564172554859e-05,
+      "loss": 0.8496,
+      "step": 1961
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.35794446701122e-05,
+      "loss": 0.8418,
+      "step": 1962
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3573246037637023e-05,
+      "loss": 0.7598,
+      "step": 1963
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3567045830854066e-05,
+      "loss": 0.8418,
+      "step": 1964
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3560844052495032e-05,
+      "loss": 0.8149,
+      "step": 1965
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3554640705292307e-05,
+      "loss": 0.8145,
+      "step": 1966
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3548435791978982e-05,
+      "loss": 0.8281,
+      "step": 1967
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3542229315288822e-05,
+      "loss": 0.8325,
+      "step": 1968
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3536021277956292e-05,
+      "loss": 0.875,
+      "step": 1969
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3529811682716542e-05,
+      "loss": 0.8003,
+      "step": 1970
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3523600532305408e-05,
+      "loss": 0.7944,
+      "step": 1971
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3517387829459412e-05,
+      "loss": 0.7983,
+      "step": 1972
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.351117357691575e-05,
+      "loss": 0.8457,
+      "step": 1973
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3504957777412322e-05,
+      "loss": 0.8721,
+      "step": 1974
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3498740433687689e-05,
+      "loss": 0.8687,
+      "step": 1975
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3492521548481104e-05,
+      "loss": 0.8794,
+      "step": 1976
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3486301124532494e-05,
+      "loss": 0.8735,
+      "step": 1977
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3480079164582467e-05,
+      "loss": 0.8389,
+      "step": 1978
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3473855671372309e-05,
+      "loss": 0.894,
+      "step": 1979
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3467630647643972e-05,
+      "loss": 0.811,
+      "step": 1980
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3461404096140093e-05,
+      "loss": 0.8633,
+      "step": 1981
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.345517601960398e-05,
+      "loss": 0.7681,
+      "step": 1982
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3448946420779607e-05,
+      "loss": 0.8896,
+      "step": 1983
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3442715302411627e-05,
+      "loss": 0.8521,
+      "step": 1984
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3436482667245359e-05,
+      "loss": 0.7793,
+      "step": 1985
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3430248518026786e-05,
+      "loss": 0.8135,
+      "step": 1986
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3424012857502559e-05,
+      "loss": 0.8154,
+      "step": 1987
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3417775688420008e-05,
+      "loss": 0.8359,
+      "step": 1988
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3411537013527106e-05,
+      "loss": 0.8174,
+      "step": 1989
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.340529683557251e-05,
+      "loss": 0.8599,
+      "step": 1990
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.339905515730552e-05,
+      "loss": 0.7773,
+      "step": 1991
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3392811981476119e-05,
+      "loss": 0.8232,
+      "step": 1992
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3386567310834928e-05,
+      "loss": 0.8574,
+      "step": 1993
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3380321148133242e-05,
+      "loss": 0.8604,
+      "step": 1994
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3374073496123004e-05,
+      "loss": 0.8359,
+      "step": 1995
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3367824357556815e-05,
+      "loss": 0.8442,
+      "step": 1996
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.336157373518794e-05,
+      "loss": 0.856,
+      "step": 1997
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.335532163177028e-05,
+      "loss": 0.7769,
+      "step": 1998
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3349068050058414e-05,
+      "loss": 0.8267,
+      "step": 1999
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3342812992807546e-05,
+      "loss": 0.8662,
+      "step": 2000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3336556462773542e-05,
+      "loss": 0.8296,
+      "step": 2001
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3330298462712921e-05,
+      "loss": 0.8354,
+      "step": 2002
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.332403899538284e-05,
+      "loss": 0.8213,
+      "step": 2003
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3317778063541113e-05,
+      "loss": 0.8418,
+      "step": 2004
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3311515669946187e-05,
+      "loss": 0.7983,
+      "step": 2005
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3305251817357165e-05,
+      "loss": 0.8315,
+      "step": 2006
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3298986508533788e-05,
+      "loss": 0.7458,
+      "step": 2007
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.329271974623643e-05,
+      "loss": 0.8325,
+      "step": 2008
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3286451533226127e-05,
+      "loss": 0.8818,
+      "step": 2009
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.328018187226453e-05,
+      "loss": 0.813,
+      "step": 2010
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3273910766113942e-05,
+      "loss": 0.8594,
+      "step": 2011
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3267638217537297e-05,
+      "loss": 0.8628,
+      "step": 2012
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.326136422929817e-05,
+      "loss": 0.8936,
+      "step": 2013
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3255088804160768e-05,
+      "loss": 0.8066,
+      "step": 2014
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3248811944889926e-05,
+      "loss": 0.8008,
+      "step": 2015
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3242533654251116e-05,
+      "loss": 0.8047,
+      "step": 2016
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3236253935010441e-05,
+      "loss": 0.8057,
+      "step": 2017
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3229972789934633e-05,
+      "loss": 0.7847,
+      "step": 2018
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.322369022179105e-05,
+      "loss": 0.8447,
+      "step": 2019
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3217406233347678e-05,
+      "loss": 0.8364,
+      "step": 2020
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3211120827373129e-05,
+      "loss": 0.8379,
+      "step": 2021
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.320483400663664e-05,
+      "loss": 0.7925,
+      "step": 2022
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3198545773908072e-05,
+      "loss": 0.8384,
+      "step": 2023
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3192256131957903e-05,
+      "loss": 0.7542,
+      "step": 2024
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3185965083557242e-05,
+      "loss": 0.7852,
+      "step": 2025
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3179672631477804e-05,
+      "loss": 0.7881,
+      "step": 2026
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3173378778491939e-05,
+      "loss": 0.877,
+      "step": 2027
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3167083527372598e-05,
+      "loss": 0.8677,
+      "step": 2028
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3160786880893355e-05,
+      "loss": 0.8643,
+      "step": 2029
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3154488841828404e-05,
+      "loss": 0.8105,
+      "step": 2030
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3148189412952542e-05,
+      "loss": 0.769,
+      "step": 2031
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.314188859704119e-05,
+      "loss": 0.8623,
+      "step": 2032
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3135586396870366e-05,
+      "loss": 0.8564,
+      "step": 2033
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3129282815216715e-05,
+      "loss": 0.8535,
+      "step": 2034
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3122977854857473e-05,
+      "loss": 0.8179,
+      "step": 2035
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3116671518570496e-05,
+      "loss": 0.8604,
+      "step": 2036
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3110363809134243e-05,
+      "loss": 0.8521,
+      "step": 2037
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3104054729327773e-05,
+      "loss": 0.873,
+      "step": 2038
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3097744281930758e-05,
+      "loss": 0.8599,
+      "step": 2039
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3091432469723462e-05,
+      "loss": 0.8823,
+      "step": 2040
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3085119295486759e-05,
+      "loss": 0.856,
+      "step": 2041
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3078804762002122e-05,
+      "loss": 0.8364,
+      "step": 2042
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3072488872051611e-05,
+      "loss": 0.8608,
+      "step": 2043
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3066171628417907e-05,
+      "loss": 0.7622,
+      "step": 2044
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3059853033884262e-05,
+      "loss": 0.8164,
+      "step": 2045
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3053533091234542e-05,
+      "loss": 0.8252,
+      "step": 2046
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3047211803253195e-05,
+      "loss": 0.8345,
+      "step": 2047
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3040889172725269e-05,
+      "loss": 0.835,
+      "step": 2048
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3034565202436405e-05,
+      "loss": 0.8418,
+      "step": 2049
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3028239895172821e-05,
+      "loss": 0.8057,
+      "step": 2050
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3021913253721339e-05,
+      "loss": 0.8125,
+      "step": 2051
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3015585280869361e-05,
+      "loss": 0.8354,
+      "step": 2052
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3009255979404882e-05,
+      "loss": 0.791,
+      "step": 2053
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3002925352116474e-05,
+      "loss": 0.8877,
+      "step": 2054
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2996593401793298e-05,
+      "loss": 0.8574,
+      "step": 2055
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.29902601312251e-05,
+      "loss": 0.7773,
+      "step": 2056
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.29839255432022e-05,
+      "loss": 0.8135,
+      "step": 2057
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2977589640515507e-05,
+      "loss": 0.8086,
+      "step": 2058
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2971252425956504e-05,
+      "loss": 0.8511,
+      "step": 2059
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2964913902317254e-05,
+      "loss": 0.8271,
+      "step": 2060
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2958574072390396e-05,
+      "loss": 0.814,
+      "step": 2061
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2952232938969145e-05,
+      "loss": 0.8711,
+      "step": 2062
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.294589050484729e-05,
+      "loss": 0.8838,
+      "step": 2063
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.293954677281919e-05,
+      "loss": 0.8394,
+      "step": 2064
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2933201745679785e-05,
+      "loss": 0.8223,
+      "step": 2065
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2926855426224574e-05,
+      "loss": 0.7969,
+      "step": 2066
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2920507817249636e-05,
+      "loss": 0.8237,
+      "step": 2067
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2914158921551607e-05,
+      "loss": 0.8628,
+      "step": 2068
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2907808741927702e-05,
+      "loss": 0.8188,
+      "step": 2069
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2901457281175693e-05,
+      "loss": 0.7803,
+      "step": 2070
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.289510454209392e-05,
+      "loss": 0.8228,
+      "step": 2071
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2888750527481285e-05,
+      "loss": 0.7856,
+      "step": 2072
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2882395240137246e-05,
+      "loss": 0.8525,
+      "step": 2073
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.287603868286184e-05,
+      "loss": 0.8359,
+      "step": 2074
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2869680858455642e-05,
+      "loss": 0.8467,
+      "step": 2075
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2863321769719798e-05,
+      "loss": 0.7988,
+      "step": 2076
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2856961419456012e-05,
+      "loss": 0.8569,
+      "step": 2077
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2850599810466529e-05,
+      "loss": 0.7686,
+      "step": 2078
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.284423694555417e-05,
+      "loss": 0.8101,
+      "step": 2079
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.283787282752229e-05,
+      "loss": 0.7563,
+      "step": 2080
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2831507459174814e-05,
+      "loss": 0.8457,
+      "step": 2081
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2825140843316202e-05,
+      "loss": 0.8184,
+      "step": 2082
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2818772982751471e-05,
+      "loss": 0.8247,
+      "step": 2083
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2812403880286188e-05,
+      "loss": 0.8394,
+      "step": 2084
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2806033538726459e-05,
+      "loss": 0.7695,
+      "step": 2085
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2799661960878948e-05,
+      "loss": 0.8623,
+      "step": 2086
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2793289149550856e-05,
+      "loss": 0.7993,
+      "step": 2087
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2786915107549927e-05,
+      "loss": 0.8555,
+      "step": 2088
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2780539837684449e-05,
+      "loss": 0.8755,
+      "step": 2089
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.277416334276325e-05,
+      "loss": 0.8594,
+      "step": 2090
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2767785625595703e-05,
+      "loss": 0.8462,
+      "step": 2091
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2761406688991713e-05,
+      "loss": 0.7661,
+      "step": 2092
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.275502653576172e-05,
+      "loss": 0.8921,
+      "step": 2093
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2748645168716712e-05,
+      "loss": 0.8057,
+      "step": 2094
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.27422625906682e-05,
+      "loss": 0.7896,
+      "step": 2095
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2735878804428233e-05,
+      "loss": 0.8315,
+      "step": 2096
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2729493812809392e-05,
+      "loss": 0.7666,
+      "step": 2097
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.272310761862479e-05,
+      "loss": 0.8813,
+      "step": 2098
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.271672022468807e-05,
+      "loss": 0.8457,
+      "step": 2099
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.27103316338134e-05,
+      "loss": 0.7783,
+      "step": 2100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.270394184881548e-05,
+      "loss": 0.8413,
+      "step": 2101
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2697550872509533e-05,
+      "loss": 0.813,
+      "step": 2102
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2691158707711306e-05,
+      "loss": 0.8672,
+      "step": 2103
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2684765357237079e-05,
+      "loss": 0.8047,
+      "step": 2104
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.267837082390364e-05,
+      "loss": 0.8359,
+      "step": 2105
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2671975110528304e-05,
+      "loss": 0.8105,
+      "step": 2106
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2665578219928914e-05,
+      "loss": 0.8716,
+      "step": 2107
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2659180154923818e-05,
+      "loss": 0.8154,
+      "step": 2108
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2652780918331894e-05,
+      "loss": 0.8496,
+      "step": 2109
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2646380512972521e-05,
+      "loss": 0.8237,
+      "step": 2110
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2639978941665614e-05,
+      "loss": 0.8286,
+      "step": 2111
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.263357620723158e-05,
+      "loss": 0.8403,
+      "step": 2112
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2627172312491354e-05,
+      "loss": 0.7959,
+      "step": 2113
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2620767260266378e-05,
+      "loss": 0.7856,
+      "step": 2114
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2614361053378594e-05,
+      "loss": 0.7837,
+      "step": 2115
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.260795369465047e-05,
+      "loss": 0.8228,
+      "step": 2116
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2601545186904968e-05,
+      "loss": 0.8701,
+      "step": 2117
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2595135532965563e-05,
+      "loss": 0.8848,
+      "step": 2118
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2588724735656235e-05,
+      "loss": 0.8179,
+      "step": 2119
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2582312797801459e-05,
+      "loss": 0.853,
+      "step": 2120
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2575899722226229e-05,
+      "loss": 0.8306,
+      "step": 2121
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2569485511756019e-05,
+      "loss": 0.8306,
+      "step": 2122
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.256307016921682e-05,
+      "loss": 0.79,
+      "step": 2123
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2556653697435119e-05,
+      "loss": 0.8208,
+      "step": 2124
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2550236099237896e-05,
+      "loss": 0.7837,
+      "step": 2125
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2543817377452628e-05,
+      "loss": 0.8228,
+      "step": 2126
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2537397534907288e-05,
+      "loss": 0.8613,
+      "step": 2127
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2530976574430342e-05,
+      "loss": 0.7808,
+      "step": 2128
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2524554498850752e-05,
+      "loss": 0.8291,
+      "step": 2129
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2518131310997968e-05,
+      "loss": 0.7637,
+      "step": 2130
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2511707013701928e-05,
+      "loss": 0.7832,
+      "step": 2131
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2505281609793066e-05,
+      "loss": 0.8843,
+      "step": 2132
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2498855102102297e-05,
+      "loss": 0.8281,
+      "step": 2133
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2492427493461024e-05,
+      "loss": 0.8062,
+      "step": 2134
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2485998786701135e-05,
+      "loss": 0.7842,
+      "step": 2135
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2479568984655002e-05,
+      "loss": 0.7832,
+      "step": 2136
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2473138090155483e-05,
+      "loss": 0.8174,
+      "step": 2137
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2466706106035911e-05,
+      "loss": 0.8154,
+      "step": 2138
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2460273035130106e-05,
+      "loss": 0.7373,
+      "step": 2139
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2453838880272358e-05,
+      "loss": 0.7944,
+      "step": 2140
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.244740364429744e-05,
+      "loss": 0.7715,
+      "step": 2141
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2440967330040606e-05,
+      "loss": 0.8628,
+      "step": 2142
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2434529940337576e-05,
+      "loss": 0.7842,
+      "step": 2143
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.242809147802455e-05,
+      "loss": 0.9165,
+      "step": 2144
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2421651945938194e-05,
+      "loss": 0.8345,
+      "step": 2145
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2415211346915657e-05,
+      "loss": 0.77,
+      "step": 2146
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2408769683794544e-05,
+      "loss": 0.7439,
+      "step": 2147
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2402326959412938e-05,
+      "loss": 0.811,
+      "step": 2148
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2395883176609391e-05,
+      "loss": 0.8188,
+      "step": 2149
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.238943833822291e-05,
+      "loss": 0.8154,
+      "step": 2150
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2382992447092983e-05,
+      "loss": 0.8716,
+      "step": 2151
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2376545506059545e-05,
+      "loss": 0.8521,
+      "step": 2152
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2370097517963006e-05,
+      "loss": 0.8213,
+      "step": 2153
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2363648485644237e-05,
+      "loss": 0.7886,
+      "step": 2154
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.235719841194456e-05,
+      "loss": 0.877,
+      "step": 2155
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2350747299705763e-05,
+      "loss": 0.7666,
+      "step": 2156
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2344295151770087e-05,
+      "loss": 0.8691,
+      "step": 2157
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.233784197098024e-05,
+      "loss": 0.8066,
+      "step": 2158
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2331387760179368e-05,
+      "loss": 0.8335,
+      "step": 2159
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2324932522211084e-05,
+      "loss": 0.8398,
+      "step": 2160
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2318476259919451e-05,
+      "loss": 0.7998,
+      "step": 2161
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2312018976148976e-05,
+      "loss": 0.8423,
+      "step": 2162
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2305560673744631e-05,
+      "loss": 0.7393,
+      "step": 2163
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2299101355551822e-05,
+      "loss": 0.8433,
+      "step": 2164
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2292641024416404e-05,
+      "loss": 0.7974,
+      "step": 2165
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.228617968318469e-05,
+      "loss": 0.8101,
+      "step": 2166
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2279717334703424e-05,
+      "loss": 0.7842,
+      "step": 2167
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.227325398181981e-05,
+      "loss": 0.7705,
+      "step": 2168
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2266789627381475e-05,
+      "loss": 0.8433,
+      "step": 2169
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2260324274236499e-05,
+      "loss": 0.8398,
+      "step": 2170
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2253857925233401e-05,
+      "loss": 0.8145,
+      "step": 2171
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.224739058322114e-05,
+      "loss": 0.8257,
+      "step": 2172
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2240922251049105e-05,
+      "loss": 0.8208,
+      "step": 2173
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2234452931567135e-05,
+      "loss": 0.8613,
+      "step": 2174
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.222798262762549e-05,
+      "loss": 0.8584,
+      "step": 2175
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2221511342074867e-05,
+      "loss": 0.8945,
+      "step": 2176
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2215039077766402e-05,
+      "loss": 0.7666,
+      "step": 2177
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2208565837551653e-05,
+      "loss": 0.793,
+      "step": 2178
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2202091624282622e-05,
+      "loss": 0.8521,
+      "step": 2179
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.219561644081172e-05,
+      "loss": 0.7979,
+      "step": 2180
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2189140289991805e-05,
+      "loss": 0.7979,
+      "step": 2181
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2182663174676147e-05,
+      "loss": 0.8003,
+      "step": 2182
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2176185097718447e-05,
+      "loss": 0.8306,
+      "step": 2183
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.216970606197283e-05,
+      "loss": 0.8447,
+      "step": 2184
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2163226070293841e-05,
+      "loss": 0.7715,
+      "step": 2185
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.215674512553645e-05,
+      "loss": 0.8354,
+      "step": 2186
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2150263230556041e-05,
+      "loss": 0.7793,
+      "step": 2187
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2143780388208424e-05,
+      "loss": 0.8745,
+      "step": 2188
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2137296601349817e-05,
+      "loss": 0.8569,
+      "step": 2189
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2130811872836865e-05,
+      "loss": 0.833,
+      "step": 2190
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2124326205526622e-05,
+      "loss": 0.8081,
+      "step": 2191
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.211783960227655e-05,
+      "loss": 0.7358,
+      "step": 2192
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2111352065944536e-05,
+      "loss": 0.8267,
+      "step": 2193
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2104863599388873e-05,
+      "loss": 0.7666,
+      "step": 2194
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2098374205468253e-05,
+      "loss": 0.8394,
+      "step": 2195
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2091883887041795e-05,
+      "loss": 0.8096,
+      "step": 2196
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.208539264696901e-05,
+      "loss": 0.8042,
+      "step": 2197
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2078900488109828e-05,
+      "loss": 0.7798,
+      "step": 2198
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2072407413324572e-05,
+      "loss": 0.8809,
+      "step": 2199
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2065913425473973e-05,
+      "loss": 0.8325,
+      "step": 2200
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.205941852741917e-05,
+      "loss": 0.7817,
+      "step": 2201
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2052922722021694e-05,
+      "loss": 0.8389,
+      "step": 2202
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2046426012143484e-05,
+      "loss": 0.8643,
+      "step": 2203
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2039928400646865e-05,
+      "loss": 0.8179,
+      "step": 2204
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2033429890394575e-05,
+      "loss": 0.8481,
+      "step": 2205
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2026930484249738e-05,
+      "loss": 0.8081,
+      "step": 2206
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2020430185075874e-05,
+      "loss": 0.8408,
+      "step": 2207
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2013928995736898e-05,
+      "loss": 0.835,
+      "step": 2208
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.200742691909712e-05,
+      "loss": 0.8052,
+      "step": 2209
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2000923958021235e-05,
+      "loss": 0.7944,
+      "step": 2210
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.199442011537433e-05,
+      "loss": 0.8169,
+      "step": 2211
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.198791539402188e-05,
+      "loss": 0.8604,
+      "step": 2212
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1981409796829753e-05,
+      "loss": 0.8179,
+      "step": 2213
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1974903326664191e-05,
+      "loss": 0.8174,
+      "step": 2214
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1968395986391832e-05,
+      "loss": 0.8535,
+      "step": 2215
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1961887778879693e-05,
+      "loss": 0.8315,
+      "step": 2216
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.195537870699517e-05,
+      "loss": 0.8398,
+      "step": 2217
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1948868773606044e-05,
+      "loss": 0.8618,
+      "step": 2218
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1942357981580472e-05,
+      "loss": 0.8525,
+      "step": 2219
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1935846333786998e-05,
+      "loss": 0.875,
+      "step": 2220
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1929333833094532e-05,
+      "loss": 0.8467,
+      "step": 2221
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1922820482372362e-05,
+      "loss": 0.8428,
+      "step": 2222
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.191630628449016e-05,
+      "loss": 0.7969,
+      "step": 2223
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1909791242317957e-05,
+      "loss": 0.8096,
+      "step": 2224
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1903275358726168e-05,
+      "loss": 0.7944,
+      "step": 2225
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.189675863658557e-05,
+      "loss": 0.8198,
+      "step": 2226
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1890241078767316e-05,
+      "loss": 0.8335,
+      "step": 2227
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1883722688142925e-05,
+      "loss": 0.8511,
+      "step": 2228
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.187720346758428e-05,
+      "loss": 0.771,
+      "step": 2229
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1870683419963631e-05,
+      "loss": 0.8013,
+      "step": 2230
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.18641625481536e-05,
+      "loss": 0.7959,
+      "step": 2231
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1857640855027157e-05,
+      "loss": 0.8218,
+      "step": 2232
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.185111834345765e-05,
+      "loss": 0.8462,
+      "step": 2233
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1844595016318774e-05,
+      "loss": 0.8296,
+      "step": 2234
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1838070876484593e-05,
+      "loss": 0.8306,
+      "step": 2235
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1831545926829522e-05,
+      "loss": 0.77,
+      "step": 2236
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1825020170228341e-05,
+      "loss": 0.8311,
+      "step": 2237
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.181849360955618e-05,
+      "loss": 0.8564,
+      "step": 2238
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1811966247688516e-05,
+      "loss": 0.8506,
+      "step": 2239
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1805438087501198e-05,
+      "loss": 0.8369,
+      "step": 2240
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1798909131870407e-05,
+      "loss": 0.8306,
+      "step": 2241
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.179237938367269e-05,
+      "loss": 0.7891,
+      "step": 2242
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.178584884578493e-05,
+      "loss": 0.8062,
+      "step": 2243
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1779317521084368e-05,
+      "loss": 0.8896,
+      "step": 2244
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1772785412448591e-05,
+      "loss": 0.8442,
+      "step": 2245
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1766252522755521e-05,
+      "loss": 0.8545,
+      "step": 2246
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1759718854883434e-05,
+      "loss": 0.8066,
+      "step": 2247
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.175318441171095e-05,
+      "loss": 0.8408,
+      "step": 2248
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1746649196117024e-05,
+      "loss": 0.8672,
+      "step": 2249
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1740113210980952e-05,
+      "loss": 0.8311,
+      "step": 2250
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1733576459182376e-05,
+      "loss": 0.8623,
+      "step": 2251
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1727038943601268e-05,
+      "loss": 0.8047,
+      "step": 2252
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1720500667117937e-05,
+      "loss": 0.8101,
+      "step": 2253
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1713961632613036e-05,
+      "loss": 0.8247,
+      "step": 2254
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1707421842967541e-05,
+      "loss": 0.8726,
+      "step": 2255
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1700881301062767e-05,
+      "loss": 0.7969,
+      "step": 2256
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1694340009780356e-05,
+      "loss": 0.8115,
+      "step": 2257
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1687797972002291e-05,
+      "loss": 0.8618,
+      "step": 2258
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1681255190610868e-05,
+      "loss": 0.8164,
+      "step": 2259
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1674711668488719e-05,
+      "loss": 0.8037,
+      "step": 2260
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1668167408518808e-05,
+      "loss": 0.7749,
+      "step": 2261
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1661622413584408e-05,
+      "loss": 0.7637,
+      "step": 2262
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1655076686569137e-05,
+      "loss": 0.8013,
+      "step": 2263
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1648530230356914e-05,
+      "loss": 0.7803,
+      "step": 2264
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1641983047832e-05,
+      "loss": 0.7988,
+      "step": 2265
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1635435141878957e-05,
+      "loss": 0.8315,
+      "step": 2266
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1628886515382675e-05,
+      "loss": 0.8115,
+      "step": 2267
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1622337171228368e-05,
+      "loss": 0.7798,
+      "step": 2268
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.161578711230155e-05,
+      "loss": 0.7861,
+      "step": 2269
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1609236341488066e-05,
+      "loss": 0.8345,
+      "step": 2270
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1602684861674064e-05,
+      "loss": 0.7437,
+      "step": 2271
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1596132675746006e-05,
+      "loss": 0.8022,
+      "step": 2272
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1589579786590673e-05,
+      "loss": 0.8018,
+      "step": 2273
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1583026197095141e-05,
+      "loss": 0.8169,
+      "step": 2274
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1576471910146815e-05,
+      "loss": 0.7485,
+      "step": 2275
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1569916928633385e-05,
+      "loss": 0.7559,
+      "step": 2276
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.156336125544286e-05,
+      "loss": 0.7769,
+      "step": 2277
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1556804893463556e-05,
+      "loss": 0.8311,
+      "step": 2278
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1550247845584084e-05,
+      "loss": 0.8286,
+      "step": 2279
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1543690114693364e-05,
+      "loss": 0.8472,
+      "step": 2280
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.153713170368061e-05,
+      "loss": 0.8071,
+      "step": 2281
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1530572615435343e-05,
+      "loss": 0.7866,
+      "step": 2282
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1524012852847377e-05,
+      "loss": 0.7822,
+      "step": 2283
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1517452418806826e-05,
+      "loss": 0.7896,
+      "step": 2284
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.15108913162041e-05,
+      "loss": 0.8125,
+      "step": 2285
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1504329547929901e-05,
+      "loss": 0.8086,
+      "step": 2286
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1497767116875227e-05,
+      "loss": 0.8291,
+      "step": 2287
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1491204025931363e-05,
+      "loss": 0.7822,
+      "step": 2288
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1484640277989894e-05,
+      "loss": 0.8701,
+      "step": 2289
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1478075875942688e-05,
+      "loss": 0.8223,
+      "step": 2290
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1471510822681898e-05,
+      "loss": 0.8271,
+      "step": 2291
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1464945121099971e-05,
+      "loss": 0.8042,
+      "step": 2292
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1458378774089638e-05,
+      "loss": 0.8105,
+      "step": 2293
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.145181178454391e-05,
+      "loss": 0.8442,
+      "step": 2294
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1445244155356087e-05,
+      "loss": 0.8369,
+      "step": 2295
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1438675889419746e-05,
+      "loss": 0.8047,
+      "step": 2296
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1432106989628748e-05,
+      "loss": 0.8159,
+      "step": 2297
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1425537458877232e-05,
+      "loss": 0.7622,
+      "step": 2298
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.141896730005961e-05,
+      "loss": 0.8052,
+      "step": 2299
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1412396516070585e-05,
+      "loss": 0.7983,
+      "step": 2300
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.140582510980512e-05,
+      "loss": 0.8135,
+      "step": 2301
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1399253084158455e-05,
+      "loss": 0.8418,
+      "step": 2302
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1392680442026114e-05,
+      "loss": 0.8208,
+      "step": 2303
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1386107186303877e-05,
+      "loss": 0.8042,
+      "step": 2304
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1379533319887805e-05,
+      "loss": 0.7959,
+      "step": 2305
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1372958845674226e-05,
+      "loss": 0.877,
+      "step": 2306
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1366383766559734e-05,
+      "loss": 0.8735,
+      "step": 2307
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1359808085441194e-05,
+      "loss": 0.769,
+      "step": 2308
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1353231805215722e-05,
+      "loss": 0.8149,
+      "step": 2309
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.134665492878072e-05,
+      "loss": 0.8057,
+      "step": 2310
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1340077459033836e-05,
+      "loss": 0.8384,
+      "step": 2311
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1333499398872983e-05,
+      "loss": 0.8042,
+      "step": 2312
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.132692075119634e-05,
+      "loss": 0.7734,
+      "step": 2313
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1320341518902336e-05,
+      "loss": 0.8037,
+      "step": 2314
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1313761704889668e-05,
+      "loss": 0.8286,
+      "step": 2315
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1307181312057275e-05,
+      "loss": 0.8325,
+      "step": 2316
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.130060034330437e-05,
+      "loss": 0.8428,
+      "step": 2317
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.12940188015304e-05,
+      "loss": 0.8286,
+      "step": 2318
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.128743668963508e-05,
+      "loss": 0.793,
+      "step": 2319
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1280854010518364e-05,
+      "loss": 0.8447,
+      "step": 2320
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1274270767080465e-05,
+      "loss": 0.7686,
+      "step": 2321
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1267686962221843e-05,
+      "loss": 0.7769,
+      "step": 2322
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1261102598843198e-05,
+      "loss": 0.8242,
+      "step": 2323
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.125451767984549e-05,
+      "loss": 0.8599,
+      "step": 2324
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1247932208129909e-05,
+      "loss": 0.8091,
+      "step": 2325
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1241346186597897e-05,
