diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,29296 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 4878,
+  "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": 1.5412,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.72108843537415e-07,
+      "loss": 1.4204,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0816326530612243e-07,
+      "loss": 1.5253,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.4421768707483e-07,
+      "loss": 1.603,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.802721088435376e-07,
+      "loss": 1.4726,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.163265306122449e-07,
+      "loss": 1.4662,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.523809523809525e-07,
+      "loss": 1.4495,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.08843537414966e-06,
+      "loss": 1.5395,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2244897959183673e-06,
+      "loss": 1.4849,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3605442176870751e-06,
+      "loss": 1.4174,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4965986394557825e-06,
+      "loss": 1.4661,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6326530612244897e-06,
+      "loss": 1.5052,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7687074829931975e-06,
+      "loss": 1.4727,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.904761904761905e-06,
+      "loss": 1.5046,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0408163265306125e-06,
+      "loss": 1.4497,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.17687074829932e-06,
+      "loss": 1.5748,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3129251700680273e-06,
+      "loss": 1.5383,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4489795918367347e-06,
+      "loss": 1.3663,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5850340136054425e-06,
+      "loss": 1.6164,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7210884353741503e-06,
+      "loss": 1.6372,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 1.7337,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.993197278911565e-06,
+      "loss": 1.7094,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1292517006802725e-06,
+      "loss": 1.504,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2653061224489794e-06,
+      "loss": 1.415,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4013605442176872e-06,
+      "loss": 1.4085,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.537414965986395e-06,
+      "loss": 1.4691,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6734693877551024e-06,
+      "loss": 1.4734,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.80952380952381e-06,
+      "loss": 1.4565,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.945578231292517e-06,
+      "loss": 1.5746,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.081632653061225e-06,
+      "loss": 1.4122,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.217687074829933e-06,
+      "loss": 1.4582,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.35374149659864e-06,
+      "loss": 1.6504,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.489795918367348e-06,
+      "loss": 1.553,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6258503401360546e-06,
+      "loss": 1.5396,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.761904761904762e-06,
+      "loss": 1.4655,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.897959183673469e-06,
+      "loss": 1.4006,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.034013605442177e-06,
+      "loss": 1.4348,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.170068027210885e-06,
+      "loss": 1.5741,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.306122448979593e-06,
+      "loss": 1.6622,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.442176870748301e-06,
+      "loss": 1.478,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.578231292517007e-06,
+      "loss": 1.5812,
+      "step": 41
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.7142857142857145e-06,
+      "loss": 1.499,
+      "step": 42
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.850340136054422e-06,
+      "loss": 1.5314,
+      "step": 43
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.98639455782313e-06,
+      "loss": 1.4496,
+      "step": 44
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.122448979591837e-06,
+      "loss": 1.4053,
+      "step": 45
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.258503401360545e-06,
+      "loss": 1.687,
+      "step": 46
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.394557823129253e-06,
+      "loss": 1.5804,
+      "step": 47
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.530612244897959e-06,
+      "loss": 1.5009,
+      "step": 48
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.5011,
+      "step": 49
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.8027210884353745e-06,
+      "loss": 1.541,
+      "step": 50
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.938775510204082e-06,
+      "loss": 1.4231,
+      "step": 51
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.07482993197279e-06,
+      "loss": 1.4189,
+      "step": 52
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.210884353741497e-06,
+      "loss": 1.5391,
+      "step": 53
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.346938775510205e-06,
+      "loss": 1.5452,
+      "step": 54
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.482993197278913e-06,
+      "loss": 1.661,
+      "step": 55
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.61904761904762e-06,
+      "loss": 1.4098,
+      "step": 56
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.755102040816327e-06,
+      "loss": 1.4403,
+      "step": 57
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.891156462585034e-06,
+      "loss": 1.4564,
+      "step": 58
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.027210884353741e-06,
+      "loss": 1.4413,
+      "step": 59
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.16326530612245e-06,
+      "loss": 1.3614,
+      "step": 60
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.299319727891157e-06,
+      "loss": 1.6139,
+      "step": 61
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.435374149659866e-06,
+      "loss": 1.4755,
+      "step": 62
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.571428571428571e-06,
+      "loss": 1.459,
+      "step": 63