+      "loss": 0.8794,
+      "step": 2326
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1234759618151137e-05,
+      "loss": 0.8027,
+      "step": 2327
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1228172505691555e-05,
+      "loss": 0.8604,
+      "step": 2328
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1221584852121307e-05,
+      "loss": 0.835,
+      "step": 2329
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1214996660342799e-05,
+      "loss": 0.8179,
+      "step": 2330
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1208407933258669e-05,
+      "loss": 0.8125,
+      "step": 2331
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1201818673771791e-05,
+      "loss": 0.8188,
+      "step": 2332
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1195228884785276e-05,
+      "loss": 0.7905,
+      "step": 2333
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1188638569202462e-05,
+      "loss": 0.8809,
+      "step": 2334
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1182047729926926e-05,
+      "loss": 0.8101,
+      "step": 2335
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1175456369862474e-05,
+      "loss": 0.7666,
+      "step": 2336
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1168864491913136e-05,
+      "loss": 0.8237,
+      "step": 2337
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1162272098983183e-05,
+      "loss": 0.8472,
+      "step": 2338
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.115567919397709e-05,
+      "loss": 0.7715,
+      "step": 2339
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1149085779799589e-05,
+      "loss": 0.7705,
+      "step": 2340
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1142491859355607e-05,
+      "loss": 0.8535,
+      "step": 2341
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1135897435550313e-05,
+      "loss": 0.8228,
+      "step": 2342
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1129302511289091e-05,
+      "loss": 0.8506,
+      "step": 2343
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1122707089477539e-05,
+      "loss": 0.7432,
+      "step": 2344
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.111611117302149e-05,
+      "loss": 0.8179,
+      "step": 2345
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1109514764826976e-05,
+      "loss": 0.8604,
+      "step": 2346
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1102917867800267e-05,
+      "loss": 0.7847,
+      "step": 2347
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.109632048484783e-05,
+      "loss": 0.8423,
+      "step": 2348
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1089722618876351e-05,
+      "loss": 0.8506,
+      "step": 2349
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1083124272792739e-05,
+      "loss": 0.7749,
+      "step": 2350
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1076525449504095e-05,
+      "loss": 0.7827,
+      "step": 2351
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1069926151917751e-05,
+      "loss": 0.7754,
+      "step": 2352
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1063326382941235e-05,
+      "loss": 0.8232,
+      "step": 2353
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1056726145482286e-05,
+      "loss": 0.8384,
+      "step": 2354
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1050125442448849e-05,
+      "loss": 0.854,
+      "step": 2355
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1043524276749078e-05,
+      "loss": 0.8071,
+      "step": 2356
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1036922651291327e-05,
+      "loss": 0.8423,
+      "step": 2357
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1030320568984148e-05,
+      "loss": 0.8066,
+      "step": 2358
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.10237180327363e-05,
+      "loss": 0.9048,
+      "step": 2359
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.101711504545675e-05,
+      "loss": 0.7949,
+      "step": 2360
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1010511610054649e-05,
+      "loss": 0.8179,
+      "step": 2361
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1003907729439349e-05,
+      "loss": 0.853,
+      "step": 2362
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0997303406520407e-05,
+      "loss": 0.8467,
+      "step": 2363
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0990698644207565e-05,
+      "loss": 0.8311,
+      "step": 2364
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0984093445410759e-05,
+      "loss": 0.8335,
+      "step": 2365
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0977487813040128e-05,
+      "loss": 0.8047,
+      "step": 2366
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0970881750005989e-05,
+      "loss": 0.8511,
+      "step": 2367
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.096427525921886e-05,
+      "loss": 0.7856,
+      "step": 2368
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.095766834358943e-05,
+      "loss": 0.7969,
+      "step": 2369
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.09510610060286e-05,
+      "loss": 0.7695,
+      "step": 2370
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0944453249447437e-05,
+      "loss": 0.8003,
+      "step": 2371
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0937845076757201e-05,
+      "loss": 0.8154,
+      "step": 2372
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0931236490869333e-05,
+      "loss": 0.8379,
+      "step": 2373
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0924627494695457e-05,
+      "loss": 0.7598,
+      "step": 2374
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.091801809114738e-05,
+      "loss": 0.8384,
+      "step": 2375
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0911408283137079e-05,
+      "loss": 0.7769,
+      "step": 2376
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0904798073576728e-05,
+      "loss": 0.8545,
+      "step": 2377
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0898187465378654e-05,
+      "loss": 0.874,
+      "step": 2378
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0891576461455378e-05,
+      "loss": 0.8091,
+      "step": 2379
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0884965064719593e-05,
+      "loss": 0.8115,
+      "step": 2380
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0878353278084148e-05,
+      "loss": 0.8364,
+      "step": 2381
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0871741104462092e-05,
+      "loss": 0.7788,
+      "step": 2382
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0865128546766619e-05,
+      "loss": 0.8057,
+      "step": 2383
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0858515607911108e-05,
+      "loss": 0.8223,
+      "step": 2384
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0851902290809102e-05,
+      "loss": 0.8408,
+      "step": 2385
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0845288598374301e-05,
+      "loss": 0.8203,
+      "step": 2386
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0838674533520587e-05,
+      "loss": 0.7959,
+      "step": 2387
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0832060099161996e-05,
+      "loss": 0.792,
+      "step": 2388
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0825445298212724e-05,
+      "loss": 0.7856,
+      "step": 2389
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0818830133587139e-05,
+      "loss": 0.7734,
+      "step": 2390
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0812214608199761e-05,
+      "loss": 0.8892,
+      "step": 2391
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0805598724965269e-05,
+      "loss": 0.8511,
+      "step": 2392
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0798982486798505e-05,
+      "loss": 0.8198,
+      "step": 2393
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0792365896614462e-05,
+      "loss": 0.7998,
+      "step": 2394
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0785748957328293e-05,
+      "loss": 0.7578,
+      "step": 2395
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0779131671855297e-05,
+      "loss": 0.7578,
+      "step": 2396
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0772514043110935e-05,
+      "loss": 0.8066,
+      "step": 2397
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0765896074010815e-05,
+      "loss": 0.8101,
+      "step": 2398
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0759277767470698e-05,
+      "loss": 0.7988,
+      "step": 2399
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0752659126406482e-05,
+      "loss": 0.7905,
+      "step": 2400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0746040153734225e-05,
+      "loss": 0.8428,
+      "step": 2401
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0739420852370128e-05,
+      "loss": 0.8003,
+      "step": 2402
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0732801225230534e-05,
+      "loss": 0.8691,
+      "step": 2403
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0726181275231935e-05,
+      "loss": 0.8428,
+      "step": 2404
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0719561005290959e-05,
+      "loss": 0.8584,
+      "step": 2405
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0712940418324375e-05,
+      "loss": 0.7886,
+      "step": 2406
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.07063195172491e-05,
+      "loss": 0.8311,
+      "step": 2407
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0699698304982178e-05,
+      "loss": 0.8296,
+      "step": 2408
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.06930767844408e-05,
+      "loss": 0.8286,
+      "step": 2409
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.068645495854229e-05,
+      "loss": 0.8271,
+      "step": 2410
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0679832830204095e-05,
+      "loss": 0.8774,
+      "step": 2411
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0673210402343818e-05,
+      "loss": 0.8354,
+      "step": 2412
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0666587677879175e-05,
+      "loss": 0.793,
+      "step": 2413
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0659964659728017e-05,
+      "loss": 0.8838,
+      "step": 2414
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0653341350808333e-05,
+      "loss": 0.812,
+      "step": 2415
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0646717754038226e-05,
+      "loss": 0.8438,
+      "step": 2416
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0640093872335936e-05,
+      "loss": 0.8594,
+      "step": 2417
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.063346970861983e-05,
+      "loss": 0.8252,
+      "step": 2418
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.062684526580839e-05,
+      "loss": 0.8198,
+      "step": 2419
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.062022054682023e-05,
+      "loss": 0.793,
+      "step": 2420
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0613595554574076e-05,
+      "loss": 0.8291,
+      "step": 2421
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0606970291988787e-05,
+      "loss": 0.8364,
+      "step": 2422
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0600344761983327e-05,
+      "loss": 0.8257,
+      "step": 2423
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0593718967476796e-05,
+      "loss": 0.8354,
+      "step": 2424
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0587092911388388e-05,
+      "loss": 0.8164,
+      "step": 2425
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0580466596637428e-05,
+      "loss": 0.8193,
+      "step": 2426
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0573840026143353e-05,
+      "loss": 0.7891,
+      "step": 2427
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0567213202825709e-05,
+      "loss": 0.8521,
+      "step": 2428
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0560586129604154e-05,
+      "loss": 0.7729,
+      "step": 2429
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0553958809398456e-05,
+      "loss": 0.8408,
+      "step": 2430
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0547331245128495e-05,
+      "loss": 0.8481,
+      "step": 2431
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.054070343971426e-05,
+      "loss": 0.7974,
+      "step": 2432
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0534075396075833e-05,
+      "loss": 0.7905,
+      "step": 2433
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.052744711713342e-05,
+      "loss": 0.7944,
+      "step": 2434
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0520818605807315e-05,
+      "loss": 0.8145,
+      "step": 2435
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0514189865017922e-05,
+      "loss": 0.7778,
+      "step": 2436
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0507560897685744e-05,
+      "loss": 0.7852,
+      "step": 2437
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.050093170673139e-05,
+      "loss": 0.7515,
+      "step": 2438
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0494302295075553e-05,
+      "loss": 0.8354,
+      "step": 2439
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0487672665639042e-05,
+      "loss": 0.8467,
+      "step": 2440
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0481042821342745e-05,
+      "loss": 0.8145,
+      "step": 2441
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0474412765107653e-05,
+      "loss": 0.7959,
+      "step": 2442
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0467782499854851e-05,
+      "loss": 0.7671,
+      "step": 2443
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0461152028505514e-05,
+      "loss": 0.8574,
+      "step": 2444
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0454521353980908e-05,
+      "loss": 0.812,
+      "step": 2445
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0447890479202388e-05,
+      "loss": 0.8003,
+      "step": 2446
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.04412594070914e-05,
+      "loss": 0.8423,
+      "step": 2447
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0434628140569471e-05,
+      "loss": 0.8491,
+      "step": 2448
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0427996682558219e-05,
+      "loss": 0.8394,
+      "step": 2449
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0421365035979346e-05,
+      "loss": 0.793,
+      "step": 2450
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0414733203754637e-05,
+      "loss": 0.7734,
+      "step": 2451
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0408101188805957e-05,
+      "loss": 0.7778,
+      "step": 2452
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0401468994055245e-05,
+      "loss": 0.8428,
+      "step": 2453
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.039483662242454e-05,
+      "loss": 0.8364,
+      "step": 2454
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0388204076835935e-05,
+      "loss": 0.8687,
+      "step": 2455
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0381571360211613e-05,
+      "loss": 0.7915,
+      "step": 2456
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0374938475473832e-05,
+      "loss": 0.7432,
+      "step": 2457
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0368305425544913e-05,
+      "loss": 0.7856,
+      "step": 2458
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0361672213347268e-05,
+      "loss": 0.8208,
+      "step": 2459
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0355038841803366e-05,
+      "loss": 0.7847,
+      "step": 2460
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0348405313835752e-05,
+      "loss": 0.8062,
+      "step": 2461
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0341771632367042e-05,
+      "loss": 0.7859,
+      "step": 2462
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0335137800319905e-05,
+      "loss": 0.7812,
+      "step": 2463
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0328503820617101e-05,
+      "loss": 0.9019,
+      "step": 2464
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0321869696181437e-05,
+      "loss": 0.7744,
+      "step": 2465
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0315235429935786e-05,
+      "loss": 0.7622,
+      "step": 2466
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0308601024803089e-05,
+      "loss": 0.7612,
+      "step": 2467
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0301966483706343e-05,
+      "loss": 0.8252,
+      "step": 2468
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0295331809568612e-05,
+      "loss": 0.8013,
+      "step": 2469
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.028869700531301e-05,
+      "loss": 0.8765,
+      "step": 2470
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.028206207386271e-05,
+      "loss": 0.8535,
+      "step": 2471
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0275427018140946e-05,
+      "loss": 0.7583,
+      "step": 2472
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0268791841071004e-05,
+      "loss": 0.8345,
+      "step": 2473
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0262156545576225e-05,
+      "loss": 0.835,
+      "step": 2474
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0255521134579997e-05,
+      "loss": 0.77,
+      "step": 2475
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0248885611005768e-05,
+      "loss": 0.8013,
+      "step": 2476
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0242249977777022e-05,
+      "loss": 0.7661,
+      "step": 2477
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0235614237817304e-05,
+      "loss": 0.7876,
+      "step": 2478
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0228978394050201e-05,
+      "loss": 0.8247,
+      "step": 2479
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0222342449399348e-05,
+      "loss": 0.8052,
+      "step": 2480
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0215706406788417e-05,
+      "loss": 0.8135,
+      "step": 2481
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0209070269141138e-05,
+      "loss": 0.7017,
+      "step": 2482
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0202434039381264e-05,
+      "loss": 0.793,
+      "step": 2483
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0195797720432601e-05,
+      "loss": 0.7769,
+      "step": 2484
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0189161315218994e-05,
+      "loss": 0.8413,
+      "step": 2485
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0182524826664322e-05,
+      "loss": 0.8218,
+      "step": 2486
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0175888257692506e-05,
+      "loss": 0.8408,
+      "step": 2487
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.016925161122749e-05,
+      "loss": 0.8569,
+      "step": 2488
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0162614890193267e-05,
+      "loss": 0.8511,
+      "step": 2489
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0155978097513854e-05,
+      "loss": 0.7651,
+      "step": 2490
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0149341236113306e-05,
+      "loss": 0.7305,
+      "step": 2491
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0142704308915702e-05,
+      "loss": 0.7749,
+      "step": 2492
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.013606731884515e-05,
+      "loss": 0.8208,
+      "step": 2493
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0129430268825795e-05,
+      "loss": 0.8203,
+      "step": 2494
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0122793161781792e-05,
+      "loss": 0.8237,
+      "step": 2495
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0116156000637338e-05,
+      "loss": 0.8569,
+      "step": 2496
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0109518788316648e-05,
+      "loss": 0.7881,
+      "step": 2497
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0102881527743948e-05,
+      "loss": 0.7725,
+      "step": 2498
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0096244221843508e-05,
+      "loss": 0.833,
+      "step": 2499
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0089606873539592e-05,
+      "loss": 0.8491,
+      "step": 2500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0082969485756509e-05,
+      "loss": 0.8306,
+      "step": 2501
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0076332061418565e-05,
+      "loss": 0.7935,
+      "step": 2502
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0069694603450092e-05,
+      "loss": 0.8057,
+      "step": 2503
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0063057114775432e-05,
+      "loss": 0.835,
+      "step": 2504
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.005641959831894e-05,
+      "loss": 0.8677,
+      "step": 2505
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0049782057004998e-05,
+      "loss": 0.8447,
+      "step": 2506
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0043144493757974e-05,
+      "loss": 0.7568,
+      "step": 2507
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0036506911502266e-05,
+      "loss": 0.8306,
+      "step": 2508
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.002986931316227e-05,
+      "loss": 0.7759,
+      "step": 2509
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0023231701662396e-05,
+      "loss": 0.7559,
+      "step": 2510
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.001659407992705e-05,
+      "loss": 0.8325,
+      "step": 2511
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0009956450880654e-05,
+      "loss": 0.7925,
+      "step": 2512
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0003318817447622e-05,
+      "loss": 0.7808,
+      "step": 2513
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.99668118255238e-06,
+      "loss": 0.793,
+      "step": 2514
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.99004354911935e-06,
+      "loss": 0.8364,
+      "step": 2515
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.983405920072954e-06,
+      "loss": 0.8491,
+      "step": 2516
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.97676829833761e-06,
+      "loss": 0.8208,
+      "step": 2517
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.97013068683773e-06,
+      "loss": 0.7861,
+      "step": 2518
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.96349308849774e-06,
+      "loss": 0.8428,
+      "step": 2519
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.956855506242027e-06,
+      "loss": 0.8691,
+      "step": 2520
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.950217942995005e-06,
+      "loss": 0.792,
+      "step": 2521
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.943580401681058e-06,
+      "loss": 0.7627,
+      "step": 2522
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.936942885224573e-06,
+      "loss": 0.8643,
+      "step": 2523
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.930305396549913e-06,
+      "loss": 0.8354,
+      "step": 2524
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.923667938581436e-06,
+      "loss": 0.7969,
+      "step": 2525
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.917030514243494e-06,
+      "loss": 0.8423,
+      "step": 2526
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.91039312646041e-06,
+      "loss": 0.855,
+      "step": 2527
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.903755778156496e-06,
+      "loss": 0.8721,
+      "step": 2528
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.897118472256054e-06,
+      "loss": 0.8042,
+      "step": 2529
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.890481211683358e-06,
+      "loss": 0.7461,
+      "step": 2530
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.883843999362664e-06,
+      "loss": 0.876,
+      "step": 2531
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.87720683821821e-06,
+      "loss": 0.7866,
+      "step": 2532
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.87056973117421e-06,
+      "loss": 0.7974,
+      "step": 2533
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.863932681154853e-06,
+      "loss": 0.8291,
+      "step": 2534
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.8572956910843e-06,
+      "loss": 0.8188,
+      "step": 2535
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.850658763886694e-06,
+      "loss": 0.8613,
+      "step": 2536
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.844021902486148e-06,
+      "loss": 0.7876,
+      "step": 2537
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.837385109806736e-06,
+      "loss": 0.8027,
+      "step": 2538
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.830748388772512e-06,
+      "loss": 0.8408,
+      "step": 2539
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.824111742307501e-06,
+      "loss": 0.8511,
+      "step": 2540
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.817475173335681e-06,
+      "loss": 0.8105,
+      "step": 2541
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.81083868478101e-06,
+      "loss": 0.7915,
+      "step": 2542
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.804202279567399e-06,
+      "loss": 0.8579,
+      "step": 2543
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.797565960618743e-06,
+      "loss": 0.8276,
+      "step": 2544
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.790929730858866e-06,
+      "loss": 0.8091,
+      "step": 2545
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.784293593211584e-06,
+      "loss": 0.7832,
+      "step": 2546
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.777657550600657e-06,
+      "loss": 0.8301,
+      "step": 2547
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.771021605949802e-06,
+      "loss": 0.8359,
+      "step": 2548
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.764385762182699e-06,
+      "loss": 0.8252,
+      "step": 2549
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.75775002222298e-06,
+      "loss": 0.7681,
+      "step": 2550
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.751114388994239e-06,
+      "loss": 0.7607,
+      "step": 2551
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.744478865420006e-06,
+      "loss": 0.8398,
+      "step": 2552
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.737843454423775e-06,
+      "loss": 0.8145,
+      "step": 2553
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.731208158929e-06,
+      "loss": 0.8525,
+      "step": 2554
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.724572981859056e-06,
+      "loss": 0.792,
+      "step": 2555
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.717937926137293e-06,
+      "loss": 0.877,
+      "step": 2556
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.711302994686994e-06,
+      "loss": 0.8057,
+      "step": 2557
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.704668190431392e-06,
+      "loss": 0.8184,
+      "step": 2558
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.698033516293659e-06,
+      "loss": 0.8828,
+      "step": 2559
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.691398975196913e-06,
+      "loss": 0.791,
+      "step": 2560
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.68476457006422e-06,
+      "loss": 0.8174,
+      "step": 2561
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.678130303818565e-06,
+      "loss": 0.8154,
+      "step": 2562
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.6714961793829e-06,
+      "loss": 0.8252,
+      "step": 2563
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.664862199680095e-06,
+      "loss": 0.8228,
+      "step": 2564
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.658228367632963e-06,
+      "loss": 0.8467,
+      "step": 2565
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.65159468616425e-06,
+      "loss": 0.8359,
+      "step": 2566
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.644961158196635e-06,
+      "loss": 0.8271,
+      "step": 2567
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.638327786652737e-06,
+      "loss": 0.813,
+      "step": 2568
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.63169457445509e-06,
+      "loss": 0.7837,
+      "step": 2569
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.625061524526171e-06,
+      "loss": 0.8311,
+      "step": 2570
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.618428639788387e-06,
+      "loss": 0.8296,
+      "step": 2571
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.611795923164068e-06,
+      "loss": 0.834,
+      "step": 2572
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.605163377575463e-06,
+      "loss": 0.7847,
+      "step": 2573
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.598531005944755e-06,
+      "loss": 0.835,
+      "step": 2574
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.591898811194048e-06,
+      "loss": 0.875,
+      "step": 2575
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.585266796245367e-06,
+      "loss": 0.8032,
+      "step": 2576
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.578634964020655e-06,
+      "loss": 0.7622,
+      "step": 2577
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.572003317441781e-06,
+      "loss": 0.7856,
+      "step": 2578
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.565371859430534e-06,
+      "loss": 0.7959,
+      "step": 2579
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.558740592908602e-06,
+      "loss": 0.7803,
+      "step": 2580
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.552109520797614e-06,
+      "loss": 0.8359,
+      "step": 2581
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.545478646019095e-06,
+      "loss": 0.7656,
+      "step": 2582
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.538847971494489e-06,
+      "loss": 0.854,
+      "step": 2583
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.53221750014515e-06,
+      "loss": 0.8594,
+      "step": 2584
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.525587234892347e-06,
+      "loss": 0.7871,
+      "step": 2585
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.51895717865726e-06,
+      "loss": 0.7749,
+      "step": 2586
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.51232733436096e-06,
+      "loss": 0.8687,
+      "step": 2587
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.505697704924447e-06,
+      "loss": 0.8262,
+      "step": 2588
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.499068293268615e-06,
+      "loss": 0.7954,
+      "step": 2589
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.492439102314257e-06,
+      "loss": 0.813,
+      "step": 2590
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.48581013498208e-06,
+      "loss": 0.7856,
+      "step": 2591
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.479181394192688e-06,
+      "loss": 0.8374,
+      "step": 2592
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.472552882866585e-06,
+      "loss": 0.8242,
+      "step": 2593
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.46592460392417e-06,
+      "loss": 0.7583,
+      "step": 2594
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.459296560285742e-06,
+      "loss": 0.8154,
+      "step": 2595
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.452668754871508e-06,
+      "loss": 0.8418,
+      "step": 2596
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.446041190601545e-06,
+      "loss": 0.8174,
+      "step": 2597
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.43941387039585e-06,
+      "loss": 0.8335,
+      "step": 2598
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.432786797174293e-06,
+      "loss": 0.8291,
+      "step": 2599
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.42615997385665e-06,
+      "loss": 0.7886,
+      "step": 2600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.419533403362575e-06,
+      "loss": 0.8164,
+      "step": 2601
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.412907088611617e-06,
+      "loss": 0.8208,
+      "step": 2602
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.40628103252321e-06,
+      "loss": 0.7954,
+      "step": 2603
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.399655238016674e-06,
+      "loss": 0.8491,
+      "step": 2604
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.393029708011214e-06,
+      "loss": 0.8008,
+      "step": 2605
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.386404445425924e-06,
+      "loss": 0.8354,
+      "step": 2606
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.379779453179773e-06,
+      "loss": 0.7734,
+      "step": 2607
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.373154734191612e-06,
+      "loss": 0.811,
+      "step": 2608
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.366530291380172e-06,
+      "loss": 0.8296,
+      "step": 2609
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.359906127664067e-06,
+      "loss": 0.813,
+      "step": 2610
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.35328224596178e-06,
+      "loss": 0.7871,
+      "step": 2611
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.34665864919167e-06,
+      "loss": 0.8003,
+      "step": 2612
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.340035340271984e-06,
+      "loss": 0.7456,
+      "step": 2613
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.33341232212083e-06,
+      "loss": 0.8203,
+      "step": 2614
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.326789597656185e-06,
+      "loss": 0.813,
+      "step": 2615
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.320167169795905e-06,
+      "loss": 0.8223,
+      "step": 2616
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.313545041457716e-06,
+      "loss": 0.8071,
+      "step": 2617
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.306923215559202e-06,
+      "loss": 0.7905,
+      "step": 2618
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.300301695017824e-06,
+      "loss": 0.8359,
+      "step": 2619
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.293680482750901e-06,
+      "loss": 0.811,
+      "step": 2620
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.287059581675628e-06,
+      "loss": 0.7852,
+      "step": 2621
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.280438994709043e-06,
+      "loss": 0.7822,
+      "step": 2622
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.273818724768066e-06,
+      "loss": 0.7959,
+      "step": 2623
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.267198774769468e-06,
+      "loss": 0.8179,
+      "step": 2624
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.260579147629875e-06,
+      "loss": 0.8491,
+      "step": 2625
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.253959846265779e-06,
+      "loss": 0.8354,
+      "step": 2626
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.247340873593522e-06,
+      "loss": 0.7729,
+      "step": 2627
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.240722232529307e-06,
+      "loss": 0.7783,
+      "step": 2628
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.234103925989186e-06,
+      "loss": 0.7954,
+      "step": 2629
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.227485956889063e-06,
+      "loss": 0.7563,
+      "step": 2630
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.220868328144708e-06,
+      "loss": 0.7637,
+      "step": 2631
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.21425104267171e-06,
+      "loss": 0.8267,
+      "step": 2632
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.207634103385541e-06,
+      "loss": 0.7974,
+      "step": 2633
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.201017513201498e-06,
+      "loss": 0.8281,
+      "step": 2634
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.194401275034734e-06,
+      "loss": 0.8374,
+      "step": 2635
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.187785391800244e-06,
+      "loss": 0.7959,
+      "step": 2636
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.18116986641286e-06,
+      "loss": 0.8423,
+      "step": 2637
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.174554701787279e-06,
+      "loss": 0.8022,
+      "step": 2638
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.167939900838007e-06,
+      "loss": 0.7866,
+      "step": 2639
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.161325466479414e-06,
+      "loss": 0.8213,
+      "step": 2640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.1547114016257e-06,
+      "loss": 0.7839,
+      "step": 2641
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.148097709190903e-06,
+      "loss": 0.7959,
+      "step": 2642
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.141484392088893e-06,
+      "loss": 0.8008,
+      "step": 2643
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.134871453233383e-06,
+      "loss": 0.791,
+      "step": 2644
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.128258895537913e-06,
+      "loss": 0.8584,
+      "step": 2645
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.121646721915854e-06,
+      "loss": 0.8105,
+      "step": 2646
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.11503493528041e-06,
+      "loss": 0.8286,
+      "step": 2647
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.108423538544622e-06,
+      "loss": 0.7617,
+      "step": 2648
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.101812534621348e-06,
+      "loss": 0.813,
+      "step": 2649
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.095201926423275e-06,
+      "loss": 0.8081,
+      "step": 2650
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.088591716862921e-06,
+      "loss": 0.8262,
+      "step": 2651
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.081981908852626e-06,
+      "loss": 0.8022,
+      "step": 2652
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.075372505304547e-06,
+      "loss": 0.7861,
+      "step": 2653
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.06876350913067e-06,
+      "loss": 0.8281,
+      "step": 2654
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.062154923242799e-06,
+      "loss": 0.8047,
+      "step": 2655
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.055546750552568e-06,
+      "loss": 0.8071,
+      "step": 2656
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.048938993971403e-06,
+      "loss": 0.7949,
+      "step": 2657
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.04233165641057e-06,
+      "loss": 0.8228,
+      "step": 2658
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.035724740781146e-06,
+      "loss": 0.853,
+      "step": 2659
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.029118249994013e-06,
+      "loss": 0.8975,
+      "step": 2660
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.022512186959875e-06,
+      "loss": 0.8301,
+      "step": 2661
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.01590655458924e-06,
+      "loss": 0.8789,
+      "step": 2662
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.009301355792442e-06,
+      "loss": 0.7896,
+      "step": 2663
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.002696593479594e-06,
+      "loss": 0.7998,
+      "step": 2664
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.996092270560653e-06,
+      "loss": 0.8062,
+      "step": 2665
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.989488389945356e-06,
+      "loss": 0.7832,
+      "step": 2666
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.982884954543252e-06,
+      "loss": 0.8208,
+      "step": 2667
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.976281967263701e-06,
+      "loss": 0.7886,
+      "step": 2668
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.969679431015857e-06,
+      "loss": 0.7651,
+      "step": 2669
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.96307734870868e-06,
+      "loss": 0.8223,
+      "step": 2670
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.956475723250927e-06,
+      "loss": 0.8477,
+      "step": 2671
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.949874557551151e-06,
+      "loss": 0.8525,
+      "step": 2672
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.94327385451772e-06,
+      "loss": 0.7817,
+      "step": 2673
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.936673617058767e-06,
+      "loss": 0.7974,
+      "step": 2674
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.93007384808225e-06,
+      "loss": 0.7827,
+      "step": 2675
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.923474550495907e-06,
+      "loss": 0.7812,
+      "step": 2676
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.916875727207266e-06,
+      "loss": 0.8232,
+      "step": 2677
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.91027738112365e-06,
+      "loss": 0.7881,
+      "step": 2678
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.903679515152174e-06,
+      "loss": 0.7773,
+      "step": 2679
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.897082132199738e-06,
+      "loss": 0.8105,
+      "step": 2680
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.890485235173026e-06,
+      "loss": 0.8403,
+      "step": 2681
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.883888826978512e-06,
+      "loss": 0.7603,
+      "step": 2682
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.877292910522461e-06,
+      "loss": 0.811,
+      "step": 2683
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.870697488710914e-06,
+      "loss": 0.8115,
+      "step": 2684
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.864102564449689e-06,
+      "loss": 0.8252,
+      "step": 2685
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.857508140644394e-06,
+      "loss": 0.8174,
+      "step": 2686
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.850914220200416e-06,
+      "loss": 0.8101,
+      "step": 2687
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.844320806022911e-06,
+      "loss": 0.7783,
+      "step": 2688
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.83772790101682e-06,
+      "loss": 0.7988,
+      "step": 2689
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.831135508086864e-06,
+      "loss": 0.8467,
+      "step": 2690
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.82454363013753e-06,
+      "loss": 0.7529,
+      "step": 2691
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.817952270073075e-06,
+      "loss": 0.793,
+      "step": 2692
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.81136143079754e-06,
+      "loss": 0.8008,
+      "step": 2693
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.804771115214729e-06,
+      "loss": 0.8262,
+      "step": 2694
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.798181326228212e-06,
+      "loss": 0.8218,
+      "step": 2695
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.791592066741335e-06,
+      "loss": 0.8481,
+      "step": 2696
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.785003339657203e-06,
+      "loss": 0.7786,
+      "step": 2697
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.778415147878698e-06,
+      "loss": 0.7949,
+      "step": 2698
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.771827494308448e-06,
+      "loss": 0.8618,
+      "step": 2699
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.765240381848864e-06,
+      "loss": 0.8398,
+      "step": 2700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.758653813402107e-06,
+      "loss": 0.854,
+      "step": 2701
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.752067791870094e-06,
+      "loss": 0.7734,
+      "step": 2702
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.745482320154514e-06,
+      "loss": 0.7974,
+      "step": 2703
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.738897401156803e-06,
+      "loss": 0.8047,
+      "step": 2704
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.732313037778162e-06,
+      "loss": 0.8325,
+      "step": 2705
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.72572923291954e-06,
+      "loss": 0.832,
+      "step": 2706
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.719145989481638e-06,
+      "loss": 0.8179,
+      "step": 2707
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.712563310364928e-06,
+      "loss": 0.7681,
+      "step": 2708
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.705981198469603e-06,
+      "loss": 0.8076,
+      "step": 2709
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.699399656695632e-06,
+      "loss": 0.77,
+      "step": 2710
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.692818687942726e-06,
+      "loss": 0.8047,
+      "step": 2711
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.686238295110337e-06,
+      "loss": 0.7583,
+      "step": 2712
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.679658481097668e-06,
+      "loss": 0.874,
+      "step": 2713
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.673079248803662e-06,
+      "loss": 0.8101,
+      "step": 2714
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.666500601127022e-06,
+      "loss": 0.7788,
+      "step": 2715
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.659922540966167e-06,
+      "loss": 0.7661,
+      "step": 2716
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.653345071219283e-06,
+      "loss": 0.8223,
+      "step": 2717
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.646768194784278e-06,
+      "loss": 0.8486,
+      "step": 2718
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.640191914558811e-06,
+      "loss": 0.8325,
+      "step": 2719
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.633616233440268e-06,
+      "loss": 0.8267,
+      "step": 2720
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.627041154325775e-06,
+      "loss": 0.8223,
+      "step": 2721
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.6204666801122e-06,
+      "loss": 0.833,
+      "step": 2722
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.613892813696128e-06,
+      "loss": 0.7715,
+      "step": 2723
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.607319557973889e-06,
+      "loss": 0.7432,
+      "step": 2724
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.600746915841545e-06,
+      "loss": 0.77,
+      "step": 2725
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.594174890194884e-06,
+      "loss": 0.832,
+      "step": 2726
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.587603483929417e-06,
+      "loss": 0.7744,
+      "step": 2727
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.58103269994039e-06,
+      "loss": 0.7998,
+      "step": 2728
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.574462541122772e-06,
+      "loss": 0.8286,
+      "step": 2729
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.567893010371255e-06,
+      "loss": 0.8208,
+      "step": 2730
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.561324110580257e-06,
+      "loss": 0.8208,
+      "step": 2731
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.554755844643915e-06,
+      "loss": 0.8169,
+      "step": 2732
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.548188215456094e-06,
+      "loss": 0.8506,
+      "step": 2733
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.541621225910366e-06,
+      "loss": 0.7881,
+      "step": 2734
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.53505487890003e-06,
+      "loss": 0.7896,
+      "step": 2735
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.528489177318107e-06,
+      "loss": 0.8369,
+      "step": 2736
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.521924124057317e-06,
+      "loss": 0.7974,
+      "step": 2737
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.515359722010108e-06,
+      "loss": 0.7612,
+      "step": 2738
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.508795974068637e-06,
+      "loss": 0.8022,
+      "step": 2739
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.50223288312478e-06,
+      "loss": 0.8193,
+      "step": 2740
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.4956704520701e-06,
+      "loss": 0.7812,
+      "step": 2741
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.489108683795902e-06,
+      "loss": 0.71,
+      "step": 2742
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.482547581193177e-06,
+      "loss": 0.7773,
+      "step": 2743
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.475987147152626e-06,
+      "loss": 0.8384,
+      "step": 2744
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.46942738456466e-06,
+      "loss": 0.8193,
+      "step": 2745
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.462868296319392e-06,
+      "loss": 0.7783,
+      "step": 2746
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.456309885306641e-06,
+      "loss": 0.8115,
+      "step": 2747
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.44975215441592e-06,
+      "loss": 0.8364,
+      "step": 2748
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.443195106536445e-06,
+      "loss": 0.8003,
+      "step": 2749
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.436638744557143e-06,
+      "loss": 0.8809,
+      "step": 2750
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.43008307136662e-06,
+      "loss": 0.7876,
+      "step": 2751
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.42352808985319e-06,
+      "loss": 0.7842,
+      "step": 2752
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.41697380290486e-06,
+      "loss": 0.8364,
+      "step": 2753
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.410420213409332e-06,
+      "loss": 0.7266,
+      "step": 2754
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.403867324253996e-06,
+      "loss": 0.8267,
+      "step": 2755
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.39731513832594e-06,
+      "loss": 0.7432,
+      "step": 2756
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.390763658511938e-06,
+      "loss": 0.8081,
+      "step": 2757
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.384212887698453e-06,
+      "loss": 0.7881,
+      "step": 2758
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.377662828771634e-06,
+      "loss": 0.7617,
+      "step": 2759
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.371113484617323e-06,
+      "loss": 0.811,
+      "step": 2760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.364564858121047e-06,
+      "loss": 0.7783,
+      "step": 2761
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.358016952168004e-06,
+      "loss": 0.856,
+      "step": 2762
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.351469769643087e-06,
+      "loss": 0.7837,
+      "step": 2763
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.344923313430868e-06,
+      "loss": 0.8066,
+      "step": 2764
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.338377586415595e-06,
+      "loss": 0.8877,
+      "step": 2765
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.331832591481194e-06,
+      "loss": 0.8101,
+      "step": 2766
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.32528833151128e-06,
+      "loss": 0.7954,
+      "step": 2767
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.318744809389135e-06,
+      "loss": 0.8154,
+      "step": 2768
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.312202027997712e-06,
+      "loss": 0.8599,
+      "step": 2769
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.305659990219644e-06,
+      "loss": 0.8545,
+      "step": 2770
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.299118698937237e-06,
+      "loss": 0.8398,
+      "step": 2771
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.292578157032462e-06,
+      "loss": 0.7808,
+      "step": 2772
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.286038367386967e-06,
+      "loss": 0.8071,
+      "step": 2773
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.279499332882063e-06,
+      "loss": 0.8008,
+      "step": 2774
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.272961056398739e-06,
+      "loss": 0.833,
+      "step": 2775
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.266423540817627e-06,
+      "loss": 0.834,
+      "step": 2776
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.25988678901905e-06,
+      "loss": 0.8804,
+      "step": 2777
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.25335080388298e-06,
+      "loss": 0.8174,
+      "step": 2778
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.246815588289052e-06,
+      "loss": 0.8252,
+      "step": 2779
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.240281145116567e-06,
+      "loss": 0.8491,
+      "step": 2780
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.23374747724448e-06,
+      "loss": 0.8062,
+      "step": 2781
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.227214587551414e-06,
+      "loss": 0.8237,
+      "step": 2782
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.220682478915635e-06,
+      "loss": 0.7837,
+      "step": 2783
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.214151154215072e-06,
+      "loss": 0.7866,
+      "step": 2784
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.207620616327316e-06,
+      "loss": 0.8154,
+      "step": 2785
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.201090868129595e-06,
+      "loss": 0.8296,
+      "step": 2786
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.194561912498807e-06,
+      "loss": 0.79,
+      "step": 2787
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.188033752311486e-06,
+      "loss": 0.8022,
+      "step": 2788
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.181506390443828e-06,
+      "loss": 0.7803,
+      "step": 2789
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.174979829771662e-06,
+      "loss": 0.7627,
+      "step": 2790
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.168454073170478e-06,
+      "loss": 0.7632,
+      "step": 2791
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.161929123515413e-06,
+      "loss": 0.8237,
+      "step": 2792
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.15540498368123e-06,
+      "loss": 0.7871,
+      "step": 2793
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.148881656542353e-06,
+      "loss": 0.8022,
+      "step": 2794
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.142359144972845e-06,
+      "loss": 0.7881,
+      "step": 2795
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.135837451846404e-06,
+      "loss": 0.8413,
+      "step": 2796
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.12931658003637e-06,
+      "loss": 0.8306,
+      "step": 2797
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.122796532415724e-06,
+      "loss": 0.7866,
+      "step": 2798
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.11627731185708e-06,
+      "loss": 0.7842,
+      "step": 2799
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.109758921232688e-06,
+      "loss": 0.8242,
+      "step": 2800
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.103241363414432e-06,
+      "loss": 0.8291,
+      "step": 2801
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.096724641273834e-06,
+      "loss": 0.7998,
+      "step": 2802
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.090208757682047e-06,
+      "loss": 0.8057,
+      "step": 2803
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.083693715509843e-06,
+      "loss": 0.8354,
+      "step": 2804
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.077179517627638e-06,
+      "loss": 0.8262,
+      "step": 2805
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.070666166905473e-06,
+      "loss": 0.8237,
+      "step": 2806
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.064153666213007e-06,
+      "loss": 0.7441,
+      "step": 2807
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.05764201841953e-06,
+      "loss": 0.8398,
+      "step": 2808
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.051131226393958e-06,
+      "loss": 0.77,
+      "step": 2809
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.044621293004836e-06,
+      "loss": 0.7769,
+      "step": 2810
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.03811222112031e-06,
+      "loss": 0.8057,
+      "step": 2811
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.03160401360817e-06,
+      "loss": 0.7793,
+      "step": 2812
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.025096673335812e-06,
+      "loss": 0.793,
+      "step": 2813
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.018590203170252e-06,
+      "loss": 0.8159,
+      "step": 2814
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.012084605978121e-06,
+      "loss": 0.8164,
+      "step": 2815
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.00557988462567e-06,
+      "loss": 0.833,
+      "step": 2816
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.999076041978769e-06,
+      "loss": 0.8584,
+      "step": 2817
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.992573080902881e-06,
+      "loss": 0.7539,
+      "step": 2818
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.986071004263104e-06,
+      "loss": 0.8022,
+      "step": 2819
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.979569814924131e-06,
+      "loss": 0.8364,
+      "step": 2820
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.973069515750266e-06,
+      "loss": 0.7852,
+      "step": 2821
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.966570109605429e-06,
+      "loss": 0.835,
+      "step": 2822
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.960071599353137e-06,
+      "loss": 0.8218,
+      "step": 2823
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.953573987856522e-06,
+      "loss": 0.8599,
+      "step": 2824
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.94707727797831e-06,
+      "loss": 0.7729,
+      "step": 2825
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.940581472580831e-06,
+      "loss": 0.7561,
+      "step": 2826
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.93408657452603e-06,
+      "loss": 0.8413,
+      "step": 2827
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.927592586675431e-06,
+      "loss": 0.8271,
+      "step": 2828
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.921099511890174e-06,
+      "loss": 0.7798,
+      "step": 2829
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.914607353030989e-06,
+      "loss": 0.7456,
+      "step": 2830
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.908116112958208e-06,
+      "loss": 0.8384,
+      "step": 2831
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.90162579453175e-06,
+      "loss": 0.7793,
+      "step": 2832
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.895136400611132e-06,
+      "loss": 0.8384,
+      "step": 2833
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.888647934055467e-06,
+      "loss": 0.814,
+      "step": 2834
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.882160397723452e-06,
+      "loss": 0.8174,
+      "step": 2835
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.875673794473382e-06,
+      "loss": 0.8271,
+      "step": 2836
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.869188127163135e-06,
+      "loss": 0.8364,
+      "step": 2837
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.862703398650185e-06,
+      "loss": 0.7798,
+      "step": 2838
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.85621961179158e-06,
+      "loss": 0.7798,
+      "step": 2839
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.84973676944396e-06,
+      "loss": 0.7549,
+      "step": 2840
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.843254874463554e-06,
+      "loss": 0.7759,
+      "step": 2841
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.836773929706162e-06,
+      "loss": 0.8184,
+      "step": 2842
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.830293938027174e-06,
+      "loss": 0.8154,
+      "step": 2843
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.823814902281553e-06,
+      "loss": 0.8047,
+      "step": 2844
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.817336825323858e-06,
+      "loss": 0.7568,
+      "step": 2845
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.810859710008197e-06,
+      "loss": 0.7202,
+      "step": 2846
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.80438355918828e-06,
+      "loss": 0.7837,
+      "step": 2847
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.797908375717382e-06,
+      "loss": 0.7949,
+      "step": 2848
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.791434162448349e-06,
+      "loss": 0.8179,
+      "step": 2849
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.784960922233603e-06,
+      "loss": 0.8252,
+      "step": 2850
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.778488657925134e-06,
+      "loss": 0.8174,
+      "step": 2851
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.772017372374516e-06,
+      "loss": 0.8174,
+      "step": 2852
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.765547068432867e-06,
+      "loss": 0.7998,
+      "step": 2853
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.759077748950895e-06,
+      "loss": 0.7681,
+      "step": 2854
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.752609416778864e-06,
+      "loss": 0.7515,
+      "step": 2855
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.746142074766602e-06,
+      "loss": 0.8145,
+      "step": 2856
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.739675725763504e-06,
+      "loss": 0.8096,
+      "step": 2857
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.73321037261853e-06,
+      "loss": 0.8398,
+      "step": 2858
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.726746018180196e-06,
+      "loss": 0.8359,
+      "step": 2859
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.720282665296577e-06,
+      "loss": 0.7822,
+      "step": 2860
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.713820316815312e-06,
+      "loss": 0.7498,
+      "step": 2861
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.7073589755836e-06,
+      "loss": 0.8027,
+      "step": 2862
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.700898644448183e-06,
+      "loss": 0.8232,
+      "step": 2863
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.694439326255372e-06,
+      "loss": 0.7388,
+      "step": 2864
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.687981023851023e-06,
+      "loss": 0.7812,
+      "step": 2865
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.681523740080554e-06,
+      "loss": 0.8198,
+      "step": 2866
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.675067477788919e-06,
+      "loss": 0.771,
+      "step": 2867
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.668612239820636e-06,
+      "loss": 0.834,
+      "step": 2868
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.662158029019767e-06,
+      "loss": 0.8267,
+      "step": 2869
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.655704848229916e-06,
+      "loss": 0.7393,
+      "step": 2870
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.64925270029424e-06,
+      "loss": 0.7808,
+      "step": 2871
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.642801588055441e-06,
+      "loss": 0.7856,
+      "step": 2872
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.636351514355766e-06,
+      "loss": 0.7969,
+      "step": 2873
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.629902482036996e-06,
+      "loss": 0.8164,
+      "step": 2874
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.623454493940457e-06,
+      "loss": 0.8457,
+      "step": 2875
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.617007552907023e-06,
+      "loss": 0.7642,
+      "step": 2876
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.610561661777094e-06,
+      "loss": 0.832,
+      "step": 2877
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.604116823390611e-06,
+      "loss": 0.8071,
+      "step": 2878
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.597673040587061e-06,
+      "loss": 0.7583,
+      "step": 2879
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.591230316205459e-06,
+      "loss": 0.7695,
+      "step": 2880
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.584788653084346e-06,
+      "loss": 0.7583,
+      "step": 2881
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.578348054061807e-06,
+      "loss": 0.8491,
+      "step": 2882
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.571908521975455e-06,
+      "loss": 0.8208,
+      "step": 2883
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.565470059662428e-06,
+      "loss": 0.8076,
+      "step": 2884
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.559032669959398e-06,
+      "loss": 0.7954,
+      "step": 2885
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5525963557025595e-06,
+      "loss": 0.8408,
+      "step": 2886
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.546161119727648e-06,
+      "loss": 0.7642,
+      "step": 2887
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.539726964869898e-06,
+      "loss": 0.7871,
+      "step": 2888
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.53329389396409e-06,
+      "loss": 0.7871,
+      "step": 2889
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.52686190984452e-06,
+      "loss": 0.6997,
+      "step": 2890
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.520431015345e-06,
+      "loss": 0.7896,
+      "step": 2891
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.514001213298868e-06,
+      "loss": 0.7158,
+      "step": 2892
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.50757250653898e-06,
+      "loss": 0.8286,
+      "step": 2893
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.501144897897707e-06,
+      "loss": 0.8247,
+      "step": 2894
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.494718390206937e-06,
+      "loss": 0.7905,
+      "step": 2895
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.488292986298072e-06,
+      "loss": 0.7905,
+      "step": 2896
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4818686890020385e-06,
+      "loss": 0.8257,
+      "step": 2897
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.47544550114925e-06,
+      "loss": 0.8374,
+      "step": 2898
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.469023425569661e-06,
+      "loss": 0.8232,
+      "step": 2899
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.462602465092716e-06,
+      "loss": 0.8447,
+      "step": 2900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.456182622547377e-06,
+      "loss": 0.7529,
+      "step": 2901
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.449763900762108e-06,
+      "loss": 0.7373,
+      "step": 2902
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.443346302564881e-06,
+      "loss": 0.812,
+      "step": 2903
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4369298307831835e-06,
+      "loss": 0.8037,
+      "step": 2904
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.430514488243985e-06,
+      "loss": 0.8325,
+      "step": 2905
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4241002777737754e-06,
+      "loss": 0.7744,
+      "step": 2906
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.417687202198542e-06,
+      "loss": 0.7842,
+      "step": 2907
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.411275264343769e-06,
+      "loss": 0.8115,
+      "step": 2908
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.404864467034439e-06,
+      "loss": 0.8506,
+      "step": 2909
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.398454813095033e-06,
+      "loss": 0.7769,
+      "step": 2910
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.392046305349534e-06,
+      "loss": 0.7446,
+      "step": 2911
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.385638946621409e-06,
+      "loss": 0.8271,
+      "step": 2912
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.379232739733625e-06,
+      "loss": 0.8291,
+      "step": 2913
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.372827687508646e-06,
+      "loss": 0.7378,
+      "step": 2914
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.366423792768423e-06,
+      "loss": 0.7578,
+      "step": 2915
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.360021058334389e-06,
+      "loss": 0.8071,
+      "step": 2916
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.353619487027479e-06,
+      "loss": 0.752,
+      "step": 2917
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3472190816681114e-06,
+      "loss": 0.8931,
+      "step": 2918
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.340819845076185e-06,
+      "loss": 0.7998,
+      "step": 2919
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.334421780071089e-06,
+      "loss": 0.7646,
+      "step": 2920
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.328024889471696e-06,
+      "loss": 0.7983,
+      "step": 2921
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3216291760963655e-06,
+      "loss": 0.751,
+      "step": 2922
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3152346427629225e-06,
+      "loss": 0.7961,
+      "step": 2923
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.308841292288693e-06,
+      "loss": 0.7739,
+      "step": 2924
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.302449127490471e-06,
+      "loss": 0.8232,
+      "step": 2925
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.296058151184524e-06,
+      "loss": 0.7373,
+      "step": 2926
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.289668366186603e-06,
+      "loss": 0.8057,
+      "step": 2927
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.283279775311931e-06,
+      "loss": 0.7681,
+      "step": 2928
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2768923813752135e-06,
+      "loss": 0.7778,
+      "step": 2929
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.27050618719061e-06,
+      "loss": 0.8809,
+      "step": 2930
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.26412119557177e-06,
+      "loss": 0.7646,
+      "step": 2931
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2577374093318045e-06,
+      "loss": 0.8003,
+      "step": 2932
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.25135483128329e-06,
+      "loss": 0.8403,
+      "step": 2933
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.244973464238281e-06,
+      "loss": 0.7646,
+      "step": 2934
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.23859331100829e-06,
+      "loss": 0.814,
+      "step": 2935
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2322143744043005e-06,
+      "loss": 0.7734,
+      "step": 2936
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.225836657236752e-06,
+      "loss": 0.7554,
+      "step": 2937
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.219460162315552e-06,
+      "loss": 0.853,
+      "step": 2938
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.213084892450079e-06,
+      "loss": 0.8096,
+      "step": 2939
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.206710850449147e-06,
+      "loss": 0.7886,
+      "step": 2940
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.200338039121053e-06,
+      "loss": 0.8281,
+      "step": 2941
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.193966461273542e-06,
+      "loss": 0.7964,
+      "step": 2942
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.187596119713816e-06,
+      "loss": 0.7676,
+      "step": 2943
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.181227017248532e-06,
+      "loss": 0.772,
+      "step": 2944
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.174859156683801e-06,
+      "loss": 0.7476,
+      "step": 2945
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.16849254082519e-06,
+      "loss": 0.7529,
+      "step": 2946
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.162127172477713e-06,
+      "loss": 0.8267,
+      "step": 2947
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.155763054445833e-06,
+      "loss": 0.812,
+      "step": 2948
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.149400189533473e-06,
+      "loss": 0.8154,
+      "step": 2949
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.143038580543995e-06,
+      "loss": 0.7866,
+      "step": 2950
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.136678230280205e-06,
+      "loss": 0.8179,
+      "step": 2951
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.130319141544362e-06,
+      "loss": 0.8154,
+      "step": 2952
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.123961317138166e-06,
+      "loss": 0.8423,
+      "step": 2953
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1176047598627575e-06,
+      "loss": 0.7964,
+      "step": 2954
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.111249472518718e-06,
+      "loss": 0.7969,
+      "step": 2955
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.104895457906082e-06,
+      "loss": 0.8608,
+      "step": 2956
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.098542718824308e-06,
+      "loss": 0.7896,
+      "step": 2957
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.092191258072298e-06,
+      "loss": 0.8115,
+      "step": 2958
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.085841078448393e-06,
+      "loss": 0.7632,
+      "step": 2959
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.079492182750368e-06,
+      "loss": 0.7793,
+      "step": 2960
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.073144573775428e-06,
+      "loss": 0.7573,
+      "step": 2961
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.066798254320219e-06,
+      "loss": 0.8687,
+      "step": 2962
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0604532271808105e-06,
+      "loss": 0.855,
+      "step": 2963
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.054109495152717e-06,
+      "loss": 0.8306,
+      "step": 2964
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.047767061030858e-06,
+      "loss": 0.7554,
+      "step": 2965
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.041425927609607e-06,
+      "loss": 0.7495,
+      "step": 2966
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.03508609768275e-06,
+      "loss": 0.813,
+      "step": 2967
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0287475740434995e-06,
+      "loss": 0.8706,
+      "step": 2968
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.022410359484495e-06,
+      "loss": 0.7837,
+      "step": 2969
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.016074456797803e-06,
+      "loss": 0.7808,
+      "step": 2970
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.009739868774905e-06,
+      "loss": 0.75,
+      "step": 2971
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.003406598206706e-06,
+      "loss": 0.8037,
+      "step": 2972
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.997074647883527e-06,
+      "loss": 0.8237,
+      "step": 2973
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9907440205951235e-06,
+      "loss": 0.8071,
+      "step": 2974
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.984414719130641e-06,
+      "loss": 0.8208,
+      "step": 2975
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.978086746278665e-06,
+      "loss": 0.7632,
+      "step": 2976
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.971760104827184e-06,
+      "loss": 0.7349,
+      "step": 2977
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9654347975636015e-06,
+      "loss": 0.8232,
+      "step": 2978
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.959110827274733e-06,
+      "loss": 0.8174,
+      "step": 2979
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.952788196746806e-06,
+      "loss": 0.8169,
+      "step": 2980
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.946466908765464e-06,
+      "loss": 0.8159,
+      "step": 2981
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.94014696611574e-06,
+      "loss": 0.8032,
+      "step": 2982
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.933828371582096e-06,
+      "loss": 0.8452,
+      "step": 2983
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9275111279483885e-06,
+      "loss": 0.7773,
+      "step": 2984
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.921195237997884e-06,
+      "loss": 0.7998,
+      "step": 2985
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9148807045132435e-06,
+      "loss": 0.8198,
+      "step": 2986
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.90856753027654e-06,
+      "loss": 0.8491,
+      "step": 2987
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.902255718069247e-06,
+      "loss": 0.8306,
+      "step": 2988
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.89594527067223e-06,
+      "loss": 0.7861,
+      "step": 2989
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.889636190865759e-06,
+      "loss": 0.8252,
+      "step": 2990
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.883328481429505e-06,
+      "loss": 0.8027,
+      "step": 2991
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.87702214514253e-06,
+      "loss": 0.8057,
+      "step": 2992
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8707171847832884e-06,
+      "loss": 0.7949,
+      "step": 2993
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8644136031296335e-06,
+      "loss": 0.7954,
+      "step": 2994
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.858111402958815e-06,
+      "loss": 0.8418,
+      "step": 2995
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.851810587047461e-06,
+      "loss": 0.7842,
+      "step": 2996
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8455111581715995e-06,
+      "loss": 0.8145,
+      "step": 2997
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.839213119106644e-06,
+      "loss": 0.8042,
+      "step": 2998
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.832916472627407e-06,
+      "loss": 0.77,
+      "step": 2999
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.826621221508063e-06,
+      "loss": 0.8735,
+      "step": 3000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.820327368522196e-06,
+      "loss": 0.8413,
+      "step": 3001
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.814034916442762e-06,
+      "loss": 0.8003,
+      "step": 3002
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.807743868042098e-06,
+      "loss": 0.7886,
+      "step": 3003
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.801454226091932e-06,
+      "loss": 0.7979,
+      "step": 3004
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.79516599336336e-06,
+      "loss": 0.7905,
+      "step": 3005
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7888791726268744e-06,
+      "loss": 0.7598,
+      "step": 3006
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7825937666523235e-06,
+      "loss": 0.8169,
+      "step": 3007
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7763097782089515e-06,
+      "loss": 0.7236,
+      "step": 3008
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.77002721006537e-06,
+      "loss": 0.7969,
+      "step": 3009
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.763746064989561e-06,
+      "loss": 0.8223,
+      "step": 3010
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.757466345748887e-06,
+      "loss": 0.7861,
+      "step": 3011
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.751188055110078e-06,
+      "loss": 0.8066,
+      "step": 3012
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.744911195839237e-06,
+      "loss": 0.8325,
+      "step": 3013
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.738635770701834e-06,
+      "loss": 0.8003,
+      "step": 3014
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.732361782462704e-06,
+      "loss": 0.7842,
+      "step": 3015
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.726089233886064e-06,
+      "loss": 0.8271,
+      "step": 3016
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7198181277354734e-06,
+      "loss": 0.8628,
+      "step": 3017
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.713548466773876e-06,
+      "loss": 0.7881,
+      "step": 3018
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.707280253763568e-06,
+      "loss": 0.8076,
+      "step": 3019
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.701013491466216e-06,
+      "loss": 0.8232,
+      "step": 3020
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.694748182642836e-06,
+      "loss": 0.7778,
+      "step": 3021
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.688484330053815e-06,
+      "loss": 0.7871,
+      "step": 3022
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.682221936458893e-06,
+      "loss": 0.8252,
+      "step": 3023
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.675961004617163e-06,
+      "loss": 0.8354,
+      "step": 3024
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.669701537287081e-06,
+      "loss": 0.7812,
+      "step": 3025
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.663443537226458e-06,
+      "loss": 0.8013,
+      "step": 3026
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.657187007192457e-06,
+      "loss": 0.8232,
+      "step": 3027
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.650931949941588e-06,
+      "loss": 0.7654,
+      "step": 3028
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6446783682297174e-06,
+      "loss": 0.8062,
+      "step": 3029
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.638426264812064e-06,
+      "loss": 0.8154,
+      "step": 3030
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.632175642443187e-06,
+      "loss": 0.7686,
+      "step": 3031
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.625926503876998e-06,
+      "loss": 0.7373,
+      "step": 3032
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6196788518667595e-06,
+      "loss": 0.811,
+      "step": 3033
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6134326891650736e-06,
+      "loss": 0.7896,
+      "step": 3034
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.607188018523883e-06,
+      "loss": 0.7964,
+      "step": 3035
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6009448426944785e-06,
+      "loss": 0.8174,
+      "step": 3036
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.594703164427495e-06,
+      "loss": 0.8486,
+      "step": 3037
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.588462986472897e-06,
+      "loss": 0.7695,
+      "step": 3038
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5822243115799965e-06,
+      "loss": 0.7725,
+      "step": 3039
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.575987142497441e-06,
+      "loss": 0.7607,
+      "step": 3040
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.569751481973221e-06,
+      "loss": 0.7734,
+      "step": 3041
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.563517332754645e-06,
+      "loss": 0.7822,
+      "step": 3042
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.557284697588374e-06,
+      "loss": 0.8247,
+      "step": 3043
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.551053579220396e-06,
+      "loss": 0.8457,
+      "step": 3044
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.544823980396025e-06,
+      "loss": 0.752,
+      "step": 3045
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.53859590385991e-06,
+      "loss": 0.8286,
+      "step": 3046
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.532369352356032e-06,
+      "loss": 0.8335,
+      "step": 3047
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.526144328627697e-06,
+      "loss": 0.8306,
+      "step": 3048
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.519920835417535e-06,
+      "loss": 0.7681,
+      "step": 3049
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5136988754675065e-06,
+      "loss": 0.7754,
+      "step": 3050
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.507478451518901e-06,
+      "loss": 0.7263,
+      "step": 3051
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.501259566312312e-06,
+      "loss": 0.7749,
+      "step": 3052
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4950422225876805e-06,
+      "loss": 0.8125,
+      "step": 3053
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.488826423084251e-06,
+      "loss": 0.7573,
+      "step": 3054
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4826121705405945e-06,
+      "loss": 0.792,
+      "step": 3055
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.476399467694595e-06,
+      "loss": 0.7852,
+      "step": 3056
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.470188317283458e-06,