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.70748299319728e-06,
+      "loss": 1.5303,
+      "step": 64
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.843537414965987e-06,
+      "loss": 1.4404,
+      "step": 65
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.979591836734695e-06,
+      "loss": 1.5945,
+      "step": 66
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.115646258503402e-06,
+      "loss": 1.4522,
+      "step": 67
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.251700680272109e-06,
+      "loss": 1.6329,
+      "step": 68
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.387755102040818e-06,
+      "loss": 1.576,
+      "step": 69
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.523809523809525e-06,
+      "loss": 1.727,
+      "step": 70
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.659863945578232e-06,
+      "loss": 1.4203,
+      "step": 71
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.795918367346939e-06,
+      "loss": 1.6167,
+      "step": 72
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.931972789115647e-06,
+      "loss": 1.4441,
+      "step": 73
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0068027210884354e-05,
+      "loss": 1.4248,
+      "step": 74
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0204081632653063e-05,
+      "loss": 1.3576,
+      "step": 75
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.034013605442177e-05,
+      "loss": 1.4722,
+      "step": 76
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0476190476190477e-05,
+      "loss": 1.5392,
+      "step": 77
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0612244897959186e-05,
+      "loss": 1.536,
+      "step": 78
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0748299319727893e-05,
+      "loss": 1.63,
+      "step": 79
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0884353741496601e-05,
+      "loss": 1.5653,
+      "step": 80
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1020408163265306e-05,
+      "loss": 1.5921,
+      "step": 81
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1156462585034013e-05,
+      "loss": 1.5837,
+      "step": 82
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1292517006802722e-05,
+      "loss": 1.4381,
+      "step": 83
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1428571428571429e-05,
+      "loss": 1.6156,
+      "step": 84
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1564625850340136e-05,
+      "loss": 1.4182,
+      "step": 85
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1700680272108845e-05,
+      "loss": 1.576,
+      "step": 86
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1836734693877552e-05,
+      "loss": 1.5249,
+      "step": 87
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.197278911564626e-05,
+      "loss": 1.4377,
+      "step": 88
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2108843537414967e-05,
+      "loss": 1.7118,
+      "step": 89
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2244897959183674e-05,
+      "loss": 1.3784,
+      "step": 90
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2380952380952383e-05,
+      "loss": 1.4776,
+      "step": 91
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.251700680272109e-05,
+      "loss": 1.588,
+      "step": 92
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2653061224489798e-05,
+      "loss": 1.736,
+      "step": 93
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2789115646258505e-05,
+      "loss": 1.4219,
+      "step": 94
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2925170068027212e-05,
+      "loss": 1.6595,
+      "step": 95
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3061224489795918e-05,
+      "loss": 1.6177,
+      "step": 96
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3197278911564626e-05,
+      "loss": 1.3722,
+      "step": 97
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 1.6181,
+      "step": 98
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3469387755102042e-05,
+      "loss": 1.6844,
+      "step": 99
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3605442176870749e-05,
+      "loss": 1.4439,
+      "step": 100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3741496598639456e-05,
+      "loss": 1.5016,
+      "step": 101
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3877551020408165e-05,
+      "loss": 1.7645,
+      "step": 102
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4013605442176872e-05,
+      "loss": 1.5281,
+      "step": 103
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.414965986394558e-05,
+      "loss": 1.498,
+      "step": 104
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4285714285714287e-05,
+      "loss": 1.3896,
+      "step": 105
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4421768707482994e-05,
+      "loss": 1.4337,
+      "step": 106
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4557823129251703e-05,
+      "loss": 1.338,
+      "step": 107
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.469387755102041e-05,
+      "loss": 1.3836,
+      "step": 108
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4829931972789118e-05,
+      "loss": 1.4506,
+      "step": 109
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4965986394557825e-05,
+      "loss": 1.5825,
+      "step": 110
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.510204081632653e-05,
+      "loss": 1.4437,
+      "step": 111
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.523809523809524e-05,
+      "loss": 1.4538,
+      "step": 112
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5374149659863945e-05,
+      "loss": 1.4314,
+      "step": 113
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5510204081632655e-05,
+      "loss": 1.5365,
+      "step": 114
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5646258503401362e-05,
+      "loss": 1.2774,
+      "step": 115
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.578231292517007e-05,
+      "loss": 1.4299,
+      "step": 116
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5918367346938776e-05,
+      "loss": 1.497,
+      "step": 117
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6054421768707483e-05,
+      "loss": 1.4233,
+      "step": 118
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6190476190476193e-05,
+      "loss": 1.4502,
+      "step": 119
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.63265306122449e-05,
+      "loss": 1.4201,
+      "step": 120