+      "loss": 0.7769,
+      "step": 3057
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.463978722043711e-06,
+      "loss": 0.8462,
+      "step": 3058
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.45777068471118e-06,
+      "loss": 0.791,
+      "step": 3059
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.451564208021022e-06,
+      "loss": 0.8257,
+      "step": 3060
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.445359294707694e-06,
+      "loss": 0.8203,
+      "step": 3061
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.439155947504973e-06,
+      "loss": 0.8335,
+      "step": 3062
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.432954169145937e-06,
+      "loss": 0.8115,
+      "step": 3063
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.426753962362981e-06,
+      "loss": 0.7632,
+      "step": 3064
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.420555329887806e-06,
+      "loss": 0.8359,
+      "step": 3065
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4143582744514135e-06,
+      "loss": 0.7798,
+      "step": 3066
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.408162798784112e-06,
+      "loss": 0.832,
+      "step": 3067
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.401968905615525e-06,
+      "loss": 0.7754,
+      "step": 3068
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.395776597674568e-06,
+      "loss": 0.7671,
+      "step": 3069
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3895858776894544e-06,
+      "loss": 0.8208,
+      "step": 3070
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.383396748387708e-06,
+      "loss": 0.7607,
+      "step": 3071
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.377209212496149e-06,
+      "loss": 0.7822,
+      "step": 3072
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.371023272740891e-06,
+      "loss": 0.8296,
+      "step": 3073
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.364838931847347e-06,
+      "loss": 0.8232,
+      "step": 3074
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.358656192540227e-06,
+      "loss": 0.7852,
+      "step": 3075
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.352475057543541e-06,
+      "loss": 0.8057,
+      "step": 3076
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.346295529580573e-06,
+      "loss": 0.7632,
+      "step": 3077
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.340117611373922e-06,
+      "loss": 0.8364,
+      "step": 3078
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.333941305645466e-06,
+      "loss": 0.7837,
+      "step": 3079
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.327766615116368e-06,
+      "loss": 0.8252,
+      "step": 3080
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.321593542507089e-06,
+      "loss": 0.8052,
+      "step": 3081
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.315422090537373e-06,
+      "loss": 0.7954,
+      "step": 3082
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.309252261926258e-06,
+      "loss": 0.7954,
+      "step": 3083
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.303084059392044e-06,
+      "loss": 0.793,
+      "step": 3084
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.29691748565234e-06,
+      "loss": 0.8696,
+      "step": 3085
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.29075254342403e-06,
+      "loss": 0.8076,
+      "step": 3086
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.284589235423267e-06,
+      "loss": 0.7344,
+      "step": 3087
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.278427564365499e-06,
+      "loss": 0.8008,
+      "step": 3088
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.272267532965448e-06,
+      "loss": 0.8047,
+      "step": 3089
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.266109143937111e-06,
+      "loss": 0.8408,
+      "step": 3090
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.259952399993762e-06,
+      "loss": 0.7876,
+      "step": 3091
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2537973038479505e-06,
+      "loss": 0.812,
+      "step": 3092
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.247643858211508e-06,
+      "loss": 0.7754,
+      "step": 3093
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.241492065795522e-06,
+      "loss": 0.7979,
+      "step": 3094
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2353419293103675e-06,
+      "loss": 0.7617,
+      "step": 3095
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.229193451465682e-06,
+      "loss": 0.8003,
+      "step": 3096
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.223046634970378e-06,
+      "loss": 0.7549,
+      "step": 3097
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.216901482532624e-06,
+      "loss": 0.8164,
+      "step": 3098
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.2107579968598685e-06,
+      "loss": 0.8262,
+      "step": 3099
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.204616180658822e-06,
+      "loss": 0.8291,
+      "step": 3100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1984760366354525e-06,
+      "loss": 0.7925,
+      "step": 3101
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.192337567494999e-06,
+      "loss": 0.7803,
+      "step": 3102
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.186200775941963e-06,
+      "loss": 0.8149,
+      "step": 3103
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.180065664680104e-06,
+      "loss": 0.8574,
+      "step": 3104
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.173932236412439e-06,
+      "loss": 0.7988,
+      "step": 3105
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.167800493841245e-06,
+      "loss": 0.8335,
+      "step": 3106
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1616704396680615e-06,
+      "loss": 0.8682,
+      "step": 3107
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.155542076593676e-06,
+      "loss": 0.835,
+      "step": 3108
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.149415407318133e-06,
+      "loss": 0.7886,
+      "step": 3109
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.143290434540738e-06,
+      "loss": 0.8301,
+      "step": 3110
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.137167160960045e-06,
+      "loss": 0.7822,
+      "step": 3111
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1310455892738495e-06,
+      "loss": 0.8037,
+      "step": 3112
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.124925722179211e-06,
+      "loss": 0.7993,
+      "step": 3113
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.118807562372434e-06,
+      "loss": 0.7944,
+      "step": 3114
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.112691112549065e-06,
+      "loss": 0.8472,
+      "step": 3115
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.106576375403904e-06,
+      "loss": 0.8291,
+      "step": 3116
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.100463353630992e-06,
+      "loss": 0.791,
+      "step": 3117
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.094352049923624e-06,
+      "loss": 0.7998,
+      "step": 3118
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.088242466974317e-06,
+      "loss": 0.7993,
+      "step": 3119
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.082134607474852e-06,
+      "loss": 0.8008,
+      "step": 3120
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.076028474116243e-06,
+      "loss": 0.8267,
+      "step": 3121
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.069924069588737e-06,
+      "loss": 0.8228,
+      "step": 3122
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.063821396581827e-06,
+      "loss": 0.7603,
+      "step": 3123
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0577204577842395e-06,
+      "loss": 0.8608,
+      "step": 3124
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.051621255883942e-06,
+      "loss": 0.8115,
+      "step": 3125
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.045523793568128e-06,
+      "loss": 0.8423,
+      "step": 3126
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.039428073523228e-06,
+      "loss": 0.7944,
+      "step": 3127
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.033334098434918e-06,
+      "loss": 0.7529,
+      "step": 3128
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0272418709880785e-06,
+      "loss": 0.772,
+      "step": 3129
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.021151393866843e-06,
+      "loss": 0.8105,
+      "step": 3130
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.015062669754565e-06,
+      "loss": 0.8154,
+      "step": 3131
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.00897570133383e-06,
+      "loss": 0.8257,
+      "step": 3132
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0028904912864395e-06,
+      "loss": 0.8071,
+      "step": 3133
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.996807042293429e-06,
+      "loss": 0.8325,
+      "step": 3134
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.990725357035064e-06,
+      "loss": 0.8296,
+      "step": 3135
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9846454381908155e-06,
+      "loss": 0.8086,
+      "step": 3136
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.978567288439389e-06,
+      "loss": 0.7915,
+      "step": 3137
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.97249091045871e-06,
+      "loss": 0.7964,
+      "step": 3138
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.966416306925923e-06,
+      "loss": 0.7944,
+      "step": 3139
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9603434805173835e-06,
+      "loss": 0.834,
+      "step": 3140
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.95427243390867e-06,
+      "loss": 0.812,
+      "step": 3141
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.948203169774581e-06,
+      "loss": 0.8247,
+      "step": 3142
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9421356907891195e-06,
+      "loss": 0.8267,
+      "step": 3143
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9360699996255075e-06,
+      "loss": 0.7886,
+      "step": 3144
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.930006098956184e-06,
+      "loss": 0.814,
+      "step": 3145
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.923943991452792e-06,
+      "loss": 0.8208,
+      "step": 3146
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.917883679786185e-06,
+      "loss": 0.7666,
+      "step": 3147
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.911825166626428e-06,
+      "loss": 0.7974,
+      "step": 3148
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.905768454642795e-06,
+      "loss": 0.8291,
+      "step": 3149
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8997135465037595e-06,
+      "loss": 0.8071,
+      "step": 3150
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.893660444877007e-06,
+      "loss": 0.8364,
+      "step": 3151
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.887609152429424e-06,
+      "loss": 0.8086,
+      "step": 3152
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.881559671827109e-06,
+      "loss": 0.7217,
+      "step": 3153
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.87551200573534e-06,
+      "loss": 0.8057,
+      "step": 3154
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.869466156818619e-06,
+      "loss": 0.8452,
+      "step": 3155
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.863422127740638e-06,
+      "loss": 0.7476,
+      "step": 3156
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.857379921164283e-06,
+      "loss": 0.7598,
+      "step": 3157
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.851339539751644e-06,
+      "loss": 0.77,
+      "step": 3158
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8453009861640006e-06,
+      "loss": 0.8149,
+      "step": 3159
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.83926426306184e-06,
+      "loss": 0.7476,
+      "step": 3160
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.833229373104823e-06,
+      "loss": 0.8154,
+      "step": 3161
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.827196318951818e-06,
+      "loss": 0.8101,
+      "step": 3162
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.82116510326088e-06,
+      "loss": 0.7681,
+      "step": 3163
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8151357286892515e-06,
+      "loss": 0.7549,
+      "step": 3164
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.809108197893368e-06,
+      "loss": 0.7612,
+      "step": 3165
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.803082513528853e-06,
+      "loss": 0.7632,
+      "step": 3166
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.79705867825051e-06,
+      "loss": 0.8457,
+      "step": 3167
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.791036694712336e-06,
+      "loss": 0.8555,
+      "step": 3168
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.785016565567505e-06,
+      "loss": 0.7627,
+      "step": 3169
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.778998293468384e-06,
+      "loss": 0.7676,
+      "step": 3170
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.772981881066506e-06,
+      "loss": 0.729,
+      "step": 3171
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.766967331012598e-06,
+      "loss": 0.7158,
+      "step": 3172
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.76095464595656e-06,
+      "loss": 0.8408,
+      "step": 3173
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.754943828547482e-06,
+      "loss": 0.793,
+      "step": 3174
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.748934881433612e-06,
+      "loss": 0.7793,
+      "step": 3175
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.742927807262385e-06,
+      "loss": 0.8032,
+      "step": 3176
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.736922608680413e-06,
+      "loss": 0.7949,
+      "step": 3177
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7309192883334745e-06,
+      "loss": 0.7764,
+      "step": 3178
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.724917848866527e-06,
+      "loss": 0.8506,
+      "step": 3179
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7189182929236965e-06,
+      "loss": 0.7578,
+      "step": 3180
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7129206231482815e-06,
+      "loss": 0.7842,
+      "step": 3181
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.706924842182739e-06,
+      "loss": 0.7278,
+      "step": 3182
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.700930952668704e-06,
+      "loss": 0.8403,
+      "step": 3183
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.694938957246988e-06,
+      "loss": 0.791,
+      "step": 3184
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.688948858557542e-06,
+      "loss": 0.7798,
+      "step": 3185
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.682960659239503e-06,
+      "loss": 0.7681,
+      "step": 3186
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.676974361931158e-06,
+      "loss": 0.8008,
+      "step": 3187
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.670989969269966e-06,
+      "loss": 0.8364,
+      "step": 3188
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.665007483892544e-06,
+      "loss": 0.8525,
+      "step": 3189
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.659026908434663e-06,
+      "loss": 0.8208,
+      "step": 3190
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.653048245531263e-06,
+      "loss": 0.8262,
+      "step": 3191
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.647071497816422e-06,
+      "loss": 0.7974,
+      "step": 3192
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6410966679234005e-06,
+      "loss": 0.7432,
+      "step": 3193
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.635123758484596e-06,
+      "loss": 0.7939,
+      "step": 3194
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6291527721315675e-06,
+      "loss": 0.75,
+      "step": 3195
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.623183711495019e-06,
+      "loss": 0.8164,
+      "step": 3196
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.617216579204814e-06,
+      "loss": 0.7627,
+      "step": 3197
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6112513778899615e-06,
+      "loss": 0.7734,
+      "step": 3198
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.605288110178621e-06,
+      "loss": 0.8208,
+      "step": 3199
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.599326778698106e-06,
+      "loss": 0.8076,
+      "step": 3200
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.593367386074867e-06,
+      "loss": 0.8018,
+      "step": 3201
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.587409934934507e-06,
+      "loss": 0.8604,
+      "step": 3202
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.581454427901769e-06,
+      "loss": 0.7842,
+      "step": 3203
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.575500867600545e-06,
+      "loss": 0.7998,
+      "step": 3204
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.569549256653868e-06,
+      "loss": 0.8013,
+      "step": 3205
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.563599597683905e-06,
+      "loss": 0.8057,
+      "step": 3206
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.557651893311971e-06,
+      "loss": 0.7837,
+      "step": 3207
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5517061461585135e-06,
+      "loss": 0.793,
+      "step": 3208
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.545762358843132e-06,
+      "loss": 0.7915,
+      "step": 3209
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.539820533984541e-06,
+      "loss": 0.7983,
+      "step": 3210
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5338806742006075e-06,
+      "loss": 0.8047,
+      "step": 3211
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.527942782108324e-06,
+      "loss": 0.7974,
+      "step": 3212
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.52200686032382e-06,
+      "loss": 0.8301,
+      "step": 3213
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.516072911462356e-06,
+      "loss": 0.7524,
+      "step": 3214
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.510140938138322e-06,
+      "loss": 0.8403,
+      "step": 3215
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.504210942965242e-06,
+      "loss": 0.814,
+      "step": 3216
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.498282928555754e-06,
+      "loss": 0.7632,
+      "step": 3217
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.492356897521644e-06,
+      "loss": 0.7612,
+      "step": 3218
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.486432852473817e-06,
+      "loss": 0.7861,
+      "step": 3219
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.480510796022294e-06,
+      "loss": 0.7578,
+      "step": 3220
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.474590730776226e-06,
+      "loss": 0.8213,
+      "step": 3221
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.46867265934389e-06,
+      "loss": 0.8457,
+      "step": 3222
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.46275658433268e-06,
+      "loss": 0.8193,
+      "step": 3223
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.456842508349112e-06,
+      "loss": 0.8325,
+      "step": 3224
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4509304339988225e-06,
+      "loss": 0.8022,
+      "step": 3225
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.445020363886568e-06,
+      "loss": 0.7385,
+      "step": 3226
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4391123006162075e-06,
+      "loss": 0.7817,
+      "step": 3227
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4332062467907365e-06,
+      "loss": 0.8091,
+      "step": 3228
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.427302205012257e-06,
+      "loss": 0.812,
+      "step": 3229
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.421400177881983e-06,
+      "loss": 0.7642,
+      "step": 3230
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.415500168000234e-06,
+      "loss": 0.812,
+      "step": 3231
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.409602177966453e-06,
+      "loss": 0.7808,
+      "step": 3232
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.403706210379187e-06,
+      "loss": 0.7891,
+      "step": 3233
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.397812267836096e-06,
+      "loss": 0.8286,
+      "step": 3234
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.39192035293394e-06,
+      "loss": 0.813,
+      "step": 3235
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.386030468268594e-06,
+      "loss": 0.812,
+      "step": 3236
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3801426164350335e-06,
+      "loss": 0.832,
+      "step": 3237
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.374256800027341e-06,
+      "loss": 0.7349,
+      "step": 3238
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3683730216387e-06,
+      "loss": 0.7856,
+      "step": 3239
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.362491283861402e-06,
+      "loss": 0.8359,
+      "step": 3240
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.356611589286824e-06,
+      "loss": 0.751,
+      "step": 3241
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.350733940505458e-06,
+      "loss": 0.7852,
+      "step": 3242
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.344858340106895e-06,
+      "loss": 0.811,
+      "step": 3243
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.338984790679819e-06,
+      "loss": 0.8354,
+      "step": 3244
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.333113294812002e-06,
+      "loss": 0.7739,
+      "step": 3245
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.327243855090324e-06,
+      "loss": 0.8267,
+      "step": 3246
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.321376474100753e-06,
+      "loss": 0.7661,
+      "step": 3247
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.315511154428354e-06,
+      "loss": 0.7949,
+      "step": 3248
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.309647898657277e-06,
+      "loss": 0.7368,
+      "step": 3249
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.30378670937077e-06,
+      "loss": 0.8066,
+      "step": 3250
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.29792758915117e-06,
+      "loss": 0.7896,
+      "step": 3251
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.292070540579888e-06,
+      "loss": 0.8164,
+      "step": 3252
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.286215566237445e-06,
+      "loss": 0.8228,
+      "step": 3253
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.28036266870344e-06,
+      "loss": 0.7832,
+      "step": 3254
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.274511850556543e-06,
+      "loss": 0.8047,
+      "step": 3255
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.268663114374524e-06,
+      "loss": 0.8149,
+      "step": 3256
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2628164627342315e-06,
+      "loss": 0.7439,
+      "step": 3257
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.256971898211593e-06,
+      "loss": 0.7871,
+      "step": 3258
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.251129423381622e-06,
+      "loss": 0.7808,
+      "step": 3259
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.245289040818403e-06,
+      "loss": 0.771,
+      "step": 3260
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.23945075309511e-06,
+      "loss": 0.7563,
+      "step": 3261
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.233614562783975e-06,
+      "loss": 0.8501,
+      "step": 3262
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.227780472456329e-06,
+      "loss": 0.8276,
+      "step": 3263
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.221948484682564e-06,
+      "loss": 0.7002,
+      "step": 3264
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.216118602032152e-06,
+      "loss": 0.8623,
+      "step": 3265
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.210290827073627e-06,
+      "loss": 0.7661,
+      "step": 3266
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.204465162374603e-06,
+      "loss": 0.8125,
+      "step": 3267
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.198641610501771e-06,
+      "loss": 0.792,
+      "step": 3268
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.192820174020874e-06,
+      "loss": 0.7495,
+      "step": 3269
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1870008554967356e-06,
+      "loss": 0.7925,
+      "step": 3270
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.181183657493241e-06,
+      "loss": 0.8164,
+      "step": 3271
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.175368582573343e-06,
+      "loss": 0.8188,
+      "step": 3272
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.16955563329906e-06,
+      "loss": 0.8052,
+      "step": 3273
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.163744812231473e-06,
+      "loss": 0.8057,
+      "step": 3274
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.157936121930728e-06,
+      "loss": 0.7275,
+      "step": 3275
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.152129564956021e-06,
+      "loss": 0.7749,
+      "step": 3276
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.146325143865617e-06,
+      "loss": 0.7964,
+      "step": 3277
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.140522861216846e-06,
+      "loss": 0.7822,
+      "step": 3278
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.134722719566092e-06,
+      "loss": 0.8159,
+      "step": 3279
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.128924721468782e-06,
+      "loss": 0.7822,
+      "step": 3280
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1231288694794144e-06,
+      "loss": 0.7891,
+      "step": 3281
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.117335166151536e-06,
+      "loss": 0.7842,
+      "step": 3282
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.111543614037751e-06,
+      "loss": 0.7236,
+      "step": 3283
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.105754215689709e-06,
+      "loss": 0.8223,
+      "step": 3284
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.099966973658117e-06,
+      "loss": 0.8135,
+      "step": 3285
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.094181890492733e-06,
+      "loss": 0.814,
+      "step": 3286
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.08839896874235e-06,
+      "loss": 0.7686,
+      "step": 3287
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.082618210954827e-06,
+      "loss": 0.7876,
+      "step": 3288
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.076839619677066e-06,
+      "loss": 0.751,
+      "step": 3289
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.071063197454997e-06,
+      "loss": 0.7979,
+      "step": 3290
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.065288946833618e-06,
+      "loss": 0.8223,
+      "step": 3291
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.059516870356951e-06,
+      "loss": 0.812,
+      "step": 3292
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.053746970568081e-06,
+      "loss": 0.8438,
+      "step": 3293
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.047979250009112e-06,
+      "loss": 0.7568,
+      "step": 3294
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.042213711221201e-06,
+      "loss": 0.7812,
+      "step": 3295
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0364503567445435e-06,
+      "loss": 0.7529,
+      "step": 3296
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.030689189118358e-06,
+      "loss": 0.7944,
+      "step": 3297
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.024930210880924e-06,
+      "loss": 0.8018,
+      "step": 3298
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.019173424569538e-06,
+      "loss": 0.7417,
+      "step": 3299
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.01341883272054e-06,
+      "loss": 0.7354,
+      "step": 3300
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.007666437869294e-06,
+      "loss": 0.7974,
+      "step": 3301
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0019162425502e-06,
+      "loss": 0.7939,
+      "step": 3302
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.996168249296701e-06,
+      "loss": 0.8247,
+      "step": 3303
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.99042246064125e-06,
+      "loss": 0.7812,
+      "step": 3304
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.984678879115339e-06,
+      "loss": 0.8276,
+      "step": 3305
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.978937507249488e-06,
+      "loss": 0.8511,
+      "step": 3306
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9731983475732415e-06,
+      "loss": 0.7759,
+      "step": 3307
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.96746140261517e-06,
+      "loss": 0.7773,
+      "step": 3308
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.961726674902868e-06,
+      "loss": 0.8389,
+      "step": 3309
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.955994166962958e-06,
+      "loss": 0.8232,
+      "step": 3310
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.950263881321069e-06,
+      "loss": 0.8008,
+      "step": 3311
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.944535820501864e-06,
+      "loss": 0.7939,
+      "step": 3312
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.938809987029032e-06,
+      "loss": 0.7686,
+      "step": 3313
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.933086383425269e-06,
+      "loss": 0.7559,
+      "step": 3314
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.927365012212285e-06,
+      "loss": 0.7529,
+      "step": 3315
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.921645875910818e-06,
+      "loss": 0.8462,
+      "step": 3316
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.915928977040615e-06,
+      "loss": 0.7214,
+      "step": 3317
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.91021431812044e-06,
+      "loss": 0.8003,
+      "step": 3318
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.904501901668066e-06,
+      "loss": 0.769,
+      "step": 3319
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.898791730200284e-06,
+      "loss": 0.7681,
+      "step": 3320
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.893083806232893e-06,
+      "loss": 0.7893,
+      "step": 3321
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.887378132280694e-06,
+      "loss": 0.749,
+      "step": 3322
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.881674710857513e-06,
+      "loss": 0.8091,
+      "step": 3323
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.875973544476173e-06,
+      "loss": 0.8271,
+      "step": 3324
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8702746356485005e-06,
+      "loss": 0.8198,
+      "step": 3325
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.864577986885335e-06,
+      "loss": 0.8003,
+      "step": 3326
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8588836006965135e-06,
+      "loss": 0.7573,
+      "step": 3327
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.85319147959089e-06,
+      "loss": 0.7075,
+      "step": 3328
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.847501626076301e-06,
+      "loss": 0.7583,
+      "step": 3329
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.841814042659595e-06,
+      "loss": 0.8091,
+      "step": 3330
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.83612873184662e-06,
+      "loss": 0.7954,
+      "step": 3331
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.830445696142222e-06,
+      "loss": 0.7549,
+      "step": 3332
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.824764938050242e-06,
+      "loss": 0.7861,
+      "step": 3333
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.819086460073523e-06,
+      "loss": 0.8018,
+      "step": 3334
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8134102647139e-06,
+      "loss": 0.7798,
+      "step": 3335
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.807736354472198e-06,
+      "loss": 0.8281,
+      "step": 3336
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.802064731848237e-06,
+      "loss": 0.79,
+      "step": 3337
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.796395399340844e-06,
+      "loss": 0.7332,
+      "step": 3338
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.790728359447813e-06,
+      "loss": 0.7964,
+      "step": 3339
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7850636146659454e-06,
+      "loss": 0.7661,
+      "step": 3340
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.779401167491023e-06,
+      "loss": 0.749,
+      "step": 3341
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.773741020417819e-06,
+      "loss": 0.7559,
+      "step": 3342
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.768083175940093e-06,
+      "loss": 0.7573,
+      "step": 3343
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.762427636550586e-06,
+      "loss": 0.8242,
+      "step": 3344
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.756774404741035e-06,
+      "loss": 0.792,
+      "step": 3345
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.751123483002141e-06,
+      "loss": 0.8647,
+      "step": 3346
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7454748738236e-06,
+      "loss": 0.8071,
+      "step": 3347
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.739828579694095e-06,
+      "loss": 0.8105,
+      "step": 3348
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.73418460310128e-06,
+      "loss": 0.7456,
+      "step": 3349
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.728542946531784e-06,
+      "loss": 0.8271,
+      "step": 3350
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.722903612471219e-06,
+      "loss": 0.7437,
+      "step": 3351
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.717266603404178e-06,
+      "loss": 0.8125,
+      "step": 3352
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.711631921814224e-06,
+      "loss": 0.7759,
+      "step": 3353
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.705999570183895e-06,
+      "loss": 0.769,
+      "step": 3354
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.700369550994707e-06,
+      "loss": 0.7202,
+      "step": 3355
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.694741866727141e-06,
+      "loss": 0.7773,
+      "step": 3356
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.689116519860655e-06,
+      "loss": 0.7603,
+      "step": 3357
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.683493512873675e-06,
+      "loss": 0.7788,
+      "step": 3358
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.677872848243601e-06,
+      "loss": 0.7236,
+      "step": 3359
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.672254528446788e-06,
+      "loss": 0.7852,
+      "step": 3360
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.666638555958572e-06,
+      "loss": 0.8423,
+      "step": 3361
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.661024933253242e-06,
+      "loss": 0.7979,
+      "step": 3362
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.655413662804073e-06,
+      "loss": 0.7754,
+      "step": 3363
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.649804747083279e-06,
+      "loss": 0.7969,
+      "step": 3364
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.644198188562048e-06,
+      "loss": 0.7778,
+      "step": 3365
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.638593989710531e-06,
+      "loss": 0.7505,
+      "step": 3366
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.632992152997836e-06,
+      "loss": 0.8257,
+      "step": 3367
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.627392680892031e-06,
+      "loss": 0.8403,
+      "step": 3368
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6217955758601426e-06,
+      "loss": 0.8232,
+      "step": 3369
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.616200840368159e-06,
+      "loss": 0.752,
+      "step": 3370
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.610608476881012e-06,
+      "loss": 0.8438,
+      "step": 3371
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.605018487862594e-06,
+      "loss": 0.8013,
+      "step": 3372
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5994308757757685e-06,
+      "loss": 0.769,
+      "step": 3373
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.593845643082323e-06,
+      "loss": 0.749,
+      "step": 3374
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.588262792243013e-06,
+      "loss": 0.8027,
+      "step": 3375
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.582682325717545e-06,
+      "loss": 0.77,
+      "step": 3376
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.577104245964569e-06,
+      "loss": 0.792,
+      "step": 3377
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.571528555441687e-06,
+      "loss": 0.7905,
+      "step": 3378
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5659552566054485e-06,
+      "loss": 0.77,
+      "step": 3379
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5603843519113525e-06,
+      "loss": 0.7285,
+      "step": 3380
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5548158438138275e-06,
+      "loss": 0.8003,
+      "step": 3381
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.549249734766269e-06,
+      "loss": 0.772,
+      "step": 3382
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.543686027221001e-06,
+      "loss": 0.8149,
+      "step": 3383
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.538124723629294e-06,
+      "loss": 0.8354,
+      "step": 3384
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.532565826441354e-06,
+      "loss": 0.7837,
+      "step": 3385
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.527009338106333e-06,
+      "loss": 0.7798,
+      "step": 3386
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.52145526107232e-06,
+      "loss": 0.7812,
+      "step": 3387
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.515903597786339e-06,
+      "loss": 0.7754,
+      "step": 3388
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.510354350694357e-06,
+      "loss": 0.7915,
+      "step": 3389
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.504807522241269e-06,
+      "loss": 0.8003,
+      "step": 3390
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.499263114870908e-06,
+      "loss": 0.7715,
+      "step": 3391
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.493721131026042e-06,
+      "loss": 0.8242,
+      "step": 3392
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.488181573148367e-06,
+      "loss": 0.8022,
+      "step": 3393
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.482644443678516e-06,
+      "loss": 0.7832,
+      "step": 3394
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.477109745056043e-06,
+      "loss": 0.8125,
+      "step": 3395
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.471577479719437e-06,
+      "loss": 0.7793,
+      "step": 3396
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.466047650106114e-06,
+      "loss": 0.8081,
+      "step": 3397
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.460520258652426e-06,
+      "loss": 0.8042,
+      "step": 3398
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4549953077936315e-06,
+      "loss": 0.792,
+      "step": 3399
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.449472799963927e-06,
+      "loss": 0.8247,
+      "step": 3400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.443952737596433e-06,
+      "loss": 0.8223,
+      "step": 3401
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.438435123123186e-06,
+      "loss": 0.77,
+      "step": 3402
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.43291995897515e-06,
+      "loss": 0.8008,
+      "step": 3403
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.427407247582205e-06,
+      "loss": 0.7578,
+      "step": 3404
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.421896991373157e-06,
+      "loss": 0.7798,
+      "step": 3405
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.416389192775714e-06,
+      "loss": 0.771,
+      "step": 3406
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.410883854216524e-06,
+      "loss": 0.8418,
+      "step": 3407
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.40538097812114e-06,
+      "loss": 0.7847,
+      "step": 3408
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.399880566914022e-06,
+      "loss": 0.7925,
+      "step": 3409
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.394382623018557e-06,
+      "loss": 0.7974,
+      "step": 3410
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.38888714885704e-06,
+      "loss": 0.8154,
+      "step": 3411
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.383394146850674e-06,
+      "loss": 0.7427,
+      "step": 3412
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.377903619419583e-06,
+      "loss": 0.8052,
+      "step": 3413
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.372415568982788e-06,
+      "loss": 0.729,
+      "step": 3414
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.366929997958235e-06,
+      "loss": 0.7896,
+      "step": 3415
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.361446908762751e-06,
+      "loss": 0.7954,
+      "step": 3416
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3559663038121e-06,
+      "loss": 0.7793,
+      "step": 3417
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3504881855209345e-06,
+      "loss": 0.7861,
+      "step": 3418
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.345012556302818e-06,
+      "loss": 0.8071,
+      "step": 3419
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.339539418570204e-06,
+      "loss": 0.7559,
+      "step": 3420
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.334068774734467e-06,
+      "loss": 0.8096,
+      "step": 3421
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.32860062720587e-06,
+      "loss": 0.7637,
+      "step": 3422
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.32313497839358e-06,
+      "loss": 0.79,
+      "step": 3423
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.317671830705667e-06,
+      "loss": 0.7988,
+      "step": 3424
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.312211186549092e-06,
+      "loss": 0.7949,
+      "step": 3425
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3067530483297185e-06,
+      "loss": 0.8145,
+      "step": 3426
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3012974184523024e-06,
+      "loss": 0.7881,
+      "step": 3427
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.295844299320496e-06,
+      "loss": 0.8076,
+      "step": 3428
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.290393693336849e-06,
+      "loss": 0.7969,
+      "step": 3429
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.284945602902793e-06,
+      "loss": 0.7808,
+      "step": 3430
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.279500030418659e-06,
+      "loss": 0.8296,
+      "step": 3431
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.274056978283675e-06,
+      "loss": 0.7578,
+      "step": 3432
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.26861644889595e-06,
+      "loss": 0.7446,
+      "step": 3433
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.263178444652478e-06,
+      "loss": 0.8286,
+      "step": 3434
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.257742967949149e-06,
+      "loss": 0.7485,
+      "step": 3435
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.252310021180735e-06,
+      "loss": 0.8257,
+      "step": 3436
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.246879606740895e-06,
+      "loss": 0.791,
+      "step": 3437
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.241451727022172e-06,
+      "loss": 0.7983,
+      "step": 3438
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.236026384415992e-06,
+      "loss": 0.812,
+      "step": 3439
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.230603581312665e-06,
+      "loss": 0.7939,
+      "step": 3440
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.225183320101372e-06,
+      "loss": 0.7979,
+      "step": 3441
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2197656031701914e-06,
+      "loss": 0.8091,
+      "step": 3442
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2143504329060724e-06,
+      "loss": 0.8022,
+      "step": 3443
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.208937811694834e-06,
+      "loss": 0.8472,
+      "step": 3444
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2035277419211824e-06,
+      "loss": 0.731,
+      "step": 3445
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1981202259686996e-06,
+      "loss": 0.7964,
+      "step": 3446
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.192715266219835e-06,
+      "loss": 0.8047,
+      "step": 3447
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.18731286505592e-06,
+      "loss": 0.793,
+      "step": 3448
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.181913024857154e-06,
+      "loss": 0.7466,
+      "step": 3449
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.176515748002613e-06,
+      "loss": 0.8086,
+      "step": 3450
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.171121036870228e-06,
+      "loss": 0.7744,
+      "step": 3451
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.165728893836825e-06,
+      "loss": 0.79,
+      "step": 3452
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.160339321278077e-06,
+      "loss": 0.8057,
+      "step": 3453
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1549523215685406e-06,
+      "loss": 0.8384,
+      "step": 3454
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.149567897081623e-06,
+      "loss": 0.7837,
+      "step": 3455
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1441860501896e-06,
+      "loss": 0.8193,
+      "step": 3456
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.138806783263633e-06,
+      "loss": 0.7686,
+      "step": 3457
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1334300986737146e-06,
+      "loss": 0.7925,
+      "step": 3458
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.128055998788722e-06,
+      "loss": 0.8032,
+      "step": 3459
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.122684485976384e-06,
+      "loss": 0.7563,
+      "step": 3460
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.117315562603295e-06,
+      "loss": 0.8096,
+      "step": 3461
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.111949231034903e-06,
+      "loss": 0.771,
+      "step": 3462
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.106585493635519e-06,
+      "loss": 0.7878,
+      "step": 3463
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.101224352768312e-06,
+      "loss": 0.8066,
+      "step": 3464
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.095865810795297e-06,
+      "loss": 0.8262,
+      "step": 3465
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.090509870077351e-06,
+      "loss": 0.8159,
+      "step": 3466
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.085156532974211e-06,
+      "loss": 0.8184,
+      "step": 3467
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.079805801844463e-06,
+      "loss": 0.7749,
+      "step": 3468
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.074457679045534e-06,
+      "loss": 0.7637,
+      "step": 3469
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.069112166933713e-06,
+      "loss": 0.8022,
+      "step": 3470
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.063769267864138e-06,
+      "loss": 0.7646,
+      "step": 3471
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.058428984190794e-06,
+      "loss": 0.7578,
+      "step": 3472
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.053091318266512e-06,
+      "loss": 0.7573,
+      "step": 3473
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.047756272442974e-06,
+      "loss": 0.8164,
+      "step": 3474
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.042423849070707e-06,
+      "loss": 0.7729,
+      "step": 3475
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.037094050499069e-06,
+      "loss": 0.8208,
+      "step": 3476
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.031766879076285e-06,
+      "loss": 0.7354,
+      "step": 3477
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.026442337149412e-06,
+      "loss": 0.749,
+      "step": 3478
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.021120427064336e-06,
+      "loss": 0.7993,
+      "step": 3479
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0158011511658035e-06,
+      "loss": 0.8159,
+      "step": 3480
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.010484511797384e-06,
+      "loss": 0.7803,
+      "step": 3481
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.005170511301505e-06,
+      "loss": 0.7534,
+      "step": 3482
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9998591520194086e-06,
+      "loss": 0.7974,
+      "step": 3483
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.994550436291188e-06,
+      "loss": 0.7866,
+      "step": 3484
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9892443664557665e-06,
+      "loss": 0.7944,
+      "step": 3485
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9839409448509034e-06,
+      "loss": 0.7905,
+      "step": 3486
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.978640173813191e-06,
+      "loss": 0.8125,
+      "step": 3487
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.973342055678052e-06,
+      "loss": 0.8159,
+      "step": 3488
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.968046592779748e-06,
+      "loss": 0.7905,
+      "step": 3489
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.962753787451353e-06,
+      "loss": 0.749,
+      "step": 3490
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9574636420247866e-06,
+      "loss": 0.8223,
+      "step": 3491
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9521761588308e-06,
+      "loss": 0.7456,
+      "step": 3492
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.946891340198951e-06,
+      "loss": 0.7549,
+      "step": 3493
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9416091884576404e-06,
+      "loss": 0.8184,
+      "step": 3494
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.936329705934089e-06,
+      "loss": 0.8164,
+      "step": 3495
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.93105289495434e-06,
+      "loss": 0.8071,
+      "step": 3496
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.925778757843264e-06,
+      "loss": 0.8105,
+      "step": 3497
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.920507296924549e-06,
+      "loss": 0.8145,
+      "step": 3498
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.915238514520709e-06,
+      "loss": 0.7534,
+      "step": 3499
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.909972412953069e-06,
+      "loss": 0.7954,
+      "step": 3500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.904708994541776e-06,
+      "loss": 0.7539,
+      "step": 3501
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.899448261605806e-06,
+      "loss": 0.7832,
+      "step": 3502
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.894190216462943e-06,
+      "loss": 0.7954,
+      "step": 3503
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.88893486142978e-06,
+      "loss": 0.8203,
+      "step": 3504
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8836821988217335e-06,
+      "loss": 0.7959,
+      "step": 3505
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8784322309530334e-06,
+      "loss": 0.7788,
+      "step": 3506
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.873184960136721e-06,
+      "loss": 0.8086,
+      "step": 3507
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.867940388684648e-06,
+      "loss": 0.7983,
+      "step": 3508
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.862698518907479e-06,
+      "loss": 0.8257,
+      "step": 3509
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8574593531146884e-06,
+      "loss": 0.7852,
+      "step": 3510
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.852222893614556e-06,
+      "loss": 0.8032,
+      "step": 3511
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.846989142714174e-06,
+      "loss": 0.8643,
+      "step": 3512
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8417581027194385e-06,
+      "loss": 0.8428,
+      "step": 3513
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.836529775935047e-06,
+      "loss": 0.769,
+      "step": 3514
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8313041646645095e-06,
+      "loss": 0.7324,
+      "step": 3515
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.826081271210131e-06,
+      "loss": 0.8169,
+      "step": 3516
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.820861097873038e-06,
+      "loss": 0.7876,
+      "step": 3517
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.81564364695313e-06,
+      "loss": 0.7979,
+      "step": 3518
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8104289207491285e-06,
+      "loss": 0.8159,
+      "step": 3519
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8052169215585465e-06,
+      "loss": 0.7739,
+      "step": 3520
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.800007651677696e-06,
+      "loss": 0.7983,
+      "step": 3521
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7948011134016903e-06,
+      "loss": 0.8062,
+      "step": 3522
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.789597309024433e-06,
+      "loss": 0.7734,
+      "step": 3523
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7843962408386325e-06,
+      "loss": 0.7788,
+      "step": 3524
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7791979111357767e-06,
+      "loss": 0.8032,
+      "step": 3525
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.774002322206157e-06,
+      "loss": 0.792,
+      "step": 3526
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7688094763388673e-06,
+      "loss": 0.7476,
+      "step": 3527
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7636193758217697e-06,
+      "loss": 0.73,
+      "step": 3528
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7584320229415337e-06,
+      "loss": 0.7734,
+      "step": 3529
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7532474199836133e-06,
+      "loss": 0.7754,
+      "step": 3530
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.748065569232251e-06,
+      "loss": 0.749,
+      "step": 3531
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.742886472970476e-06,
+      "loss": 0.7881,
+      "step": 3532
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.737710133480106e-06,
+      "loss": 0.8086,
+      "step": 3533
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7325365530417458e-06,
+      "loss": 0.8115,
+      "step": 3534
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.72736573393477e-06,
+      "loss": 0.7764,
+      "step": 3535
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7221976784373616e-06,
+      "loss": 0.8071,
+      "step": 3536
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.717032388826466e-06,
+      "loss": 0.7661,
+      "step": 3537
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.711869867377822e-06,
+      "loss": 0.7998,
+      "step": 3538
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7067101163659367e-06,
+      "loss": 0.793,
+      "step": 3539
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7015531380641055e-06,
+      "loss": 0.7749,
+      "step": 3540
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.696398934744402e-06,
+      "loss": 0.7739,
+      "step": 3541
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6912475086776732e-06,
+      "loss": 0.7871,
+      "step": 3542
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.686098862133546e-06,
+      "loss": 0.7563,
+      "step": 3543
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.680952997380419e-06,
+      "loss": 0.7441,
+      "step": 3544
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.675809916685471e-06,
+      "loss": 0.7903,
+      "step": 3545
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6706696223146477e-06,
+      "loss": 0.7705,
+      "step": 3546
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.665532116532672e-06,
+      "loss": 0.728,
+      "step": 3547
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6603974016030387e-06,
+      "loss": 0.7808,
+      "step": 3548
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6552654797880026e-06,
+      "loss": 0.7778,
+      "step": 3549
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6501363533486024e-06,
+      "loss": 0.7964,
+      "step": 3550
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.645010024544633e-06,
+      "loss": 0.7773,
+      "step": 3551
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6398864956346724e-06,
+      "loss": 0.8379,
+      "step": 3552
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6347657688760475e-06,
+      "loss": 0.7969,
+      "step": 3553
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6296478465248585e-06,
+      "loss": 0.7891,
+      "step": 3554
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6245327308359712e-06,
+      "loss": 0.8047,
+      "step": 3555
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6194204240630137e-06,
+      "loss": 0.7891,
+      "step": 3556
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6143109284583754e-06,
+      "loss": 0.7656,
+      "step": 3557
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.609204246273208e-06,
+      "loss": 0.7744,
+      "step": 3558
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.604100379757428e-06,
+      "loss": 0.7544,
+      "step": 3559
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.598999331159695e-06,
+      "loss": 0.7788,
+      "step": 3560
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5939011027274505e-06,
+      "loss": 0.7866,
+      "step": 3561
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5888056967068818e-06,
+      "loss": 0.7422,
+      "step": 3562
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5837131153429263e-06,
+      "loss": 0.7451,
+      "step": 3563
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5786233608792862e-06,
+      "loss": 0.7725,
+      "step": 3564
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.573536435558417e-06,
+      "loss": 0.791,
+      "step": 3565
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5684523416215244e-06,
+      "loss": 0.8042,
+      "step": 3566
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5633710813085687e-06,
+      "loss": 0.8101,
+      "step": 3567
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.558292656858264e-06,
+      "loss": 0.7515,
+      "step": 3568
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5532170705080737e-06,
+      "loss": 0.7246,
+      "step": 3569
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5481443244941995e-06,
+      "loss": 0.8345,
+      "step": 3570
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5430744210516134e-06,
+      "loss": 0.7671,
+      "step": 3571
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5380073624140175e-06,
+      "loss": 0.7793,
+      "step": 3572
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5329431508138734e-06,
+      "loss": 0.7891,
+      "step": 3573
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5278817884823723e-06,
+      "loss": 0.8022,
+      "step": 3574
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.522823277649463e-06,
+      "loss": 0.8076,
+      "step": 3575
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.517767620543834e-06,
+      "loss": 0.7231,
+      "step": 3576
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.512714819392916e-06,
+      "loss": 0.7808,
+      "step": 3577
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.507664876422885e-06,
+      "loss": 0.7722,
+      "step": 3578
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5026177938586515e-06,
+      "loss": 0.7671,
+      "step": 3579
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.49757357392387e-06,
+      "loss": 0.8325,
+      "step": 3580
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.492532218840935e-06,
+      "loss": 0.7642,
+      "step": 3581
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.487493730830973e-06,
+      "loss": 0.7651,
+      "step": 3582
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.482458112113858e-06,
+      "loss": 0.7134,
+      "step": 3583
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4774253649081824e-06,
+      "loss": 0.8193,
+      "step": 3584
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4723954914312874e-06,
+      "loss": 0.7734,
+      "step": 3585
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.467368493899249e-06,
+      "loss": 0.7588,
+      "step": 3586
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4623443745268724e-06,
+      "loss": 0.8003,
+      "step": 3587
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4573231355276872e-06,
+      "loss": 0.7627,
+      "step": 3588
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.452304779113964e-06,
+      "loss": 0.8154,
+      "step": 3589
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.447289307496701e-06,
+      "loss": 0.7412,
+      "step": 3590
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4422767228856235e-06,
+      "loss": 0.7437,
+      "step": 3591
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.437267027489186e-06,
+      "loss": 0.7607,
+      "step": 3592
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4322602235145708e-06,
+      "loss": 0.7974,
+      "step": 3593
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4272563131676884e-06,
+      "loss": 0.7432,
+      "step": 3594
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.422255298653161e-06,
+      "loss": 0.7905,
+      "step": 3595
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.417257182174356e-06,
+      "loss": 0.7627,
+      "step": 3596
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.412261965933353e-06,
+      "loss": 0.7666,
+      "step": 3597
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.407269652130948e-06,
+      "loss": 0.8237,
+      "step": 3598
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4022802429666667e-06,
+      "loss": 0.7695,
+      "step": 3599
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.397293740638754e-06,
+      "loss": 0.7954,
+      "step": 3600
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3923101473441734e-06,
+      "loss": 0.7866,
+      "step": 3601
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.387329465278605e-06,
+      "loss": 0.8018,
+      "step": 3602
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3823516966364467e-06,
+      "loss": 0.7109,
+      "step": 3603
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3773768436108188e-06,
+      "loss": 0.793,
+      "step": 3604
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.372404908393542e-06,
+      "loss": 0.7598,
+      "step": 3605
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3674358931751705e-06,
+      "loss": 0.7749,
+      "step": 3606
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3624698001449584e-06,
+      "loss": 0.7852,
+      "step": 3607
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3575066314908823e-06,
+      "loss": 0.769,
+      "step": 3608
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3525463893996167e-06,
+      "loss": 0.8662,
+      "step": 3609
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3475890760565544e-06,
+      "loss": 0.8096,
+      "step": 3610
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.34263469364581e-06,
+      "loss": 0.7456,
+      "step": 3611
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3376832443501838e-06,
+      "loss": 0.7329,
+      "step": 3612
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.332734730351199e-06,
+      "loss": 0.7393,
+      "step": 3613
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3277891538290806e-06,
+      "loss": 0.7573,
+      "step": 3614
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.322846516962762e-06,
+      "loss": 0.8252,
+      "step": 3615
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3179068219298792e-06,
+      "loss": 0.771,
+      "step": 3616
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3129700709067724e-06,
+      "loss": 0.7974,
+      "step": 3617
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3080362660684885e-06,
+      "loss": 0.8179,
+      "step": 3618
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3031054095887673e-06,
+      "loss": 0.7749,
+      "step": 3619
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.298177503640053e-06,
+      "loss": 0.771,
+      "step": 3620
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.293252550393502e-06,
+      "loss": 0.7842,
+      "step": 3621
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.288330552018957e-06,
+      "loss": 0.8003,
+      "step": 3622
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2834115106849564e-06,
+      "loss": 0.7632,
+      "step": 3623
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2784954285587444e-06,
+      "loss": 0.8242,
+      "step": 3624
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.273582307806258e-06,
+      "loss": 0.7949,
+      "step": 3625
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.268672150592129e-06,
+      "loss": 0.77,
+      "step": 3626
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.263764959079686e-06,
+      "loss": 0.7798,
+      "step": 3627
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2588607354309465e-06,
+      "loss": 0.7681,
+      "step": 3628
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.253959481806629e-06,
+      "loss": 0.8232,
+      "step": 3629
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.249061200366126e-06,
+      "loss": 0.7366,
+      "step": 3630
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2441658932675413e-06,
+      "loss": 0.7847,
+      "step": 3631
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2392735626676607e-06,
+      "loss": 0.8086,
+      "step": 3632
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.234384210721949e-06,
+      "loss": 0.7568,
+      "step": 3633
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.22949783958457e-06,
+      "loss": 0.8042,
+      "step": 3634
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2246144514083677e-06,
+      "loss": 0.8101,
+      "step": 3635
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.219734048344885e-06,
+      "loss": 0.7759,
+      "step": 3636
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2148566325443285e-06,
+      "loss": 0.7983,
+      "step": 3637
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2099822061556053e-06,
+      "loss": 0.7783,
+      "step": 3638
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2051107713262998e-06,
+      "loss": 0.7344,
+      "step": 3639
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2002423302026707e-06,
+      "loss": 0.7432,
+      "step": 3640
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.195376884929675e-06,
+      "loss": 0.7524,
+      "step": 3641
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1905144376509367e-06,
+      "loss": 0.8086,
+      "step": 3642
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.185654990508765e-06,
+      "loss": 0.7603,
+      "step": 3643
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1807985456441404e-06,
+      "loss": 0.7949,
+      "step": 3644
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1759451051967217e-06,
+      "loss": 0.771,
+      "step": 3645
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1710946713048584e-06,
+      "loss": 0.7339,
+      "step": 3646
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.166247246105556e-06,
+      "loss": 0.7788,
+      "step": 3647
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1614028317345035e-06,
+      "loss": 0.8086,
+      "step": 3648
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1565614303260637e-06,
+      "loss": 0.814,
+      "step": 3649
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1517230440132717e-06,
+      "loss": 0.8027,
+      "step": 3650
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.146887674927831e-06,
+      "loss": 0.8027,
+      "step": 3651
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.142055325200121e-06,
+      "loss": 0.7417,
+      "step": 3652
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.137225996959188e-06,
+      "loss": 0.7681,
+      "step": 3653
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1323996923327417e-06,
+      "loss": 0.7798,
+      "step": 3654
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1275764134471654e-06,
+      "loss": 0.7563,
+      "step": 3655
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.122756162427515e-06,
+      "loss": 0.8076,
+      "step": 3656
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1179389413975057e-06,
+      "loss": 0.7446,
+      "step": 3657
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.113124752479513e-06,
+      "loss": 0.7974,
+      "step": 3658
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1083135977945834e-06,
+      "loss": 0.7935,
+      "step": 3659
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1035054794624263e-06,
+      "loss": 0.7964,
+      "step": 3660
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0987003996014107e-06,
+      "loss": 0.8018,
+      "step": 3661
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.093898360328569e-06,
+      "loss": 0.79,
+      "step": 3662
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0890993637595934e-06,
+      "loss": 0.7759,
+      "step": 3663
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.084303412008838e-06,
+      "loss": 0.8408,
+      "step": 3664
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0795105071893017e-06,
+      "loss": 0.8672,
+      "step": 3665
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0747206514126647e-06,
+      "loss": 0.7651,
+      "step": 3666
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.069933846789248e-06,
+      "loss": 0.8027,
+      "step": 3667
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0651500954280267e-06,
+      "loss": 0.8184,
+      "step": 3668
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.060369399436638e-06,
+      "loss": 0.7441,
+      "step": 3669
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.055591760921368e-06,
+      "loss": 0.7808,
+      "step": 3670
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.050817181987167e-06,
+      "loss": 0.8247,
+      "step": 3671
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.046045664737618e-06,
+      "loss": 0.7388,
+      "step": 3672
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.04127721127497e-06,
+      "loss": 0.6663,
+      "step": 3673
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0365118237001166e-06,
+      "loss": 0.7788,
+      "step": 3674
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.031749504112602e-06,
+      "loss": 0.772,
+      "step": 3675
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.02699025461062e-06,
+      "loss": 0.8252,
+      "step": 3676
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0222340772910074e-06,
+      "loss": 0.7373,
+      "step": 3677
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.017480974249255e-06,
+      "loss": 0.7812,
+      "step": 3678
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0127309475794875e-06,
+      "loss": 0.7793,
+      "step": 3679
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.007983999374481e-06,
+      "loss": 0.7681,
+      "step": 3680
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0032401317256644e-06,
+      "loss": 0.7783,
+      "step": 3681
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.998499346723092e-06,
+      "loss": 0.7617,
+      "step": 3682
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9937616464554696e-06,
+      "loss": 0.8276,
+      "step": 3683
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9890270330101447e-06,
+      "loss": 0.7812,
+      "step": 3684
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9842955084730994e-06,
+      "loss": 0.7485,
+      "step": 3685
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.979567074928962e-06,
+      "loss": 0.7646,
+      "step": 3686
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9748417344609925e-06,
+      "loss": 0.8149,
+      "step": 3687
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.970119489151094e-06,
+      "loss": 0.7524,
+      "step": 3688
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9654003410797984e-06,
+      "loss": 0.8252,
+      "step": 3689
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9606842923262746e-06,
+      "loss": 0.7646,
+      "step": 3690
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9559713449683357e-06,
+      "loss": 0.7373,
+      "step": 3691
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.951261501082423e-06,
+      "loss": 0.7515,
+      "step": 3692
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9465547627436e-06,
+      "loss": 0.793,
+      "step": 3693
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.941851132025575e-06,
+      "loss": 0.7915,
+      "step": 3694
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.937150611000684e-06,
+      "loss": 0.7412,
+      "step": 3695
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.93245320173989e-06,
+      "loss": 0.752,
+      "step": 3696
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9277589063127876e-06,
+      "loss": 0.8311,
+      "step": 3697
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9230677267875997e-06,
+      "loss": 0.7939,
+      "step": 3698
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9183796652311736e-06,
+      "loss": 0.7769,
+      "step": 3699
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9136947237089862e-06,
+      "loss": 0.8149,
+      "step": 3700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9090129042851365e-06,
+      "loss": 0.8042,
+      "step": 3701
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9043342090223513e-06,
+      "loss": 0.7798,
+      "step": 3702
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8996586399819826e-06,
+      "loss": 0.7681,
+      "step": 3703
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8949861992239947e-06,
+      "loss": 0.709,
+      "step": 3704
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.890316888806981e-06,
+      "loss": 0.7876,
+      "step": 3705
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8856507107881657e-06,
+      "loss": 0.7627,
+      "step": 3706
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8809876672233718e-06,
+      "loss": 0.7651,
+      "step": 3707
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.876327760167056e-06,
+      "loss": 0.8071,
+      "step": 3708
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8716709916722895e-06,
+      "loss": 0.8364,
+      "step": 3709
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8670173637907605e-06,
+      "loss": 0.7944,
+      "step": 3710
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.862366878572774e-06,
+      "loss": 0.769,
+      "step": 3711
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8577195380672495e-06,
+      "loss": 0.769,
+      "step": 3712
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.853075344321724e-06,
+      "loss": 0.7861,
+      "step": 3713
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.84843429938234e-06,
+      "loss": 0.7275,
+      "step": 3714
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8437964052938584e-06,
+      "loss": 0.8203,
+      "step": 3715
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8391616640996565e-06,
+      "loss": 0.8442,
+      "step": 3716
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8345300778417194e-06,
+      "loss": 0.7998,
+      "step": 3717
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8299016485606336e-06,
+      "loss": 0.8389,
+      "step": 3718
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.825276378295605e-06,
+      "loss": 0.8267,
+      "step": 3719
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.820654269084444e-06,
+      "loss": 0.7744,
+      "step": 3720
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8160353229635684e-06,
+      "loss": 0.8579,
+      "step": 3721
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8114195419680023e-06,
+      "loss": 0.8174,
+      "step": 3722
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.806806928131377e-06,
+      "loss": 0.7803,
+      "step": 3723
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8021974834859258e-06,
+      "loss": 0.7549,
+      "step": 3724
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.797591210062487e-06,
+      "loss": 0.7783,
+      "step": 3725
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.792988109890502e-06,
+      "loss": 0.8135,
+      "step": 3726
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7883881849980168e-06,
+      "loss": 0.7808,
+      "step": 3727
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7837914374116683e-06,
+      "loss": 0.7842,
+      "step": 3728
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7791978691567047e-06,
+      "loss": 0.7749,
+      "step": 3729
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.774607482256967e-06,
+      "loss": 0.7847,
+      "step": 3730
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.770020278734904e-06,
+      "loss": 0.7764,
+      "step": 3731
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.765436260611547e-06,
+      "loss": 0.7764,
+      "step": 3732
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.760855429906535e-06,
+      "loss": 0.8525,
+      "step": 3733
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.756277788638099e-06,
+      "loss": 0.7661,
+      "step": 3734
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.751703338823064e-06,
+      "loss": 0.7964,
+      "step": 3735
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.747132082476852e-06,
+      "loss": 0.8223,
+      "step": 3736
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.742564021613475e-06,
+      "loss": 0.772,
+      "step": 3737
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7379991582455434e-06,
+      "loss": 0.792,
+      "step": 3738
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7334374943842444e-06,
+      "loss": 0.7319,
+      "step": 3739
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7288790320393667e-06,
+      "loss": 0.7407,
+      "step": 3740
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.724323773219296e-06,
+      "loss": 0.8403,
+      "step": 3741
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.719771719930987e-06,
+      "loss": 0.7444,
+      "step": 3742
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.715222874179996e-06,
+      "loss": 0.749,
+      "step": 3743
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7106772379704614e-06,
+      "loss": 0.8188,
+      "step": 3744
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.706134813305111e-06,
+      "loss": 0.7666,
+      "step": 3745
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.701595602185253e-06,
+      "loss": 0.7881,
+      "step": 3746
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6970596066107825e-06,
+      "loss": 0.7612,
+      "step": 3747
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6925268285801807e-06,
+      "loss": 0.8623,
+      "step": 3748
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6879972700904987e-06,
+      "loss": 0.6992,
+      "step": 3749
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6834709331373863e-06,
+      "loss": 0.7979,
+      "step": 3750
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6789478197150654e-06,
+      "loss": 0.7729,
+      "step": 3751
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.674427931816339e-06,
+      "loss": 0.7881,
+      "step": 3752
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6699112714325827e-06,
+      "loss": 0.7915,
+      "step": 3753
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6653978405537594e-06,
+      "loss": 0.7397,
+      "step": 3754
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6608876411684047e-06,
+      "loss": 0.8237,
+      "step": 3755
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6563806752636314e-06,
+      "loss": 0.8071,
+      "step": 3756
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6518769448251267e-06,
+      "loss": 0.8354,
+      "step": 3757
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.647376451837155e-06,
+      "loss": 0.8203,
+      "step": 3758
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6428791982825497e-06,
+      "loss": 0.7456,
+      "step": 3759
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.638385186142721e-06,
+      "loss": 0.8086,
+      "step": 3760
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.63389441739765e-06,
+      "loss": 0.7769,
+      "step": 3761
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.629406894025891e-06,
+      "loss": 0.7778,
+      "step": 3762
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6249226180045606e-06,
+      "loss": 0.7646,
+      "step": 3763
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6204415913093505e-06,
+      "loss": 0.8262,
+      "step": 3764
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6159638159145205e-06,
+      "loss": 0.8013,
+      "step": 3765
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.611489293792905e-06,
+      "loss": 0.7783,
+      "step": 3766
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6070180269158884e-06,
+      "loss": 0.812,
+      "step": 3767
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6025500172534345e-06,
+      "loss": 0.8174,
+      "step": 3768
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5980852667740664e-06,
+      "loss": 0.7988,
+      "step": 3769
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5936237774448738e-06,
+      "loss": 0.7905,
+      "step": 3770
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5891655512315083e-06,
+      "loss": 0.8281,
+      "step": 3771
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.584710590098184e-06,
+      "loss": 0.7729,
+      "step": 3772
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.580258896007678e-06,
+      "loss": 0.7817,
+      "step": 3773
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5758104709213195e-06,
+      "loss": 0.8091,
+      "step": 3774
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5713653167990126e-06,
+      "loss": 0.7954,
+      "step": 3775
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.566923435599212e-06,
+      "loss": 0.7793,
+      "step": 3776
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.562484829278925e-06,
+      "loss": 0.7393,
+      "step": 3777
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5580494997937232e-06,
+      "loss": 0.7529,
+      "step": 3778
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5536174490977337e-06,
+      "loss": 0.6797,
+      "step": 3779
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5491886791436384e-06,
+      "loss": 0.8105,
+      "step": 3780
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5447631918826733e-06,
+      "loss": 0.791,
+      "step": 3781
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.540340989264628e-06,
+      "loss": 0.7515,
+      "step": 3782
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5359220732378478e-06,
+      "loss": 0.7666,
+      "step": 3783
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.531506445749219e-06,
+      "loss": 0.8159,
+      "step": 3784
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.527094108744197e-06,
+      "loss": 0.8013,
+      "step": 3785
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5226850641667745e-06,
+      "loss": 0.769,
+      "step": 3786
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5182793139594998e-06,
+      "loss": 0.7842,
+      "step": 3787