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6462585034013607e-05,
+      "loss": 1.4587,
+      "step": 121
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6598639455782314e-05,
+      "loss": 1.4168,
+      "step": 122
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.673469387755102e-05,
+      "loss": 1.4133,
+      "step": 123
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.687074829931973e-05,
+      "loss": 1.2086,
+      "step": 124
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7006802721088435e-05,
+      "loss": 1.4507,
+      "step": 125
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7142857142857142e-05,
+      "loss": 1.5075,
+      "step": 126
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7278911564625852e-05,
+      "loss": 1.3626,
+      "step": 127
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.741496598639456e-05,
+      "loss": 1.5406,
+      "step": 128
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7551020408163266e-05,
+      "loss": 1.4562,
+      "step": 129
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7687074829931973e-05,
+      "loss": 1.4265,
+      "step": 130
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.782312925170068e-05,
+      "loss": 1.5058,
+      "step": 131
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.795918367346939e-05,
+      "loss": 1.3645,
+      "step": 132
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8095238095238097e-05,
+      "loss": 1.5957,
+      "step": 133
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8231292517006804e-05,
+      "loss": 1.4585,
+      "step": 134
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.836734693877551e-05,
+      "loss": 1.578,
+      "step": 135
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8503401360544218e-05,
+      "loss": 1.4656,
+      "step": 136
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.863945578231293e-05,
+      "loss": 1.4107,
+      "step": 137
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8775510204081636e-05,
+      "loss": 1.4062,
+      "step": 138
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8911564625850343e-05,
+      "loss": 1.4162,
+      "step": 139
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.904761904761905e-05,
+      "loss": 1.6134,
+      "step": 140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9183673469387756e-05,
+      "loss": 1.4602,
+      "step": 141
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9319727891156463e-05,
+      "loss": 1.3472,
+      "step": 142
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.945578231292517e-05,
+      "loss": 1.4645,
+      "step": 143
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9591836734693877e-05,
+      "loss": 1.3922,
+      "step": 144
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9727891156462588e-05,
+      "loss": 1.3337,
+      "step": 145
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9863945578231295e-05,
+      "loss": 1.2756,
+      "step": 146
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 1.3596,
+      "step": 147
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999997795227226e-05,
+      "loss": 1.5198,
+      "step": 148
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999999118090987e-05,
+      "loss": 1.5528,
+      "step": 149
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999980157050852e-05,
+      "loss": 1.3028,
+      "step": 150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999964723655033e-05,
+      "loss": 1.4145,
+      "step": 151
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999944880729214e-05,
+      "loss": 1.5004,
+      "step": 152
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999992062828215e-05,
+      "loss": 1.4255,
+      "step": 153
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999891966324538e-05,
+      "loss": 1.3304,
+      "step": 154
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999858894869007e-05,
+      "loss": 1.3287,
+      "step": 155
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999821413930146e-05,
+      "loss": 1.417,
+      "step": 156
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999779523524485e-05,
+      "loss": 1.2831,
+      "step": 157
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999973322367049e-05,
+      "loss": 1.2192,
+      "step": 158
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999968251438858e-05,
+      "loss": 1.2426,
+      "step": 159
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999627395701112e-05,
+      "loss": 1.46,
+      "step": 160
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999567867632396e-05,
+      "loss": 1.4341,
+      "step": 161
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999950393020868e-05,
+      "loss": 1.3358,
+      "step": 162
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999435583458157e-05,
+      "loss": 1.4748,
+      "step": 163
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999362827410965e-05,
+      "loss": 1.4494,
+      "step": 164
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999285662099182e-05,
+      "loss": 1.3232,
+      "step": 165
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999920408755684e-05,
+      "loss": 1.6212,
+      "step": 166
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999118103819906e-05,
+      "loss": 1.4268,
+      "step": 167
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.99990277109263e-05,
+      "loss": 1.3777,
+      "step": 168
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998932908915873e-05,
+      "loss": 1.3009,
+      "step": 169
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998833697830434e-05,
+      "loss": 1.3969,
+      "step": 170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998730077713732e-05,
+      "loss": 1.4864,
+      "step": 171
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998622048611453e-05,
+      "loss": 1.2586,
+      "step": 172
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999850961057124e-05,
+      "loss": 1.4113,
+      "step": 173
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999839276364267e-05,
+      "loss": 1.4923,
+      "step": 174
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998271507877264e-05,
+      "loss": 1.3681,
+      "step": 175
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999814584332849e-05,
+      "loss": 1.2826,
+      "step": 176
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998015770051766e-05,
+      "loss": 1.3349,
+      "step": 177