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.51387686006346e-06,
+      "loss": 0.7715,
+      "step": 3788
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.509477704418303e-06,
+      "loss": 0.7769,
+      "step": 3789
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5050818489622144e-06,
+      "loss": 0.7773,
+      "step": 3790
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5006892956319296e-06,
+      "loss": 0.8213,
+      "step": 3791
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4963000463627297e-06,
+      "loss": 0.7422,
+      "step": 3792
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4919141030884377e-06,
+      "loss": 0.8003,
+      "step": 3793
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4875314677414196e-06,
+      "loss": 0.7329,
+      "step": 3794
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4831521422525895e-06,
+      "loss": 0.7905,
+      "step": 3795
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4787761285513957e-06,
+      "loss": 0.7568,
+      "step": 3796
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.474403428565836e-06,
+      "loss": 0.8042,
+      "step": 3797
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.470034044222436e-06,
+      "loss": 0.7773,
+      "step": 3798
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.465667977446268e-06,
+      "loss": 0.7358,
+      "step": 3799
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4613052301609474e-06,
+      "loss": 0.7905,
+      "step": 3800
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4569458042886242e-06,
+      "loss": 0.7424,
+      "step": 3801
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4525897017499747e-06,
+      "loss": 0.7471,
+      "step": 3802
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4482369244642234e-06,
+      "loss": 0.793,
+      "step": 3803
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4438874743491246e-06,
+      "loss": 0.7852,
+      "step": 3804
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.439541353320969e-06,
+      "loss": 0.812,
+      "step": 3805
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4351985632945773e-06,
+      "loss": 0.7612,
+      "step": 3806
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4308591061833064e-06,
+      "loss": 0.8154,
+      "step": 3807
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4265229838990433e-06,
+      "loss": 0.7832,
+      "step": 3808
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4221901983522e-06,
+      "loss": 0.7368,
+      "step": 3809
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4178607514517292e-06,
+      "loss": 0.7617,
+      "step": 3810
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.413534645105109e-06,
+      "loss": 0.7686,
+      "step": 3811
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4092118812183375e-06,
+      "loss": 0.7734,
+      "step": 3812
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.404892461695949e-06,
+      "loss": 0.7983,
+      "step": 3813
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.400576388441003e-06,
+      "loss": 0.7964,
+      "step": 3814
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3962636633550815e-06,
+      "loss": 0.7803,
+      "step": 3815
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3919542883382964e-06,
+      "loss": 0.7371,
+      "step": 3816
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3876482652892773e-06,
+      "loss": 0.7324,
+      "step": 3817
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3833455961051866e-06,
+      "loss": 0.7803,
+      "step": 3818
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.379046282681691e-06,
+      "loss": 0.7988,
+      "step": 3819
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3747503269130013e-06,
+      "loss": 0.7749,
+      "step": 3820
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3704577306918343e-06,
+      "loss": 0.7632,
+      "step": 3821
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3661684959094335e-06,
+      "loss": 0.7896,
+      "step": 3822
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.361882624455553e-06,
+      "loss": 0.749,
+      "step": 3823
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.357600118218469e-06,
+      "loss": 0.7886,
+      "step": 3824
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3533209790849874e-06,
+      "loss": 0.7412,
+      "step": 3825
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3490452089404102e-06,
+      "loss": 0.769,
+      "step": 3826
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3447728096685685e-06,
+      "loss": 0.7568,
+      "step": 3827
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.340503783151802e-06,
+      "loss": 0.7686,
+      "step": 3828
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.336238131270968e-06,
+      "loss": 0.7754,
+      "step": 3829
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3319758559054363e-06,
+      "loss": 0.7861,
+      "step": 3830
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.327716958933088e-06,
+      "loss": 0.8652,
+      "step": 3831
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3234614422303192e-06,
+      "loss": 0.7402,
+      "step": 3832
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.319209307672029e-06,
+      "loss": 0.7852,
+      "step": 3833
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3149605571316304e-06,
+      "loss": 0.7334,
+      "step": 3834
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.310715192481051e-06,
+      "loss": 0.7944,
+      "step": 3835
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3064732155907244e-06,
+      "loss": 0.7539,
+      "step": 3836
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3022346283295814e-06,
+      "loss": 0.6738,
+      "step": 3837
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2979994325650713e-06,
+      "loss": 0.8398,
+      "step": 3838
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.293767630163143e-06,
+      "loss": 0.8008,
+      "step": 3839
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.289539222988255e-06,
+      "loss": 0.7783,
+      "step": 3840
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2853142129033656e-06,
+      "loss": 0.7234,
+      "step": 3841
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2810926017699375e-06,
+      "loss": 0.8198,
+      "step": 3842
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.276874391447941e-06,
+      "loss": 0.7246,
+      "step": 3843
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2726595837958344e-06,
+      "loss": 0.7988,
+      "step": 3844
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.268448180670594e-06,
+      "loss": 0.79,
+      "step": 3845
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2642401839276883e-06,
+      "loss": 0.7935,
+      "step": 3846
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2600355954210796e-06,
+      "loss": 0.7612,
+      "step": 3847
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2558344170032355e-06,
+      "loss": 0.7896,
+      "step": 3848
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2516366505251176e-06,
+      "loss": 0.7739,
+      "step": 3849
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.247442297836194e-06,
+      "loss": 0.7949,
+      "step": 3850
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.243251360784413e-06,
+      "loss": 0.7632,
+      "step": 3851
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.239063841216229e-06,
+      "loss": 0.8496,
+      "step": 3852
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2348797409765864e-06,
+      "loss": 0.7314,
+      "step": 3853
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.230699061908925e-06,
+      "loss": 0.7485,
+      "step": 3854
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2265218058551776e-06,
+      "loss": 0.7656,
+      "step": 3855
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.222347974655766e-06,
+      "loss": 0.7695,
+      "step": 3856
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.218177570149609e-06,
+      "loss": 0.7627,
+      "step": 3857
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.214010594174105e-06,
+      "loss": 0.7759,
+      "step": 3858
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.20984704856515e-06,
+      "loss": 0.7734,
+      "step": 3859
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2056869351571353e-06,
+      "loss": 0.792,
+      "step": 3860
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.201530255782922e-06,
+      "loss": 0.811,
+      "step": 3861
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1973770122738715e-06,
+      "loss": 0.7793,
+      "step": 3862
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.193227206459829e-06,
+      "loss": 0.8013,
+      "step": 3863
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.189080840169123e-06,
+      "loss": 0.8516,
+      "step": 3864
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.184937915228569e-06,
+      "loss": 0.7666,
+      "step": 3865
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1807984334634626e-06,
+      "loss": 0.8125,
+      "step": 3866
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1766623966975907e-06,
+      "loss": 0.811,
+      "step": 3867
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1725298067532087e-06,
+      "loss": 0.7168,
+      "step": 3868
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1684006654510616e-06,
+      "loss": 0.7866,
+      "step": 3869
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1642749746103807e-06,
+      "loss": 0.7456,
+      "step": 3870
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.160152736048872e-06,
+      "loss": 0.7876,
+      "step": 3871
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1560339515827123e-06,
+      "loss": 0.7871,
+      "step": 3872
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1519186230265663e-06,
+      "loss": 0.7588,
+      "step": 3873
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.147806752193574e-06,
+      "loss": 0.7412,
+      "step": 3874
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1436983408953526e-06,
+      "loss": 0.771,
+      "step": 3875
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.139593390941994e-06,
+      "loss": 0.7446,
+      "step": 3876
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.135491904142063e-06,
+      "loss": 0.8047,
+      "step": 3877
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1313938823026026e-06,
+      "loss": 0.834,
+      "step": 3878
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1272993272291276e-06,
+      "loss": 0.7271,
+      "step": 3879
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.123208240725624e-06,
+      "loss": 0.7974,
+      "step": 3880
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.119120624594554e-06,
+      "loss": 0.8042,
+      "step": 3881
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1150364806368416e-06,
+      "loss": 0.791,
+      "step": 3882
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1109558106518903e-06,
+      "loss": 0.8037,
+      "step": 3883
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1068786164375667e-06,
+      "loss": 0.7371,
+      "step": 3884
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1028048997902175e-06,
+      "loss": 0.7578,
+      "step": 3885
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.098734662504641e-06,
+      "loss": 0.73,
+      "step": 3886
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0946679063741115e-06,
+      "loss": 0.7979,
+      "step": 3887
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0906046331903708e-06,
+      "loss": 0.8115,
+      "step": 3888
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.086544844743622e-06,
+      "loss": 0.7866,
+      "step": 3889
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.082488542822535e-06,
+      "loss": 0.7905,
+      "step": 3890
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.078435729214243e-06,
+      "loss": 0.7568,
+      "step": 3891
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0743864057043474e-06,
+      "loss": 0.7354,
+      "step": 3892
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.070340574076899e-06,
+      "loss": 0.7324,
+      "step": 3893
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.066298236114419e-06,
+      "loss": 0.7192,
+      "step": 3894
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0622593935978974e-06,
+      "loss": 0.812,
+      "step": 3895
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0582240483067674e-06,
+      "loss": 0.771,
+      "step": 3896
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.054192202018932e-06,
+      "loss": 0.7656,
+      "step": 3897
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0501638565107486e-06,
+      "loss": 0.7532,
+      "step": 3898
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0461390135570357e-06,
+      "loss": 0.791,
+      "step": 3899
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0421176749310657e-06,
+      "loss": 0.7471,
+      "step": 3900
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0380998424045663e-06,
+      "loss": 0.7607,
+      "step": 3901
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.034085517747727e-06,
+      "loss": 0.7437,
+      "step": 3902
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.030074702729178e-06,
+      "loss": 0.7007,
+      "step": 3903
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0260673991160206e-06,
+      "loss": 0.8164,
+      "step": 3904
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0220636086737967e-06,
+      "loss": 0.7988,
+      "step": 3905
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0180633331665077e-06,
+      "loss": 0.7036,
+      "step": 3906
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0140665743565966e-06,
+      "loss": 0.7871,
+      "step": 3907
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0100733340049683e-06,
+      "loss": 0.8013,
+      "step": 3908
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0060836138709696e-06,
+      "loss": 0.8218,
+      "step": 3909
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0020974157124008e-06,
+      "loss": 0.7695,
+      "step": 3910
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9981147412855085e-06,
+      "loss": 0.856,
+      "step": 3911
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.994135592344988e-06,
+      "loss": 0.7112,
+      "step": 3912
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.990159970643979e-06,
+      "loss": 0.812,
+      "step": 3913
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9861878779340703e-06,
+      "loss": 0.8076,
+      "step": 3914
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9822193159652937e-06,
+      "loss": 0.79,
+      "step": 3915
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.978254286486129e-06,
+      "loss": 0.7505,
+      "step": 3916
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9742927912434907e-06,
+      "loss": 0.7627,
+      "step": 3917
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9703348319827452e-06,
+      "loss": 0.7222,
+      "step": 3918
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9663804104476957e-06,
+      "loss": 0.8384,
+      "step": 3919
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9624295283805973e-06,
+      "loss": 0.7847,
+      "step": 3920
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9584821875221283e-06,
+      "loss": 0.7979,
+      "step": 3921
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9545383896114188e-06,
+      "loss": 0.7383,
+      "step": 3922
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.950598136386036e-06,
+      "loss": 0.7905,
+      "step": 3923
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.946661429581983e-06,
+      "loss": 0.7603,
+      "step": 3924
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9427282709337025e-06,
+      "loss": 0.7959,
+      "step": 3925
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.938798662174075e-06,
+      "loss": 0.728,
+      "step": 3926
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9348726050344145e-06,
+      "loss": 0.7393,
+      "step": 3927
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9309501012444664e-06,
+      "loss": 0.7144,
+      "step": 3928
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.92703115253242e-06,
+      "loss": 0.7715,
+      "step": 3929
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9231157606248963e-06,
+      "loss": 0.8203,
+      "step": 3930
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.91920392724694e-06,
+      "loss": 0.7261,
+      "step": 3931
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9152956541220357e-06,
+      "loss": 0.7969,
+      "step": 3932
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9113909429721e-06,
+      "loss": 0.7842,
+      "step": 3933
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9074897955174764e-06,
+      "loss": 0.7571,
+      "step": 3934
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9035922134769425e-06,
+      "loss": 0.7285,
+      "step": 3935
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.899698198567701e-06,
+      "loss": 0.8071,
+      "step": 3936
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8958077525053863e-06,
+      "loss": 0.7778,
+      "step": 3937
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8919208770040531e-06,
+      "loss": 0.7559,
+      "step": 3938
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8880375737761958e-06,
+      "loss": 0.8184,
+      "step": 3939
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8841578445327258e-06,
+      "loss": 0.7769,
+      "step": 3940
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8802816909829825e-06,
+      "loss": 0.8496,
+      "step": 3941
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8764091148347253e-06,
+      "loss": 0.7749,
+      "step": 3942
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8725401177941449e-06,
+      "loss": 0.793,
+      "step": 3943
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8686747015658492e-06,
+      "loss": 0.7529,
+      "step": 3944
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8648128678528731e-06,
+      "loss": 0.7397,
+      "step": 3945
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8609546183566707e-06,
+      "loss": 0.7563,
+      "step": 3946
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8570999547771163e-06,
+      "loss": 0.7622,
+      "step": 3947
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8532488788125057e-06,
+      "loss": 0.7676,
+      "step": 3948
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8494013921595533e-06,
+      "loss": 0.7383,
+      "step": 3949
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8455574965133927e-06,
+      "loss": 0.7808,
+      "step": 3950
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.841717193567577e-06,
+      "loss": 0.7544,
+      "step": 3951
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8378804850140686e-06,
+      "loss": 0.751,
+      "step": 3952
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8340473725432573e-06,
+      "loss": 0.7515,
+      "step": 3953
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8302178578439389e-06,
+      "loss": 0.7734,
+      "step": 3954
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.826391942603336e-06,
+      "loss": 0.8057,
+      "step": 3955
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8225696285070704e-06,
+      "loss": 0.7734,
+      "step": 3956
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8187509172391881e-06,
+      "loss": 0.8179,
+      "step": 3957
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8149358104821435e-06,
+      "loss": 0.7705,
+      "step": 3958
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8111243099168053e-06,
+      "loss": 0.7944,
+      "step": 3959
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.80731641722245e-06,
+      "loss": 0.7739,
+      "step": 3960
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.803512134076768e-06,
+      "loss": 0.7725,
+      "step": 3961
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7997114621558597e-06,
+      "loss": 0.7769,
+      "step": 3962
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7959144031342245e-06,
+      "loss": 0.7964,
+      "step": 3963
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7921209586847865e-06,
+      "loss": 0.7559,
+      "step": 3964
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7883311304788698e-06,
+      "loss": 0.7969,
+      "step": 3965
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7845449201861975e-06,
+      "loss": 0.8086,
+      "step": 3966
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7807623294749087e-06,
+      "loss": 0.7925,
+      "step": 3967
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.776983360011545e-06,
+      "loss": 0.8457,
+      "step": 3968
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7732080134610519e-06,
+      "loss": 0.7461,
+      "step": 3969
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7694362914867803e-06,
+      "loss": 0.7983,
+      "step": 3970
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7656681957504807e-06,
+      "loss": 0.7959,
+      "step": 3971
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.761903727912312e-06,
+      "loss": 0.8413,
+      "step": 3972
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.758142889630824e-06,
+      "loss": 0.686,
+      "step": 3973
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.754385682562979e-06,
+      "loss": 0.7959,
+      "step": 3974
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7506321083641353e-06,
+      "loss": 0.7505,
+      "step": 3975
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.746882168688051e-06,
+      "loss": 0.7314,
+      "step": 3976
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.743135865186878e-06,
+      "loss": 0.7808,
+      "step": 3977
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7393931995111713e-06,
+      "loss": 0.7588,
+      "step": 3978
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.735654173309882e-06,
+      "loss": 0.7676,
+      "step": 3979
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7319187882303579e-06,
+      "loss": 0.7656,
+      "step": 3980
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7281870459183425e-06,
+      "loss": 0.7217,
+      "step": 3981
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7244589480179742e-06,
+      "loss": 0.8125,
+      "step": 3982
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7207344961717854e-06,
+      "loss": 0.7632,
+      "step": 3983
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7170136920207025e-06,
+      "loss": 0.7964,
+      "step": 3984
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7132965372040456e-06,
+      "loss": 0.8047,
+      "step": 3985
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7095830333595277e-06,
+      "loss": 0.8223,
+      "step": 3986
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7058731821232466e-06,
+      "loss": 0.8105,
+      "step": 3987
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7021669851296962e-06,
+      "loss": 0.7759,
+      "step": 3988
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6984644440117648e-06,
+      "loss": 0.7495,
+      "step": 3989
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6947655604007252e-06,
+      "loss": 0.7739,
+      "step": 3990
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6910703359262338e-06,
+      "loss": 0.7812,
+      "step": 3991
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6873787722163426e-06,
+      "loss": 0.7783,
+      "step": 3992
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6836908708974886e-06,
+      "loss": 0.7373,
+      "step": 3993
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6800066335944941e-06,
+      "loss": 0.79,
+      "step": 3994
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6763260619305666e-06,
+      "loss": 0.793,
+      "step": 3995
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.672649157527302e-06,
+      "loss": 0.7148,
+      "step": 3996
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6689759220046785e-06,
+      "loss": 0.79,
+      "step": 3997
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6653063569810502e-06,
+      "loss": 0.7896,
+      "step": 3998
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6616404640731698e-06,
+      "loss": 0.7764,
+      "step": 3999
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6579782448961622e-06,
+      "loss": 0.7407,
+      "step": 4000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6543197010635315e-06,
+      "loss": 0.79,
+      "step": 4001
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6506648341871679e-06,
+      "loss": 0.8237,
+      "step": 4002
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6470136458773412e-06,
+      "loss": 0.7891,
+      "step": 4003
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6433661377426968e-06,
+      "loss": 0.7578,
+      "step": 4004
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6397223113902627e-06,
+      "loss": 0.7314,
+      "step": 4005
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6360821684254424e-06,
+      "loss": 0.7832,
+      "step": 4006
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6324457104520198e-06,
+      "loss": 0.7744,
+      "step": 4007
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6288129390721452e-06,
+      "loss": 0.7314,
+      "step": 4008
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6251838558863597e-06,
+      "loss": 0.7539,
+      "step": 4009
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.621558462493571e-06,
+      "loss": 0.7803,
+      "step": 4010
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6179367604910612e-06,
+      "loss": 0.8071,
+      "step": 4011
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.614318751474485e-06,
+      "loss": 0.8208,
+      "step": 4012
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.610704437037871e-06,
+      "loss": 0.8086,
+      "step": 4013
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6070938187736274e-06,
+      "loss": 0.7197,
+      "step": 4014
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6034868982725215e-06,
+      "loss": 0.7217,
+      "step": 4015
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.599883677123699e-06,
+      "loss": 0.7979,
+      "step": 4016
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5962841569146747e-06,
+      "loss": 0.7588,
+      "step": 4017
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5926883392313319e-06,
+      "loss": 0.7524,
+      "step": 4018
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5890962256579234e-06,
+      "loss": 0.7573,
+      "step": 4019
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.58550781777707e-06,
+      "loss": 0.8052,
+      "step": 4020
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5819231171697602e-06,
+      "loss": 0.7573,
+      "step": 4021
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5783421254153452e-06,
+      "loss": 0.7632,
+      "step": 4022
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5747648440915453e-06,
+      "loss": 0.791,
+      "step": 4023
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5711912747744517e-06,
+      "loss": 0.7744,
+      "step": 4024
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.567621419038512e-06,
+      "loss": 0.6973,
+      "step": 4025
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5640552784565378e-06,
+      "loss": 0.6812,
+      "step": 4026
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5604928545997078e-06,
+      "loss": 0.7559,
+      "step": 4027
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5569341490375623e-06,
+      "loss": 0.791,
+      "step": 4028
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5533791633380025e-06,
+      "loss": 0.7271,
+      "step": 4029
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5498278990672911e-06,
+      "loss": 0.71,
+      "step": 4030
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5462803577900508e-06,
+      "loss": 0.8086,
+      "step": 4031
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5427365410692663e-06,
+      "loss": 0.7568,
+      "step": 4032
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5391964504662737e-06,
+      "loss": 0.7803,
+      "step": 4033
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.535660087540779e-06,
+      "loss": 0.8037,
+      "step": 4034
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5321274538508402e-06,
+      "loss": 0.7998,
+      "step": 4035
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5285985509528666e-06,
+      "loss": 0.7266,
+      "step": 4036
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5250733804016338e-06,
+      "loss": 0.731,
+      "step": 4037
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5215519437502635e-06,
+      "loss": 0.8315,
+      "step": 4038
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5180342425502447e-06,
+      "loss": 0.7783,
+      "step": 4039
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5145202783514079e-06,
+      "loss": 0.7119,
+      "step": 4040
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5110100527019432e-06,
+      "loss": 0.7676,
+      "step": 4041
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5075035671483929e-06,
+      "loss": 0.7471,
+      "step": 4042
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5040008232356518e-06,
+      "loss": 0.7832,
+      "step": 4043
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5005018225069646e-06,
+      "loss": 0.7798,
+      "step": 4044
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.497006566503929e-06,
+      "loss": 0.7954,
+      "step": 4045
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4935150567664936e-06,
+      "loss": 0.7144,
+      "step": 4046
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4900272948329509e-06,
+      "loss": 0.8188,
+      "step": 4047
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4865432822399439e-06,
+      "loss": 0.7661,
+      "step": 4048
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.483063020522475e-06,
+      "loss": 0.8169,
+      "step": 4049
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4795865112138763e-06,
+      "loss": 0.7646,
+      "step": 4050
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4761137558458382e-06,
+      "loss": 0.7588,
+      "step": 4051
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4726447559483937e-06,
+      "loss": 0.7202,
+      "step": 4052
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.469179513049921e-06,
+      "loss": 0.8213,
+      "step": 4053
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4657180286771454e-06,
+      "loss": 0.8003,
+      "step": 4054
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4622603043551331e-06,
+      "loss": 0.6855,
+      "step": 4055
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4588063416072984e-06,
+      "loss": 0.8022,
+      "step": 4056
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4553561419553885e-06,
+      "loss": 0.8066,
+      "step": 4057
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.451909706919501e-06,
+      "loss": 0.7686,
+      "step": 4058
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4484670380180776e-06,
+      "loss": 0.7563,
+      "step": 4059
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4450281367678954e-06,
+      "loss": 0.7593,
+      "step": 4060
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4415930046840675e-06,
+      "loss": 0.8003,
+      "step": 4061
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4381616432800538e-06,
+      "loss": 0.7559,
+      "step": 4062
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4347340540676513e-06,
+      "loss": 0.7959,
+      "step": 4063
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4313102385569922e-06,
+      "loss": 0.7192,
+      "step": 4064
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4278901982565484e-06,
+      "loss": 0.7422,
+      "step": 4065
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4244739346731296e-06,
+      "loss": 0.7466,
+      "step": 4066
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.421061449311878e-06,
+      "loss": 0.7349,
+      "step": 4067
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4176527436762732e-06,
+      "loss": 0.8188,
+      "step": 4068
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.414247819268131e-06,
+      "loss": 0.7285,
+      "step": 4069
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4108466775876007e-06,
+      "loss": 0.71,
+      "step": 4070
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4074493201331595e-06,
+      "loss": 0.7344,
+      "step": 4071
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.404055748401625e-06,
+      "loss": 0.7905,
+      "step": 4072
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4006659638881404e-06,
+      "loss": 0.7646,
+      "step": 4073
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3972799680861904e-06,
+      "loss": 0.8105,
+      "step": 4074
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3938977624875783e-06,
+      "loss": 0.8511,
+      "step": 4075
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.390519348582443e-06,
+      "loss": 0.7598,
+      "step": 4076
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3871447278592554e-06,
+      "loss": 0.8042,
+      "step": 4077
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3837739018048101e-06,
+      "loss": 0.7603,
+      "step": 4078
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3804068719042341e-06,
+      "loss": 0.7827,
+      "step": 4079
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3770436396409781e-06,
+      "loss": 0.8462,
+      "step": 4080
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3736842064968259e-06,
+      "loss": 0.812,
+      "step": 4081
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3703285739518779e-06,
+      "loss": 0.7686,
+      "step": 4082
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.366976743484565e-06,
+      "loss": 0.8047,
+      "step": 4083
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.36362871657165e-06,
+      "loss": 0.7891,
+      "step": 4084
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3602844946882077e-06,
+      "loss": 0.729,
+      "step": 4085
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3569440793076415e-06,
+      "loss": 0.7954,
+      "step": 4086
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3536074719016811e-06,
+      "loss": 0.7793,
+      "step": 4087
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3502746739403738e-06,
+      "loss": 0.7827,
+      "step": 4088
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3469456868920916e-06,
+      "loss": 0.7788,
+      "step": 4089
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3436205122235246e-06,
+      "loss": 0.8071,
+      "step": 4090
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.340299151399689e-06,
+      "loss": 0.7876,
+      "step": 4091
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.336981605883907e-06,
+      "loss": 0.7642,
+      "step": 4092
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3336678771378386e-06,
+      "loss": 0.7578,
+      "step": 4093
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3303579666214494e-06,
+      "loss": 0.8066,
+      "step": 4094
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3270518757930295e-06,
+      "loss": 0.7358,
+      "step": 4095
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3237496061091771e-06,
+      "loss": 0.7734,
+      "step": 4096
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.320451159024817e-06,
+      "loss": 0.7988,
+      "step": 4097
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3171565359931826e-06,
+      "loss": 0.8145,
+      "step": 4098
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3138657384658281e-06,
+      "loss": 0.751,
+      "step": 4099
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3105787678926175e-06,
+      "loss": 0.8159,
+      "step": 4100
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.307295625721733e-06,
+      "loss": 0.7798,
+      "step": 4101
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3040163133996652e-06,
+      "loss": 0.7729,
+      "step": 4102
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3007408323712223e-06,
+      "loss": 0.7773,
+      "step": 4103
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2974691840795206e-06,
+      "loss": 0.8008,
+      "step": 4104
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2942013699659928e-06,
+      "loss": 0.7593,
+      "step": 4105
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2909373914703737e-06,
+      "loss": 0.8013,
+      "step": 4106
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2876772500307156e-06,
+      "loss": 0.8179,
+      "step": 4107
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2844209470833758e-06,
+      "loss": 0.7861,
+      "step": 4108
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.281168484063029e-06,
+      "loss": 0.77,
+      "step": 4109
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2779198624026468e-06,
+      "loss": 0.791,
+      "step": 4110
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.274675083533513e-06,
+      "loss": 0.8086,
+      "step": 4111
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.271434148885221e-06,
+      "loss": 0.8091,
+      "step": 4112
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2681970598856674e-06,
+      "loss": 0.7661,
+      "step": 4113
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2649638179610547e-06,
+      "loss": 0.7305,
+      "step": 4114
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.261734424535891e-06,
+      "loss": 0.8193,
+      "step": 4115
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2585088810329915e-06,
+      "loss": 0.7925,
+      "step": 4116
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2552871888734652e-06,
+      "loss": 0.7949,
+      "step": 4117
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.25206934947674e-06,
+      "loss": 0.7603,
+      "step": 4118
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2488553642605362e-06,
+      "loss": 0.7725,
+      "step": 4119
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2456452346408731e-06,
+      "loss": 0.792,
+      "step": 4120
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2424389620320797e-06,
+      "loss": 0.7451,
+      "step": 4121
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2392365478467806e-06,
+      "loss": 0.8247,
+      "step": 4122
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2360379934959022e-06,
+      "loss": 0.7847,
+      "step": 4123
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2328433003886698e-06,
+      "loss": 0.7339,
+      "step": 4124
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2296524699326073e-06,
+      "loss": 0.7759,
+      "step": 4125
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2264655035335405e-06,
+      "loss": 0.7393,
+      "step": 4126
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2232824025955803e-06,
+      "loss": 0.7427,
+      "step": 4127
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.220103168521153e-06,
+      "loss": 0.7734,
+      "step": 4128
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2169278027109688e-06,
+      "loss": 0.8135,
+      "step": 4129
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2137563065640367e-06,
+      "loss": 0.791,
+      "step": 4130
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.210588681477659e-06,
+      "loss": 0.7363,
+      "step": 4131
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.207424928847436e-06,
+      "loss": 0.7119,
+      "step": 4132
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2042650500672581e-06,
+      "loss": 0.7681,
+      "step": 4133
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.201109046529313e-06,
+      "loss": 0.7598,
+      "step": 4134
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1979569196240781e-06,
+      "loss": 0.7405,
+      "step": 4135
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1948086707403238e-06,
+      "loss": 0.79,
+      "step": 4136
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.191664301265112e-06,
+      "loss": 0.7583,
+      "step": 4137
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.188523812583795e-06,
+      "loss": 0.8247,
+      "step": 4138
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1853872060800153e-06,
+      "loss": 0.7759,
+      "step": 4139
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1822544831357074e-06,
+      "loss": 0.7881,
+      "step": 4140
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1791256451310883e-06,
+      "loss": 0.8213,
+      "step": 4141
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1760006934446676e-06,
+      "loss": 0.7725,
+      "step": 4142
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1728796294532475e-06,
+      "loss": 0.7983,
+      "step": 4143
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1697624545319108e-06,
+      "loss": 0.7656,
+      "step": 4144
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1666491700540272e-06,
+      "loss": 0.7563,
+      "step": 4145
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1635397773912538e-06,
+      "loss": 0.812,
+      "step": 4146
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1604342779135325e-06,
+      "loss": 0.7988,
+      "step": 4147
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.157332672989092e-06,
+      "loss": 0.7788,
+      "step": 4148
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.154234963984442e-06,
+      "loss": 0.7407,
+      "step": 4149
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.151141152264379e-06,
+      "loss": 0.748,
+      "step": 4150
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1480512391919808e-06,
+      "loss": 0.752,
+      "step": 4151
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1449652261286026e-06,
+      "loss": 0.8047,
+      "step": 4152
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1418831144338915e-06,
+      "loss": 0.7764,
+      "step": 4153
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1388049054657702e-06,
+      "loss": 0.6772,
+      "step": 4154
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.135730600580439e-06,
+      "loss": 0.7744,
+      "step": 4155
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1326602011323816e-06,
+      "loss": 0.77,
+      "step": 4156
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1295937084743636e-06,
+      "loss": 0.8213,
+      "step": 4157
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1265311239574228e-06,
+      "loss": 0.7891,
+      "step": 4158
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1234724489308813e-06,
+      "loss": 0.7251,
+      "step": 4159
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1204176847423343e-06,
+      "loss": 0.77,
+      "step": 4160
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1173668327376597e-06,
+      "loss": 0.7622,
+      "step": 4161
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.114319894261e-06,
+      "loss": 0.7808,
+      "step": 4162
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1112768706547893e-06,
+      "loss": 0.748,
+      "step": 4163
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1082377632597252e-06,
+      "loss": 0.8003,
+      "step": 4164
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1052025734147865e-06,
+      "loss": 0.7554,
+      "step": 4165
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.102171302457219e-06,
+      "loss": 0.8052,
+      "step": 4166
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0991439517225467e-06,
+      "loss": 0.7788,
+      "step": 4167
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0961205225445703e-06,
+      "loss": 0.8345,
+      "step": 4168
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0931010162553534e-06,
+      "loss": 0.7451,
+      "step": 4169
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0900854341852375e-06,
+      "loss": 0.7715,
+      "step": 4170
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0870737776628327e-06,
+      "loss": 0.752,
+      "step": 4171
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0840660480150234e-06,
+      "loss": 0.8125,
+      "step": 4172
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0810622465669574e-06,
+      "loss": 0.7798,
+      "step": 4173
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0780623746420592e-06,
+      "loss": 0.7686,
+      "step": 4174
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0750664335620176e-06,
+      "loss": 0.7661,
+      "step": 4175
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0720744246467862e-06,
+      "loss": 0.8032,
+      "step": 4176
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.069086349214592e-06,
+      "loss": 0.8608,
+      "step": 4177
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0661022085819305e-06,
+      "loss": 0.8101,
+      "step": 4178
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0631220040635593e-06,
+      "loss": 0.8511,
+      "step": 4179
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0601457369724988e-06,
+      "loss": 0.8311,
+      "step": 4180
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0571734086200424e-06,
+      "loss": 0.8066,
+      "step": 4181
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0542050203157428e-06,
+      "loss": 0.772,
+      "step": 4182
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.051240573367419e-06,
+      "loss": 0.7759,
+      "step": 4183
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0482800690811523e-06,
+      "loss": 0.7407,
+      "step": 4184
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.045323508761288e-06,
+      "loss": 0.7397,
+      "step": 4185
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.042370893710436e-06,
+      "loss": 0.7417,
+      "step": 4186
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0394222252294572e-06,
+      "loss": 0.7656,
+      "step": 4187
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.036477504617489e-06,
+      "loss": 0.7598,
+      "step": 4188
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0335367331719226e-06,
+      "loss": 0.7295,
+      "step": 4189
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0305999121884047e-06,
+      "loss": 0.8267,
+      "step": 4190
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.027667042960847e-06,
+      "loss": 0.8223,
+      "step": 4191
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0247381267814183e-06,
+      "loss": 0.7461,
+      "step": 4192
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0218131649405494e-06,
+      "loss": 0.7119,
+      "step": 4193
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0188921587269229e-06,
+      "loss": 0.7695,
+      "step": 4194
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0159751094274818e-06,
+      "loss": 0.8047,
+      "step": 4195
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0130620183274254e-06,
+      "loss": 0.7676,
+      "step": 4196
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0101528867102095e-06,
+      "loss": 0.8105,
+      "step": 4197
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0072477158575455e-06,
+      "loss": 0.8247,
+      "step": 4198
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0043465070493984e-06,
+      "loss": 0.8105,
+      "step": 4199
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0014492615639905e-06,
+      "loss": 0.7495,
+      "step": 4200
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.985559806777933e-07,
+      "loss": 0.8345,
+      "step": 4201
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.956666656655324e-07,
+      "loss": 0.7593,
+      "step": 4202
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.927813178001956e-07,
+      "loss": 0.7563,
+      "step": 4203
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.898999383530072e-07,
+      "loss": 0.7612,
+      "step": 4204
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.870225285934542e-07,
+      "loss": 0.7744,
+      "step": 4205
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.841490897892713e-07,
+      "loss": 0.7852,
+      "step": 4206
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.812796232064447e-07,
+      "loss": 0.7817,
+      "step": 4207
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.784141301092086e-07,
+      "loss": 0.7808,
+      "step": 4208
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.755526117600477e-07,
+      "loss": 0.7451,
+      "step": 4209
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.726950694196968e-07,
+      "loss": 0.8071,
+      "step": 4210
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.698415043471355e-07,
+      "loss": 0.7222,
+      "step": 4211
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.669919177995913e-07,
+      "loss": 0.7827,
+      "step": 4212
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.641463110325445e-07,
+      "loss": 0.718,
+      "step": 4213
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.613046852997198e-07,
+      "loss": 0.7324,
+      "step": 4214
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.584670418530805e-07,
+      "loss": 0.7026,
+      "step": 4215
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.556333819428443e-07,
+      "loss": 0.7822,
+      "step": 4216
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.5280370681747e-07,
+      "loss": 0.769,
+      "step": 4217
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.499780177236628e-07,
+      "loss": 0.7881,
+      "step": 4218
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.47156315906369e-07,
+      "loss": 0.7373,
+      "step": 4219
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.443386026087786e-07,
+      "loss": 0.7646,
+      "step": 4220
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.415248790723274e-07,
+      "loss": 0.7607,
+      "step": 4221
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.387151465366906e-07,
+      "loss": 0.7759,
+      "step": 4222
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.359094062397855e-07,
+      "loss": 0.8242,
+      "step": 4223
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.331076594177724e-07,
+      "loss": 0.7168,
+      "step": 4224
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.30309907305047e-07,
+      "loss": 0.7388,
+      "step": 4225
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.275161511342501e-07,
+      "loss": 0.8105,
+      "step": 4226
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.247263921362581e-07,
+      "loss": 0.8208,
+      "step": 4227
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.219406315401957e-07,
+      "loss": 0.8564,
+      "step": 4228
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.191588705734111e-07,
+      "loss": 0.7974,
+      "step": 4229
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.163811104615006e-07,
+      "loss": 0.7974,
+      "step": 4230
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.136073524282951e-07,
+      "loss": 0.7803,
+      "step": 4231
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.108375976958639e-07,
+      "loss": 0.751,
+      "step": 4232
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.080718474845085e-07,
+      "loss": 0.791,
+      "step": 4233
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.053101030127709e-07,
+      "loss": 0.7344,
+      "step": 4234
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.025523654974266e-07,
+      "loss": 0.71,
+      "step": 4235
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.997986361534805e-07,
+      "loss": 0.8018,
+      "step": 4236
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.970489161941786e-07,
+      "loss": 0.7827,
+      "step": 4237
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.943032068310009e-07,
+      "loss": 0.7905,
+      "step": 4238
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.915615092736529e-07,
+      "loss": 0.729,
+      "step": 4239
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.888238247300796e-07,
+      "loss": 0.7378,
+      "step": 4240
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.860901544064538e-07,
+      "loss": 0.75,
+      "step": 4241
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.833604995071832e-07,
+      "loss": 0.7246,
+      "step": 4242
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.806348612349036e-07,
+      "loss": 0.8032,
+      "step": 4243
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.779132407904822e-07,
+      "loss": 0.7915,
+      "step": 4244
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.751956393730177e-07,
+      "loss": 0.7749,
+      "step": 4245
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.724820581798332e-07,
+      "loss": 0.7832,
+      "step": 4246
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.697724984064826e-07,
+      "loss": 0.7622,
+      "step": 4247
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.67066961246753e-07,
+      "loss": 0.7759,
+      "step": 4248
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.643654478926556e-07,
+      "loss": 0.813,
+      "step": 4249
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.616679595344258e-07,
+      "loss": 0.811,
+      "step": 4250
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.589744973605285e-07,
+      "loss": 0.7861,
+      "step": 4251
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.562850625576546e-07,
+      "loss": 0.7617,
+      "step": 4252
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.535996563107219e-07,
+      "loss": 0.7964,
+      "step": 4253
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.509182798028703e-07,
+      "loss": 0.7983,
+      "step": 4254
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.482409342154685e-07,
+      "loss": 0.7983,
+      "step": 4255
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.455676207281039e-07,
+      "loss": 0.7471,
+      "step": 4256
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.42898340518592e-07,
+      "loss": 0.7949,
+      "step": 4257
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.402330947629688e-07,
+      "loss": 0.7466,
+      "step": 4258
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.375718846354953e-07,
+      "loss": 0.7559,
+      "step": 4259
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.349147113086487e-07,
+      "loss": 0.7749,
+      "step": 4260
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.32261575953135e-07,
+      "loss": 0.7405,
+      "step": 4261
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.296124797378747e-07,
+      "loss": 0.7808,
+      "step": 4262
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.269674238300174e-07,
+      "loss": 0.7271,
+      "step": 4263
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.243264093949222e-07,
+      "loss": 0.7695,
+      "step": 4264
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.216894375961737e-07,
+      "loss": 0.688,
+      "step": 4265
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.190565095955749e-07,
+      "loss": 0.7129,
+      "step": 4266
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.164276265531446e-07,
+      "loss": 0.8271,
+      "step": 4267
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.138027896271228e-07,
+      "loss": 0.8066,
+      "step": 4268
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.111819999739656e-07,
+      "loss": 0.8052,
+      "step": 4269
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.085652587483461e-07,
+      "loss": 0.8062,
+      "step": 4270
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.059525671031509e-07,
+      "loss": 0.7632,
+      "step": 4271
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.033439261894849e-07,
+      "loss": 0.7607,
+      "step": 4272
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.007393371566729e-07,
+      "loss": 0.7715,
+      "step": 4273
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.98138801152244e-07,
+      "loss": 0.7603,
+      "step": 4274
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.955423193219503e-07,
+      "loss": 0.731,
+      "step": 4275
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.929498928097545e-07,
+      "loss": 0.7827,
+      "step": 4276
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.903615227578331e-07,
+      "loss": 0.7632,
+      "step": 4277
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.877772103065762e-07,
+      "loss": 0.7881,
+      "step": 4278
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.851969565945827e-07,
+      "loss": 0.79,
+      "step": 4279
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.826207627586701e-07,
+      "loss": 0.7998,
+      "step": 4280
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.800486299338561e-07,
+      "loss": 0.8047,
+      "step": 4281
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.77480559253383e-07,
+      "loss": 0.7808,
+      "step": 4282
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.749165518486934e-07,
+      "loss": 0.7378,
+      "step": 4283
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.723566088494449e-07,
+      "loss": 0.7637,
+      "step": 4284
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.698007313834977e-07,
+      "loss": 0.7983,
+      "step": 4285
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.672489205769285e-07,
+      "loss": 0.7534,
+      "step": 4286
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.647011775540192e-07,
+      "loss": 0.8135,
+      "step": 4287
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.621575034372586e-07,
+      "loss": 0.8179,
+      "step": 4288
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.596178993473436e-07,
+      "loss": 0.7544,
+      "step": 4289
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.57082366403179e-07,
+      "loss": 0.8501,
+      "step": 4290
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.545509057218736e-07,
+      "loss": 0.7734,
+      "step": 4291
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.52023518418743e-07,
+      "loss": 0.7266,
+      "step": 4292
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.495002056073109e-07,
+      "loss": 0.8164,
+      "step": 4293
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.469809683993034e-07,
+      "loss": 0.7656,
+      "step": 4294
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.444658079046474e-07,
+      "loss": 0.7896,
+      "step": 4295
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.419547252314796e-07,
+      "loss": 0.8262,
+      "step": 4296
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.39447721486135e-07,
+      "loss": 0.7524,
+      "step": 4297
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.369447977731614e-07,
+      "loss": 0.7202,
+      "step": 4298
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.344459551952954e-07,
+      "loss": 0.728,
+      "step": 4299
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.31951194853483e-07,
+      "loss": 0.769,
+      "step": 4300
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.294605178468727e-07,
+      "loss": 0.7397,
+      "step": 4301
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.269739252728091e-07,
+      "loss": 0.7642,
+      "step": 4302
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.244914182268425e-07,
+      "loss": 0.7656,
+      "step": 4303
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.220129978027202e-07,
+      "loss": 0.7925,
+      "step": 4304
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.195386650923908e-07,
+      "loss": 0.7544,
+      "step": 4305
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.170684211859947e-07,
+      "loss": 0.7627,
+      "step": 4306
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.146022671718833e-07,
+      "loss": 0.7422,
+      "step": 4307
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.121402041365999e-07,
+      "loss": 0.7568,
+      "step": 4308
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.0968223316488e-07,
+      "loss": 0.7661,
+      "step": 4309
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.072283553396653e-07,
+      "loss": 0.834,
+      "step": 4310
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.047785717420875e-07,
+      "loss": 0.7744,
+      "step": 4311
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.023328834514786e-07,
+      "loss": 0.7031,
+      "step": 4312
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.998912915453659e-07,
+      "loss": 0.7612,
+      "step": 4313
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.974537970994677e-07,
+      "loss": 0.7607,
+      "step": 4314
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.950204011877038e-07,
+      "loss": 0.7446,
+      "step": 4315
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.925911048821798e-07,
+      "loss": 0.7695,
+      "step": 4316
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.901659092532043e-07,
+      "loss": 0.8286,
+      "step": 4317
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.877448153692734e-07,
+      "loss": 0.79,
+      "step": 4318
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.853278242970784e-07,
+      "loss": 0.7847,
+      "step": 4319
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.829149371014987e-07,
+      "loss": 0.7632,
+      "step": 4320
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.805061548456104e-07,
+      "loss": 0.8052,
+      "step": 4321
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.781014785906815e-07,
+      "loss": 0.7871,
+      "step": 4322
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.757009093961664e-07,
+      "loss": 0.8091,
+      "step": 4323
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.733044483197148e-07,
+      "loss": 0.7881,
+      "step": 4324
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.709120964171623e-07,
+      "loss": 0.7444,
+      "step": 4325
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.685238547425376e-07,
+      "loss": 0.7837,
+      "step": 4326
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.66139724348055e-07,
+      "loss": 0.7847,
+      "step": 4327
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.637597062841217e-07,
+      "loss": 0.8369,
+      "step": 4328
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.613838015993312e-07,
+      "loss": 0.771,
+      "step": 4329
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.590120113404608e-07,
+      "loss": 0.7881,
+      "step": 4330
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.566443365524788e-07,
+      "loss": 0.8145,
+      "step": 4331
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.542807782785432e-07,
+      "loss": 0.7178,
+      "step": 4332
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.519213375599954e-07,
+      "loss": 0.7671,
+      "step": 4333
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.495660154363592e-07,
+      "loss": 0.7627,
+      "step": 4334
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.472148129453481e-07,
+      "loss": 0.8159,
+      "step": 4335
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.448677311228601e-07,
+      "loss": 0.8135,
+      "step": 4336
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.42524771002977e-07,
+      "loss": 0.7432,
+      "step": 4337
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.401859336179639e-07,
+      "loss": 0.7065,
+      "step": 4338
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.378512199982723e-07,
+      "loss": 0.7891,
+      "step": 4339
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.355206311725359e-07,
+      "loss": 0.7642,
+      "step": 4340
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.331941681675635e-07,
+      "loss": 0.7358,
+      "step": 4341
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.308718320083595e-07,
+      "loss": 0.7891,
+      "step": 4342
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.285536237181033e-07,
+      "loss": 0.7417,
+      "step": 4343
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.262395443181513e-07,
+      "loss": 0.8218,
+      "step": 4344
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.239295948280477e-07,
+      "loss": 0.812,
+      "step": 4345
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.216237762655142e-07,
+      "loss": 0.7617,
+      "step": 4346
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.193220896464536e-07,
+      "loss": 0.7332,
+      "step": 4347
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.170245359849458e-07,
+      "loss": 0.7402,
+      "step": 4348
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.147311162932534e-07,
+      "loss": 0.8291,
+      "step": 4349
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.124418315818159e-07,
+      "loss": 0.8022,
+      "step": 4350
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.101566828592476e-07,
+      "loss": 0.7598,
+      "step": 4351
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.078756711323475e-07,
+      "loss": 0.8203,
+      "step": 4352
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.055987974060873e-07,
+      "loss": 0.8071,
+      "step": 4353
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.033260626836168e-07,
+      "loss": 0.7668,
+      "step": 4354
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.010574679662606e-07,
+      "loss": 0.7476,
+      "step": 4355
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.987930142535192e-07,
+      "loss": 0.7729,
+      "step": 4356
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.965327025430756e-07,
+      "loss": 0.7749,
+      "step": 4357
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.942765338307776e-07,
+      "loss": 0.7471,
+      "step": 4358
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.920245091106536e-07,
+      "loss": 0.8003,
+      "step": 4359
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.897766293749052e-07,
+      "loss": 0.7212,
+      "step": 4360
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.875328956139082e-07,
+      "loss": 0.8149,
+      "step": 4361
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.852933088162105e-07,
+      "loss": 0.7842,
+      "step": 4362
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.83057869968534e-07,
+      "loss": 0.7383,
+      "step": 4363
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.808265800557744e-07,
+      "loss": 0.8306,
+      "step": 4364
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.78599440060994e-07,
+      "loss": 0.8047,
+      "step": 4365
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.763764509654324e-07,
+      "loss": 0.7832,
+      "step": 4366
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.741576137484994e-07,
+      "loss": 0.7798,
+      "step": 4367
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.719429293877755e-07,
+      "loss": 0.7593,
+      "step": 4368
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.697323988590086e-07,
+      "loss": 0.7695,
+      "step": 4369
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.675260231361191e-07,
+      "loss": 0.7808,
+      "step": 4370
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.653238031911956e-07,
+      "loss": 0.7549,
+      "step": 4371
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.631257399944989e-07,
+      "loss": 0.7798,
+      "step": 4372
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.609318345144533e-07,
+      "loss": 0.769,
+      "step": 4373
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.587420877176563e-07,
+      "loss": 0.7527,
+      "step": 4374
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.5655650056887e-07,
+      "loss": 0.7866,
+      "step": 4375
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.543750740310227e-07,
+      "loss": 0.7886,
+      "step": 4376
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.521978090652147e-07,
+      "loss": 0.793,
+      "step": 4377
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.500247066307096e-07,
+      "loss": 0.7598,
+      "step": 4378
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.478557676849339e-07,
+      "loss": 0.7095,
+      "step": 4379
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.456909931834853e-07,
+      "loss": 0.7866,
+      "step": 4380
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.435303840801231e-07,
+      "loss": 0.7217,
+      "step": 4381
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.413739413267771e-07,
+      "loss": 0.8008,
+      "step": 4382
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.392216658735316e-07,
+      "loss": 0.7944,
+      "step": 4383
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.370735586686426e-07,
+      "loss": 0.7759,
+      "step": 4384
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.349296206585275e-07,
+      "loss": 0.7637,
+      "step": 4385
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.327898527877662e-07,
+      "loss": 0.7817,
+      "step": 4386
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.306542559991035e-07,
+      "loss": 0.7822,
+      "step": 4387
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.285228312334423e-07,
+      "loss": 0.7871,
+      "step": 4388
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.263955794298525e-07,
+      "loss": 0.7456,
+      "step": 4389
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.242725015255601e-07,
+      "loss": 0.7842,
+      "step": 4390
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.221535984559556e-07,
+      "loss": 0.8066,
+      "step": 4391
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.200388711545934e-07,
+      "loss": 0.7407,
+      "step": 4392
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.179283205531793e-07,
+      "loss": 0.771,
+      "step": 4393
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.158219475815851e-07,
+      "loss": 0.7568,
+      "step": 4394
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.137197531678417e-07,
+      "loss": 0.8438,
+      "step": 4395
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.116217382381373e-07,
+      "loss": 0.7632,
+      "step": 4396
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.095279037168211e-07,
+      "loss": 0.7817,
+      "step": 4397
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.074382505263964e-07,
+      "loss": 0.8081,
+      "step": 4398
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.05352779587529e-07,
+      "loss": 0.7886,
+      "step": 4399
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.032714918190384e-07,
+      "loss": 0.8369,
+      "step": 4400
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.011943881379e-07,
+      "loss": 0.7451,
+      "step": 4401
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.991214694592539e-07,
+      "loss": 0.8438,
+      "step": 4402
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.970527366963884e-07,
+      "loss": 0.7397,
+      "step": 4403
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.949881907607468e-07,
+      "loss": 0.7402,
+      "step": 4404
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.929278325619347e-07,
+      "loss": 0.7939,
+      "step": 4405
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.908716630077071e-07,
+      "loss": 0.793,
+      "step": 4406
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.888196830039748e-07,
+      "loss": 0.8076,
+      "step": 4407
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.867718934548038e-07,
+      "loss": 0.7798,
+      "step": 4408
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.847282952624143e-07,
+      "loss": 0.8003,
+      "step": 4409
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.826888893271764e-07,
+      "loss": 0.8135,
+      "step": 4410
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.806536765476177e-07,
+      "loss": 0.8286,
+      "step": 4411
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.786226578204157e-07,
+      "loss": 0.7378,
+      "step": 4412
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.76595834040402e-07,
+      "loss": 0.7371,
+      "step": 4413
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.745732061005548e-07,
+      "loss": 0.7593,
+      "step": 4414
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.725547748920112e-07,
+      "loss": 0.7695,
+      "step": 4415
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7054054130405204e-07,
+      "loss": 0.7422,
+      "step": 4416
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6853050622411635e-07,
+      "loss": 0.7915,
+      "step": 4417
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6652467053778685e-07,
+      "loss": 0.7783,
+      "step": 4418
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.645230351287999e-07,
+      "loss": 0.8701,
+      "step": 4419
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6252560087903795e-07,
+      "loss": 0.8359,
+      "step": 4420
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6053236866853703e-07,
+      "loss": 0.8491,
+      "step": 4421
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.585433393754779e-07,
+      "loss": 0.7388,
+      "step": 4422
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5655851387619075e-07,
+      "loss": 0.7729,
+      "step": 4423
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.545778930451561e-07,
+      "loss": 0.8027,
+      "step": 4424
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5260147775499586e-07,
+      "loss": 0.7202,
+      "step": 4425
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.506292688764846e-07,
+      "loss": 0.7437,
+      "step": 4426
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.486612672785451e-07,
+      "loss": 0.793,
+      "step": 4427
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4669747382823923e-07,
+      "loss": 0.811,
+      "step": 4428
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.447378893907817e-07,
+      "loss": 0.7153,
+      "step": 4429
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.427825148295273e-07,
+      "loss": 0.8281,
+      "step": 4430
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.408313510059814e-07,
+      "loss": 0.7407,
+      "step": 4431
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3888439877979086e-07,
+      "loss": 0.7935,
+      "step": 4432
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.369416590087472e-07,
+      "loss": 0.8325,
+      "step": 4433
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.35003132548788e-07,
+      "loss": 0.7529,
+      "step": 4434
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.330688202539901e-07,
+      "loss": 0.855,
+      "step": 4435
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3113872297657955e-07,
+      "loss": 0.7915,
+      "step": 4436
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.292128415669206e-07,
+      "loss": 0.7632,
+      "step": 4437
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.272911768735244e-07,
+      "loss": 0.7798,
+      "step": 4438
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2537372974303824e-07,
+      "loss": 0.7715,
+      "step": 4439
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.234605010202575e-07,
+      "loss": 0.749,
+      "step": 4440
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2155149154811446e-07,
+      "loss": 0.7388,
+      "step": 4441
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1964670216768534e-07,
+      "loss": 0.7847,
+      "step": 4442
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.177461337181854e-07,
+      "loss": 0.8188,
+      "step": 4443
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1584978703697154e-07,
+      "loss": 0.7339,
+      "step": 4444
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1395766295953875e-07,
+      "loss": 0.7881,
+      "step": 4445
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1206976231952464e-07,
+      "loss": 0.7417,
+      "step": 4446
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1018608594870277e-07,
+      "loss": 0.7764,
+      "step": 4447
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0830663467698817e-07,
+      "loss": 0.771,
+      "step": 4448
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.064314093324306e-07,
+      "loss": 0.7544,
+      "step": 4449
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.045604107412238e-07,
+      "loss": 0.7476,
+      "step": 4450
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.026936397276926e-07,
+      "loss": 0.7812,
+      "step": 4451
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.00831097114307e-07,
+      "loss": 0.7573,
+      "step": 4452
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9897278372166505e-07,
+      "loss": 0.7827,
+      "step": 4453
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.971187003685095e-07,
+      "loss": 0.7593,
+      "step": 4454
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.952688478717148e-07,
+      "loss": 0.7773,
+      "step": 4455
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.934232270462923e-07,
+      "loss": 0.8296,
+      "step": 4456
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9158183870538936e-07,
+      "loss": 0.7754,
+      "step": 4457
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8974468366028917e-07,
+      "loss": 0.793,
+      "step": 4458
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8791176272040986e-07,
+      "loss": 0.749,
+      "step": 4459
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.860830766932999e-07,
+      "loss": 0.8423,
+      "step": 4460
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8425862638464816e-07,
+      "loss": 0.7969,
+      "step": 4461
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.824384125982761e-07,
+      "loss": 0.8105,
+      "step": 4462
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8062243613613437e-07,
+      "loss": 0.7612,
+      "step": 4463
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7881069779830857e-07,
+      "loss": 0.7212,
+      "step": 4464
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7700319838302026e-07,
+      "loss": 0.7432,
+      "step": 4465
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7519993868662117e-07,
+      "loss": 0.7983,
+      "step": 4466
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.734009195035926e-07,
+      "loss": 0.7964,
+      "step": 4467
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.716061416265526e-07,
+      "loss": 0.7261,
+      "step": 4468
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6981560584624765e-07,
+      "loss": 0.7764,
+      "step": 4469
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.680293129515533e-07,
+      "loss": 0.7827,
+      "step": 4470
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6624726372948003e-07,
+      "loss": 0.8262,
+      "step": 4471
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.644694589651654e-07,
+      "loss": 0.7241,
+      "step": 4472
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6269589944188056e-07,
+      "loss": 0.7588,
+      "step": 4473
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6092658594102157e-07,
+      "loss": 0.7319,
+      "step": 4474
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5916151924211606e-07,
+      "loss": 0.7998,
+      "step": 4475
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.574007001228208e-07,
+      "loss": 0.7578,
+      "step": 4476
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.556441293589208e-07,
+      "loss": 0.7651,
+      "step": 4477
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.538918077243303e-07,
+      "loss": 0.7637,
+      "step": 4478
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.521437359910906e-07,
+      "loss": 0.7686,
+      "step": 4479
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.503999149293702e-07,
+      "loss": 0.7554,
+      "step": 4480
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4866034530746553e-07,
+      "loss": 0.7593,
+      "step": 4481
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4692502789179796e-07,
+      "loss": 0.7573,
+      "step": 4482
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.451939634469215e-07,
+      "loss": 0.772,
+      "step": 4483
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.43467152735506e-07,
+      "loss": 0.6865,
+      "step": 4484
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4174459651835524e-07,
+      "loss": 0.7485,
+      "step": 4485
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.400262955543987e-07,
+      "loss": 0.7456,
+      "step": 4486
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3831225060068863e-07,
+      "loss": 0.7944,
+      "step": 4487
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.366024624123987e-07,
+      "loss": 0.8291,
+      "step": 4488
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3489693174283413e-07,
+      "loss": 0.7871,
+      "step": 4489
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3319565934341937e-07,
+      "loss": 0.7876,
+      "step": 4490
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3149864596370483e-07,
+      "loss": 0.749,
+      "step": 4491
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.298058923513647e-07,
+      "loss": 0.7476,
+      "step": 4492
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.281173992521947e-07,
+      "loss": 0.7842,
+      "step": 4493
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.264331674101151e-07,
+      "loss": 0.7988,
+      "step": 4494
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2475319756716585e-07,
+      "loss": 0.7832,
+      "step": 4495
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2307749046351475e-07,
+      "loss": 0.7954,
+      "step": 4496
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2140604683744805e-07,
+      "loss": 0.7197,
+      "step": 4497
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.19738867425371e-07,
+      "loss": 0.769,
+      "step": 4498
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1807595296181383e-07,
+      "loss": 0.7759,
+      "step": 4499
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.164173041794272e-07,
+      "loss": 0.8462,
+      "step": 4500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1476292180898206e-07,
+      "loss": 0.7671,
+      "step": 4501
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.131128065793676e-07,
+      "loss": 0.7593,
+      "step": 4502
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1146695921759784e-07,
+      "loss": 0.7656,
+      "step": 4503
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.098253804488016e-07,
+      "loss": 0.7642,
+      "step": 4504
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0818807099622816e-07,
+      "loss": 0.7686,
+      "step": 4505
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.065550315812471e-07,
+      "loss": 0.792,
+      "step": 4506
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.049262629233474e-07,
+      "loss": 0.7134,
+      "step": 4507
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0330176574013603e-07,
+      "loss": 0.8188,
+      "step": 4508
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.016815407473339e-07,
+      "loss": 0.7549,
+      "step": 4509
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.000655886587833e-07,
+      "loss": 0.7617,
+      "step": 4510
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9845391018644696e-07,
+      "loss": 0.7671,
+      "step": 4511
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.96846506040398e-07,
+      "loss": 0.7998,
+      "step": 4512
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.95243376928831e-07,
+      "loss": 0.7812,
+      "step": 4513
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.936445235580554e-07,
+      "loss": 0.7881,
+      "step": 4514
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9204994663249664e-07,
+      "loss": 0.7681,
+      "step": 4515
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.904596468546983e-07,
+      "loss": 0.7192,
+      "step": 4516
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8887362492531656e-07,
+      "loss": 0.7363,
+      "step": 4517
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.872918815431258e-07,
+      "loss": 0.7246,