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997881288104445e-05,
+      "loss": 1.47,
+      "step": 178
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999774239754583e-05,
+      "loss": 1.3829,
+      "step": 179
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997599098437162e-05,
+      "loss": 1.4318,
+      "step": 180
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997451390841628e-05,
+      "loss": 1.3251,
+      "step": 181
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997299274824366e-05,
+      "loss": 1.3449,
+      "step": 182
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999714275045245e-05,
+      "loss": 1.3612,
+      "step": 183
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996981817794898e-05,
+      "loss": 1.3308,
+      "step": 184
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999681647692268e-05,
+      "loss": 1.3197,
+      "step": 185
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999664672790869e-05,
+      "loss": 1.3199,
+      "step": 186
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996472570827796e-05,
+      "loss": 1.4737,
+      "step": 187
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996294005756785e-05,
+      "loss": 1.3621,
+      "step": 188
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996111032774398e-05,
+      "loss": 1.3712,
+      "step": 189
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995923651961316e-05,
+      "loss": 1.5121,
+      "step": 190
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995731863400165e-05,
+      "loss": 1.3169,
+      "step": 191
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995535667175517e-05,
+      "loss": 1.3165,
+      "step": 192
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995335063373888e-05,
+      "loss": 1.4391,
+      "step": 193
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999513005208373e-05,
+      "loss": 1.2819,
+      "step": 194
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994920633395445e-05,
+      "loss": 1.3932,
+      "step": 195
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999470680740138e-05,
+      "loss": 1.4275,
+      "step": 196
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994488574195822e-05,
+      "loss": 1.3975,
+      "step": 197
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994265933875e-05,
+      "loss": 1.2695,
+      "step": 198
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994038886537086e-05,
+      "loss": 1.3546,
+      "step": 199
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993807432282205e-05,
+      "loss": 1.3055,
+      "step": 200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993571571212408e-05,
+      "loss": 1.4277,
+      "step": 201
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993331303431707e-05,
+      "loss": 1.3482,
+      "step": 202
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993086629046045e-05,
+      "loss": 1.4288,
+      "step": 203
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992837548163315e-05,
+      "loss": 1.3639,
+      "step": 204
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999258406089335e-05,
+      "loss": 1.2814,
+      "step": 205
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992326167347923e-05,
+      "loss": 1.1091,
+      "step": 206
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992063867640757e-05,
+      "loss": 1.6147,
+      "step": 207
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991797161887514e-05,
+      "loss": 1.329,
+      "step": 208
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99915260502058e-05,
+      "loss": 1.5283,
+      "step": 209
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991250532715158e-05,
+      "loss": 1.3285,
+      "step": 210
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999097060953708e-05,
+      "loss": 1.2206,
+      "step": 211
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990686280795003e-05,
+      "loss": 1.5427,
+      "step": 212
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99903975466143e-05,
+      "loss": 1.3751,
+      "step": 213
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990104407122294e-05,
+      "loss": 1.378,
+      "step": 214
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998980686244824e-05,
+      "loss": 1.3656,
+      "step": 215
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989504912723347e-05,
+      "loss": 1.2892,
+      "step": 216
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998919855808076e-05,
+      "loss": 1.2959,
+      "step": 217
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988887798655564e-05,
+      "loss": 1.3592,
+      "step": 218
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988572634584792e-05,
+      "loss": 1.4703,
+      "step": 219
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998825306600742e-05,
+      "loss": 1.2513,
+      "step": 220
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987929093064355e-05,
+      "loss": 1.2678,
+      "step": 221
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987600715898462e-05,
+      "loss": 1.4465,
+      "step": 222
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998726793465454e-05,
+      "loss": 1.3471,
+      "step": 223
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986930749479328e-05,
+      "loss": 1.3186,
+      "step": 224
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986589160521508e-05,
+      "loss": 1.3406,
+      "step": 225
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986243167931708e-05,
+      "loss": 1.3646,
+      "step": 226
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985892771862493e-05,
+      "loss": 1.383,
+      "step": 227
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985537972468374e-05,
+      "loss": 1.2998,
+      "step": 228
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.99851787699058e-05,
+      "loss": 1.4051,
+      "step": 229
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984815164333163e-05,
+      "loss": 1.4455,
+      "step": 230
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984447155910796e-05,
+      "loss": 1.327,
+      "step": 231
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998407474480098e-05,
+      "loss": 1.3544,
+      "step": 232
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983697931167922e-05,
+      "loss": 1.3715,
+      "step": 233
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983316715177784e-05,
+      "loss": 1.2822,
+      "step": 234