+      "step": 4518
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8571441740501083e-07,
+      "loss": 0.7817,
+      "step": 4519
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.841412332059734e-07,
+      "loss": 0.7593,
+      "step": 4520
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8257232963913363e-07,
+      "loss": 0.7568,
+      "step": 4521
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.810077073957218e-07,
+      "loss": 0.7764,
+      "step": 4522
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.794473671650799e-07,
+      "loss": 0.7578,
+      "step": 4523
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.778913096346658e-07,
+      "loss": 0.7715,
+      "step": 4524
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.763395354900522e-07,
+      "loss": 0.8442,
+      "step": 4525
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.747920454149222e-07,
+      "loss": 0.8267,
+      "step": 4526
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7324884009107046e-07,
+      "loss": 0.7876,
+      "step": 4527
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.717099201984075e-07,
+      "loss": 0.7837,
+      "step": 4528
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.701752864149543e-07,
+      "loss": 0.7979,
+      "step": 4529
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.686449394168378e-07,
+      "loss": 0.7446,
+      "step": 4530
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6711887987830756e-07,
+      "loss": 0.7241,
+      "step": 4531
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.655971084717157e-07,
+      "loss": 0.7539,
+      "step": 4532
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.64079625867526e-07,
+      "loss": 0.7676,
+      "step": 4533
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.625664327343136e-07,
+      "loss": 0.7451,
+      "step": 4534
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.610575297387652e-07,
+      "loss": 0.7896,
+      "step": 4535
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.59552917545679e-07,
+      "loss": 0.7925,
+      "step": 4536
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.580525968179548e-07,
+      "loss": 0.7451,
+      "step": 4537
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5655656821661047e-07,
+      "loss": 0.7705,
+      "step": 4538
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5506483240076873e-07,
+      "loss": 0.7026,
+      "step": 4539
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5357739002765947e-07,
+      "loss": 0.8135,
+      "step": 4540
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.520942417526251e-07,
+      "loss": 0.772,
+      "step": 4541
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.506153882291129e-07,
+      "loss": 0.8052,
+      "step": 4542
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4914083010868064e-07,
+      "loss": 0.7896,
+      "step": 4543
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4767056804098966e-07,
+      "loss": 0.7407,
+      "step": 4544
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4620460267381184e-07,
+      "loss": 0.7466,
+      "step": 4545
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.447429346530272e-07,
+      "loss": 0.8042,
+      "step": 4546
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4328556462261734e-07,
+      "loss": 0.7891,
+      "step": 4547
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4183249322467404e-07,
+      "loss": 0.7495,
+      "step": 4548
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4038372109939533e-07,
+      "loss": 0.7705,
+      "step": 4549
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.389392488850828e-07,
+      "loss": 0.7378,
+      "step": 4550
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3749907721814626e-07,
+      "loss": 0.8027,
+      "step": 4551
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3606320673309812e-07,
+      "loss": 0.814,
+      "step": 4552
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3463163806255906e-07,
+      "loss": 0.8276,
+      "step": 4553
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3320437183725008e-07,
+      "loss": 0.7212,
+      "step": 4554
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.317814086859982e-07,
+      "loss": 0.7578,
+      "step": 4555
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3036274923573742e-07,
+      "loss": 0.7812,
+      "step": 4556
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.289483941115056e-07,
+      "loss": 0.7788,
+      "step": 4557
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.275383439364376e-07,
+      "loss": 0.7524,
+      "step": 4558
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2613259933177757e-07,
+      "loss": 0.7627,
+      "step": 4559
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.247311609168723e-07,
+      "loss": 0.7896,
+      "step": 4560
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2333402930916904e-07,
+      "loss": 0.8198,
+      "step": 4561
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2194120512421978e-07,
+      "loss": 0.7778,
+      "step": 4562
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2055268897567818e-07,
+      "loss": 0.7856,
+      "step": 4563
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1916848147529924e-07,
+      "loss": 0.7759,
+      "step": 4564
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1778858323293627e-07,
+      "loss": 0.749,
+      "step": 4565
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1641299485655188e-07,
+      "loss": 0.8027,
+      "step": 4566
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1504171695220567e-07,
+      "loss": 0.7112,
+      "step": 4567
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1367475012405547e-07,
+      "loss": 0.749,
+      "step": 4568
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1231209497436288e-07,
+      "loss": 0.7729,
+      "step": 4569
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.109537521034888e-07,
+      "loss": 0.7056,
+      "step": 4570
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0959972210989664e-07,
+      "loss": 0.7515,
+      "step": 4571
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0825000559014595e-07,
+      "loss": 0.7466,
+      "step": 4572
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.069046031388966e-07,
+      "loss": 0.7148,
+      "step": 4573
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0556351534891107e-07,
+      "loss": 0.7964,
+      "step": 4574
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.042267428110467e-07,
+      "loss": 0.7837,
+      "step": 4575
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0289428611426244e-07,
+      "loss": 0.7729,
+      "step": 4576
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0156614584561307e-07,
+      "loss": 0.772,
+      "step": 4577
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.002423225902561e-07,
+      "loss": 0.7471,
+      "step": 4578
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9892281693144167e-07,
+      "loss": 0.791,
+      "step": 4579
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9760762945051915e-07,
+      "loss": 0.7515,
+      "step": 4580
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.962967607269406e-07,
+      "loss": 0.771,
+      "step": 4581
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9499021133824847e-07,
+      "loss": 0.7446,
+      "step": 4582
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.936879818600834e-07,
+      "loss": 0.7861,
+      "step": 4583
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.923900728661876e-07,
+      "loss": 0.7568,
+      "step": 4584
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.910964849283936e-07,
+      "loss": 0.7817,
+      "step": 4585
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8980721861663554e-07,
+      "loss": 0.7441,
+      "step": 4586
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.885222744989379e-07,
+      "loss": 0.8086,
+      "step": 4587
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8724165314142672e-07,
+      "loss": 0.853,
+      "step": 4588
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8596535510831958e-07,
+      "loss": 0.7295,
+      "step": 4589
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.846933809619278e-07,
+      "loss": 0.7729,
+      "step": 4590
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8342573126266417e-07,
+      "loss": 0.75,
+      "step": 4591
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8216240656903195e-07,
+      "loss": 0.7769,
+      "step": 4592
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.809034074376259e-07,
+      "loss": 0.7451,
+      "step": 4593
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7964873442314233e-07,
+      "loss": 0.7612,
+      "step": 4594
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7839838807836461e-07,
+      "loss": 0.8086,
+      "step": 4595
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.771523689541732e-07,
+      "loss": 0.7632,
+      "step": 4596
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7591067759954338e-07,
+      "loss": 0.7783,
+      "step": 4597
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.746733145615409e-07,
+      "loss": 0.7988,
+      "step": 4598
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7344028038532522e-07,
+      "loss": 0.8066,
+      "step": 4599
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7221157561414958e-07,
+      "loss": 0.748,
+      "step": 4600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7098720078935872e-07,
+      "loss": 0.79,
+      "step": 4601
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6976715645039222e-07,
+      "loss": 0.7788,
+      "step": 4602
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6855144313477568e-07,
+      "loss": 0.7305,
+      "step": 4603
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6734006137813284e-07,
+      "loss": 0.7456,
+      "step": 4604
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6613301171417573e-07,
+      "loss": 0.792,
+      "step": 4605
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.649302946747111e-07,
+      "loss": 0.7559,
+      "step": 4606
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6373191078963068e-07,
+      "loss": 0.7715,
+      "step": 4607
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6253786058692323e-07,
+      "loss": 0.7393,
+      "step": 4608
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.613481445926657e-07,
+      "loss": 0.7812,
+      "step": 4609
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6016276333102542e-07,
+      "loss": 0.7944,
+      "step": 4610
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5898171732425805e-07,
+      "loss": 0.8042,
+      "step": 4611
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5780500709271395e-07,
+      "loss": 0.7549,
+      "step": 4612
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5663263315482957e-07,
+      "loss": 0.7329,
+      "step": 4613
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.554645960271306e-07,
+      "loss": 0.8208,
+      "step": 4614
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5430089622423317e-07,
+      "loss": 0.7583,
+      "step": 4615
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5314153425884492e-07,
+      "loss": 0.853,
+      "step": 4616
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5198651064175728e-07,
+      "loss": 0.7505,
+      "step": 4617
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.508358258818532e-07,
+      "loss": 0.7383,
+      "step": 4618
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4968948048610488e-07,
+      "loss": 0.7729,
+      "step": 4619
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4854747495956944e-07,
+      "loss": 0.7549,
+      "step": 4620
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4740980980539555e-07,
+      "loss": 0.7476,
+      "step": 4621
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4627648552481777e-07,
+      "loss": 0.7578,
+      "step": 4622
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4514750261715894e-07,
+      "loss": 0.7812,
+      "step": 4623
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.440228615798256e-07,
+      "loss": 0.771,
+      "step": 4624
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4290256290831805e-07,
+      "loss": 0.7915,
+      "step": 4625
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4178660709621707e-07,
+      "loss": 0.7678,
+      "step": 4626
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4067499463519486e-07,
+      "loss": 0.8262,
+      "step": 4627
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3956772601500633e-07,
+      "loss": 0.8057,
+      "step": 4628
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3846480172349352e-07,
+      "loss": 0.7764,
+      "step": 4629
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.373662222465866e-07,
+      "loss": 0.8174,
+      "step": 4630
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3627198806829833e-07,
+      "loss": 0.7832,
+      "step": 4631
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.351820996707298e-07,
+      "loss": 0.707,
+      "step": 4632
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.340965575340658e-07,
+      "loss": 0.7769,
+      "step": 4633
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3301536213657707e-07,
+      "loss": 0.8013,
+      "step": 4634
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.31938513954617e-07,
+      "loss": 0.7593,
+      "step": 4635
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.308660134626283e-07,
+      "loss": 0.7236,
+      "step": 4636
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2979786113313408e-07,
+      "loss": 0.772,
+      "step": 4637
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2873405743674127e-07,
+      "loss": 0.7539,
+      "step": 4638
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2767460284214495e-07,
+      "loss": 0.7407,
+      "step": 4639
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2661949781611948e-07,
+      "loss": 0.7827,
+      "step": 4640
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2556874282352749e-07,
+      "loss": 0.8057,
+      "step": 4641
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2452233832731197e-07,
+      "loss": 0.8115,
+      "step": 4642
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.234802847884997e-07,
+      "loss": 0.7476,
+      "step": 4643
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.22442582666199e-07,
+      "loss": 0.7927,
+      "step": 4644
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2140923241760637e-07,
+      "loss": 0.7266,
+      "step": 4645
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.203802344979932e-07,
+      "loss": 0.8237,
+      "step": 4646
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1935558936072123e-07,
+      "loss": 0.749,
+      "step": 4647
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1833529745722827e-07,
+      "loss": 0.7563,
+      "step": 4648
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1731935923703586e-07,
+      "loss": 0.8022,
+      "step": 4649
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1630777514775038e-07,
+      "loss": 0.7388,
+      "step": 4650
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1530054563505644e-07,
+      "loss": 0.7466,
+      "step": 4651
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1429767114272127e-07,
+      "loss": 0.7905,
+      "step": 4652
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1329915211259368e-07,
+      "loss": 0.7925,
+      "step": 4653
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1230498898460173e-07,
+      "loss": 0.7183,
+      "step": 4654
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.113151821967573e-07,
+      "loss": 0.7739,
+      "step": 4655
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1032973218515042e-07,
+      "loss": 0.7466,
+      "step": 4656
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0934863938395269e-07,
+      "loss": 0.8047,
+      "step": 4657
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0837190422541721e-07,
+      "loss": 0.7451,
+      "step": 4658
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0739952713987423e-07,
+      "loss": 0.7812,
+      "step": 4659
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0643150855573658e-07,
+      "loss": 0.7104,
+      "step": 4660
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0546784889949536e-07,
+      "loss": 0.7876,
+      "step": 4661
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0450854859572202e-07,
+      "loss": 0.7666,
+      "step": 4662
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0355360806706738e-07,
+      "loss": 0.771,
+      "step": 4663
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0260302773425934e-07,
+      "loss": 0.7773,
+      "step": 4664
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0165680801610734e-07,
+      "loss": 0.8276,
+      "step": 4665
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0071494932949899e-07,
+      "loss": 0.8052,
+      "step": 4666
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.977745208940015e-08,
+      "loss": 0.8213,
+      "step": 4667
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.884431670885486e-08,
+      "loss": 0.7295,
+      "step": 4668
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.791554359898647e-08,
+      "loss": 0.7964,
+      "step": 4669
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.699113316899544e-08,
+      "loss": 0.7505,
+      "step": 4670
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.607108582615932e-08,
+      "loss": 0.8032,
+      "step": 4671
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.515540197583716e-08,
+      "loss": 0.7778,
+      "step": 4672
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.424408202145852e-08,
+      "loss": 0.8599,
+      "step": 4673
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.333712636453773e-08,
+      "loss": 0.813,
+      "step": 4674
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.243453540466185e-08,
+      "loss": 0.7812,
+      "step": 4675
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.153630953949722e-08,
+      "loss": 0.7798,
+      "step": 4676
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.064244916478504e-08,
+      "loss": 0.8164,
+      "step": 4677
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.975295467434364e-08,
+      "loss": 0.7373,
+      "step": 4678
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.886782646006953e-08,
+      "loss": 0.7983,
+      "step": 4679
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.798706491193298e-08,
+      "loss": 0.7686,
+      "step": 4680
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.71106704179836e-08,
+      "loss": 0.7568,
+      "step": 4681
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.623864336434251e-08,
+      "loss": 0.77,
+      "step": 4682
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.537098413521127e-08,
+      "loss": 0.7612,
+      "step": 4683
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.4507693112863e-08,
+      "loss": 0.7461,
+      "step": 4684
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.364877067764899e-08,
+      "loss": 0.75,
+      "step": 4685
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.279421720799541e-08,
+      "loss": 0.7461,
+      "step": 4686
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.194403308040333e-08,
+      "loss": 0.8086,
+      "step": 4687
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.109821866944645e-08,
+      "loss": 0.7744,
+      "step": 4688
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.02567743477778e-08,
+      "loss": 0.8135,
+      "step": 4689
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.941970048612191e-08,
+      "loss": 0.7266,
+      "step": 4690
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.858699745327825e-08,
+      "loss": 0.7783,
+      "step": 4691
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.775866561612e-08,
+      "loss": 0.8418,
+      "step": 4692
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.693470533959746e-08,
+      "loss": 0.7715,
+      "step": 4693
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.611511698672913e-08,
+      "loss": 0.7715,
+      "step": 4694
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.529990091861394e-08,
+      "loss": 0.7676,
+      "step": 4695
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.448905749441904e-08,
+      "loss": 0.7798,
+      "step": 4696
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.368258707139087e-08,
+      "loss": 0.7573,
+      "step": 4697
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.288049000484187e-08,
+      "loss": 0.7803,
+      "step": 4698
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.208276664816382e-08,
+      "loss": 0.7314,
+      "step": 4699
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.128941735281891e-08,
+      "loss": 0.7559,
+      "step": 4700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.050044246834198e-08,
+      "loss": 0.7778,
+      "step": 4701
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.971584234234164e-08,
+      "loss": 0.7573,
+      "step": 4702
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.893561732049803e-08,
+      "loss": 0.7778,
+      "step": 4703
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.815976774656507e-08,
+      "loss": 0.751,
+      "step": 4704
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.738829396236712e-08,
+      "loss": 0.7358,
+      "step": 4705
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.662119630780339e-08,
+      "loss": 0.7964,
+      "step": 4706
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.585847512084132e-08,
+      "loss": 0.7524,
+      "step": 4707
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.510013073752208e-08,
+      "loss": 0.8101,
+      "step": 4708
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.434616349195955e-08,
+      "loss": 0.792,
+      "step": 4709
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.359657371633798e-08,
+      "loss": 0.7456,
+      "step": 4710
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.285136174091322e-08,
+      "loss": 0.7944,
+      "step": 4711
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.211052789401151e-08,
+      "loss": 0.7788,
+      "step": 4712
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.137407250203064e-08,
+      "loss": 0.7046,
+      "step": 4713
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.064199588944108e-08,
+      "loss": 0.8018,
+      "step": 4714
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.991429837878149e-08,
+      "loss": 0.7295,
+      "step": 4715
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.919098029066317e-08,
+      "loss": 0.7036,
+      "step": 4716
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.847204194376566e-08,
+      "loss": 0.7568,
+      "step": 4717
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.775748365484224e-08,
+      "loss": 0.7993,
+      "step": 4718
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.704730573871109e-08,
+      "loss": 0.748,
+      "step": 4719
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.634150850826747e-08,
+      "loss": 0.7241,
+      "step": 4720
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.564009227447042e-08,
+      "loss": 0.7593,
+      "step": 4721
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.4943057346352744e-08,
+      "loss": 0.7881,
+      "step": 4722
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.425040403101434e-08,
+      "loss": 0.8022,
+      "step": 4723
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.356213263362664e-08,
+      "loss": 0.77,
+      "step": 4724
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.287824345742931e-08,
+      "loss": 0.7544,
+      "step": 4725
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.219873680373133e-08,
+      "loss": 0.7842,
+      "step": 4726
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.152361297191211e-08,
+      "loss": 0.8027,
+      "step": 4727
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.085287225941704e-08,
+      "loss": 0.8086,
+      "step": 4728
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.018651496176419e-08,
+      "loss": 0.7114,
+      "step": 4729
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9524541372537595e-08,
+      "loss": 0.7368,
+      "step": 4730
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.886695178339173e-08,
+      "loss": 0.731,
+      "step": 4731
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.82137464840482e-08,
+      "loss": 0.8062,
+      "step": 4732
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.756492576229788e-08,
+      "loss": 0.8228,
+      "step": 4733
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6920489903997666e-08,
+      "loss": 0.7764,
+      "step": 4734
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.628043919307712e-08,
+      "loss": 0.8032,
+      "step": 4735
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.564477391152955e-08,
+      "loss": 0.7251,
+      "step": 4736
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5013494339417596e-08,
+      "loss": 0.7832,
+      "step": 4737
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4386600754872114e-08,
+      "loss": 0.7803,
+      "step": 4738
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3764093434091045e-08,
+      "loss": 0.7563,
+      "step": 4739
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3145972651339464e-08,
+      "loss": 0.7915,
+      "step": 4740
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.253223867894951e-08,
+      "loss": 0.7754,
+      "step": 4741
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.192289178732378e-08,
+      "loss": 0.7632,
+      "step": 4742
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.131793224492753e-08,
+      "loss": 0.7886,
+      "step": 4743
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.071736031829532e-08,
+      "loss": 0.7822,
+      "step": 4744
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.012117627202883e-08,
+      "loss": 0.8013,
+      "step": 4745
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.952938036879572e-08,
+      "loss": 0.8008,
+      "step": 4746
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.894197286933077e-08,
+      "loss": 0.6885,
+      "step": 4747
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.835895403243472e-08,
+      "loss": 0.7568,
+      "step": 4748
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7780324114975454e-08,
+      "loss": 0.7056,
+      "step": 4749
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.720608337188569e-08,
+      "loss": 0.7695,
+      "step": 4750
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.663623205616862e-08,
+      "loss": 0.8071,
+      "step": 4751
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6070770418886736e-08,
+      "loss": 0.7515,
+      "step": 4752
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.55096987091752e-08,
+      "loss": 0.7695,
+      "step": 4753
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.495301717423072e-08,
+      "loss": 0.7295,
+      "step": 4754
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4400726059317105e-08,
+      "loss": 0.7832,
+      "step": 4755
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.385282560776415e-08,
+      "loss": 0.7139,
+      "step": 4756
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3309316060966545e-08,
+      "loss": 0.8232,
+      "step": 4757
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.277019765838607e-08,
+      "loss": 0.7144,
+      "step": 4758
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2235470637546064e-08,
+      "loss": 0.7695,
+      "step": 4759
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1705135234040284e-08,
+      "loss": 0.7827,
+      "step": 4760
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1179191681524055e-08,
+      "loss": 0.7046,
+      "step": 4761
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.065764021171758e-08,
+      "loss": 0.7559,
+      "step": 4762
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0140481054408144e-08,
+      "loss": 0.8213,
+      "step": 4763
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.962771443744572e-08,
+      "loss": 0.7734,
+      "step": 4764
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9119340586748478e-08,
+      "loss": 0.7441,
+      "step": 4765
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8615359726293923e-08,
+      "loss": 0.7837,
+      "step": 4766
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8115772078127767e-08,
+      "loss": 0.7993,
+      "step": 4767
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7620577862360608e-08,
+      "loss": 0.7832,
+      "step": 4768
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7129777297163484e-08,
+      "loss": 0.7759,
+      "step": 4769
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6643370598777862e-08,
+      "loss": 0.8115,
+      "step": 4770
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6161357981502323e-08,
+      "loss": 0.7407,
+      "step": 4771
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5683739657704765e-08,
+      "loss": 0.748,
+      "step": 4772
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.521051583781575e-08,
+      "loss": 0.8091,
+      "step": 4773
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4741686730327398e-08,
+      "loss": 0.7905,
+      "step": 4774
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4277252541797803e-08,
+      "loss": 0.7773,
+      "step": 4775
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3817213476848844e-08,
+      "loss": 0.8047,
+      "step": 4776
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3361569738166168e-08,
+      "loss": 0.7515,
+      "step": 4777
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.291032152649697e-08,
+      "loss": 0.7754,
+      "step": 4778
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.246346904065333e-08,
+      "loss": 0.771,
+      "step": 4779
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2021012477509983e-08,
+      "loss": 0.7568,
+      "step": 4780
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1582952032006556e-08,
+      "loss": 0.7974,
+      "step": 4781
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.11492878971431e-08,
+      "loss": 0.7915,
+      "step": 4782
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0720020263985676e-08,
+      "loss": 0.7905,
+      "step": 4783
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0295149321660767e-08,
+      "loss": 0.7563,
+      "step": 4784
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9874675257359754e-08,
+      "loss": 0.7993,
+      "step": 4785
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.945859825633556e-08,
+      "loss": 0.812,
+      "step": 4786
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9046918501902655e-08,
+      "loss": 0.73,
+      "step": 4787
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8639636175442622e-08,
+      "loss": 0.7437,
+      "step": 4788
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8236751456394143e-08,
+      "loss": 0.7861,
+      "step": 4789
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7838264522263004e-08,
+      "loss": 0.7993,
+      "step": 4790
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.744417554861433e-08,
+      "loss": 0.7607,
+      "step": 4791
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.705448470907589e-08,
+      "loss": 0.7437,
+      "step": 4792
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6669192175339245e-08,
+      "loss": 0.7725,
+      "step": 4793
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6288298117157487e-08,
+      "loss": 0.7637,
+      "step": 4794
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.591180270234638e-08,
+      "loss": 0.811,
+      "step": 4795
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5539706096781016e-08,
+      "loss": 0.7998,
+      "step": 4796
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5172008464402477e-08,
+      "loss": 0.772,
+      "step": 4797
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4808709967210067e-08,
+      "loss": 0.7681,
+      "step": 4798
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.444981076526797e-08,
+      "loss": 0.7935,
+      "step": 4799
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4095311016699698e-08,
+      "loss": 0.7979,
+      "step": 4800
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3745210877691428e-08,
+      "loss": 0.7627,
+      "step": 4801
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3399510502491996e-08,
+      "loss": 0.8784,
+      "step": 4802
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.305821004340957e-08,
+      "loss": 0.7993,
+      "step": 4803
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2721309650816082e-08,
+      "loss": 0.7446,
+      "step": 4804
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2388809473142805e-08,
+      "loss": 0.7866,
+      "step": 4805
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2060709656883662e-08,
+      "loss": 0.7473,
+      "step": 4806
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1737010346593025e-08,
+      "loss": 0.813,
+      "step": 4807
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1417711684887922e-08,
+      "loss": 0.8105,
+      "step": 4808
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1102813812445823e-08,
+      "loss": 0.7319,
+      "step": 4809
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0792316868003527e-08,
+      "loss": 0.7622,
+      "step": 4810
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0486220988361607e-08,
+      "loss": 0.772,
+      "step": 4811
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0184526308379961e-08,
+      "loss": 0.7407,
+      "step": 4812
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.887232960980042e-09,
+      "loss": 0.7617,
+      "step": 4813
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.594341077143743e-09,
+      "loss": 0.7827,
+      "step": 4814
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.305850785914505e-09,
+      "loss": 0.7832,
+      "step": 4815
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.021762214395102e-09,
+      "loss": 0.7378,
+      "step": 4816
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.742075487749858e-09,
+      "loss": 0.8135,
+      "step": 4817
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.466790729203534e-09,
+      "loss": 0.7588,
+      "step": 4818
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.195908060042445e-09,
+      "loss": 0.7554,
+      "step": 4819
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.929427599613348e-09,
+      "loss": 0.7578,
+      "step": 4820
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.667349465320106e-09,
+      "loss": 0.7427,
+      "step": 4821
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.409673772632575e-09,
+      "loss": 0.7612,
+      "step": 4822
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.1564006350755e-09,
+      "loss": 0.7495,
+      "step": 4823
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.907530164238508e-09,
+      "loss": 0.7422,
+      "step": 4824
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.6630624697683336e-09,
+      "loss": 0.8052,
+      "step": 4825
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.422997659373265e-09,
+      "loss": 0.7324,
+      "step": 4826
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.1873358388209185e-09,
+      "loss": 0.7002,
+      "step": 4827
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.956077111940462e-09,
+      "loss": 0.73,
+      "step": 4828
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.729221580620392e-09,
+      "loss": 0.7681,
+      "step": 4829
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.506769344808538e-09,
+      "loss": 0.7822,
+      "step": 4830
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.288720502513167e-09,
+      "loss": 0.7842,
+      "step": 4831
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.075075149804099e-09,
+      "loss": 0.7808,
+      "step": 4832
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.86583338080826e-09,
+      "loss": 0.7964,
+      "step": 4833
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.660995287714132e-09,
+      "loss": 0.7744,
+      "step": 4834
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.460560960769522e-09,
+      "loss": 0.8013,
+      "step": 4835
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.264530488283791e-09,
+      "loss": 0.7661,
+      "step": 4836
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.072903956621188e-09,
+      "loss": 0.7637,
+      "step": 4837
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.885681450211954e-09,
+      "loss": 0.7393,
+      "step": 4838
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.702863051541217e-09,
+      "loss": 0.7925,
+      "step": 4839
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.524448841156769e-09,
+      "loss": 0.8179,
+      "step": 4840
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3504388976635104e-09,
+      "loss": 0.7495,
+      "step": 4841
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1808332977278923e-09,
+      "loss": 0.8213,
+      "step": 4842
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0156321160734747e-09,
+      "loss": 0.7656,
+      "step": 4843
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8548354254875897e-09,
+      "loss": 0.7944,
+      "step": 4844
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6984432968124585e-09,
+      "loss": 0.7573,
+      "step": 4845
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5464557989518524e-09,
+      "loss": 0.8125,
+      "step": 4846
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.398872998868873e-09,
+      "loss": 0.7842,
+      "step": 4847
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2556949615870625e-09,
+      "loss": 0.7954,
+      "step": 4848
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.116921750185963e-09,
+      "loss": 0.7905,
+      "step": 4849
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.982553425807776e-09,
+      "loss": 0.7612,
+      "step": 4850
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8525900476529246e-09,
+      "loss": 0.7827,
+      "step": 4851
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7270316729800507e-09,
+      "loss": 0.8311,
+      "step": 4852
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.605878357108237e-09,
+      "loss": 0.7271,
+      "step": 4853
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4891301534158964e-09,
+      "loss": 0.8076,
+      "step": 4854
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3767871133396616e-09,
+      "loss": 0.8657,
+      "step": 4855
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2688492863766056e-09,
+      "loss": 0.8071,
+      "step": 4856
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1653167200820215e-09,
+      "loss": 0.7959,
+      "step": 4857
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0661894600694223e-09,
+      "loss": 0.8174,
+      "step": 4858
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.714675500127613e-10,
+      "loss": 0.8179,
+      "step": 4859
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.811510316464322e-10,
+      "loss": 0.7847,
+      "step": 4860
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.952399447608284e-10,
+      "loss": 0.7666,
+      "step": 4861
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.137343272067832e-10,
+      "loss": 0.8086,
+      "step": 4862
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.366342148944604e-10,
+      "loss": 0.7744,
+      "step": 4863
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.639396417922438e-10,
+      "loss": 0.7993,
+      "step": 4864
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.956506399300676e-10,
+      "loss": 0.792,
+      "step": 4865
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3176723939275524e-10,
+      "loss": 0.7402,
+      "step": 4866
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.722894683266809e-10,
+      "loss": 0.7832,
+      "step": 4867
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1721735293643865e-10,
+      "loss": 0.7549,
+      "step": 4868
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6655091748706287e-10,
+      "loss": 0.7778,
+      "step": 4869
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2029018429958747e-10,
+      "loss": 0.7476,
+      "step": 4870
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.78435173756597e-10,
+      "loss": 0.7251,
+      "step": 4871
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.409859042988959e-10,
+      "loss": 0.7534,
+      "step": 4872
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0794239242661875e-10,
+      "loss": 0.7891,
+      "step": 4873
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.93046526958996e-11,
+      "loss": 0.8179,
+      "step": 4874
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.5072697725533366e-11,
+      "loss": 0.7734,
+      "step": 4875
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.5246538190314425e-11,
+      "loss": 0.791,
+      "step": 4876
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9826182827697992e-11,
+      "loss": 0.7402,
+      "step": 4877
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.811638430028524e-12,
+      "loss": 0.772,
+      "step": 4878
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2029098500908617e-12,
+      "loss": 0.7642,
+      "step": 4879
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.791,
+      "step": 4880
+    },
+    {
+      "epoch": 1.0,
+      "step": 4880,
+      "total_flos": 3383069183344640.0,
+      "train_loss": 0.8311814104924437,
+      "train_runtime": 146283.8209,
+      "train_samples_per_second": 4.27,
+      "train_steps_per_second": 0.033
+    }
+  ],
+  "max_steps": 4880,
+  "num_train_epochs": 1,
+  "total_flos": 3383069183344640.0,
+  "trial_name": null,
+  "trial_params": null
+}