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9982931096998662e-05,
+      "loss": 1.2221,
+      "step": 235
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9982541076800602e-05,
+      "loss": 1.4726,
+      "step": 236
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998214665475558e-05,
+      "loss": 1.3328,
+      "step": 237
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981747831037524e-05,
+      "loss": 1.448,
+      "step": 238
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998134460582229e-05,
+      "loss": 1.4131,
+      "step": 239
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980936979287685e-05,
+      "loss": 1.2918,
+      "step": 240
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980524951613454e-05,
+      "loss": 1.403,
+      "step": 241
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980108522981287e-05,
+      "loss": 1.3683,
+      "step": 242
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.99796876935748e-05,
+      "loss": 1.3502,
+      "step": 243
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997926246357957e-05,
+      "loss": 1.226,
+      "step": 244
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978832833183098e-05,
+      "loss": 1.3497,
+      "step": 245
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997839880257483e-05,
+      "loss": 1.3512,
+      "step": 246
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997796037194616e-05,
+      "loss": 1.4111,
+      "step": 247
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977517541490412e-05,
+      "loss": 1.3489,
+      "step": 248
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977070311402854e-05,
+      "loss": 1.4551,
+      "step": 249
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976618681880696e-05,
+      "loss": 1.4149,
+      "step": 250
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976162653123083e-05,
+      "loss": 1.2821,
+      "step": 251
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9975702225331108e-05,
+      "loss": 1.3147,
+      "step": 252
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9975237398707793e-05,
+      "loss": 1.3086,
+      "step": 253
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997476817345811e-05,
+      "loss": 1.3033,
+      "step": 254
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974294549788964e-05,
+      "loss": 1.2304,
+      "step": 255
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.99738165279092e-05,
+      "loss": 1.2498,
+      "step": 256
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973334108029607e-05,
+      "loss": 1.289,
+      "step": 257
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972847290362906e-05,
+      "loss": 1.2782,
+      "step": 258
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972356075123767e-05,
+      "loss": 1.3554,
+      "step": 259
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997186046252879e-05,
+      "loss": 1.5085,
+      "step": 260
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971360452796523e-05,
+      "loss": 1.2036,
+      "step": 261
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997085604614744e-05,
+      "loss": 1.2549,
+      "step": 262
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970347242803962e-05,
+      "loss": 1.2666,
+      "step": 263
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969834042990452e-05,
+      "loss": 1.5079,
+      "step": 264
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969316446933207e-05,
+      "loss": 1.3155,
+      "step": 265
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996879445486046e-05,
+      "loss": 1.1964,
+      "step": 266
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968268067002396e-05,
+      "loss": 1.2856,
+      "step": 267
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9967737283591116e-05,
+      "loss": 1.3096,
+      "step": 268
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9967202104860676e-05,
+      "loss": 1.4238,
+      "step": 269
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966662531047065e-05,
+      "loss": 1.3108,
+      "step": 270
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996611856238821e-05,
+      "loss": 1.3131,
+      "step": 271
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996557019912398e-05,
+      "loss": 1.3641,
+      "step": 272
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965017441496174e-05,
+      "loss": 1.3261,
+      "step": 273
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964460289748533e-05,
+      "loss": 1.353,
+      "step": 274
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963898744126743e-05,
+      "loss": 1.1505,
+      "step": 275
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996333280487841e-05,
+      "loss": 1.3014,
+      "step": 276
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962762472253097e-05,
+      "loss": 1.3578,
+      "step": 277
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962187746502287e-05,
+      "loss": 1.3271,
+      "step": 278
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996160862787941e-05,
+      "loss": 1.3237,
+      "step": 279
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996102511663983e-05,
+      "loss": 1.1896,
+      "step": 280
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9960437213040852e-05,
+      "loss": 1.428,
+      "step": 281
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.995984491734172e-05,
+      "loss": 1.2631,
+      "step": 282
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.995924822980359e-05,
+      "loss": 1.3513,
+      "step": 283
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9958647150689598e-05,
+      "loss": 1.2745,
+      "step": 284
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9958041680264777e-05,
+      "loss": 1.3157,
+      "step": 285
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957431818796116e-05,
+      "loss": 1.3455,
+      "step": 286
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995681756655254e-05,
+      "loss": 1.3269,
+      "step": 287
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956198923804898e-05,
+      "loss": 1.1351,
+      "step": 288
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995557589082599e-05,
+      "loss": 1.4808,
+      "step": 289
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954948467890547e-05,
+      "loss": 1.3081,
+      "step": 290
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995431665527523e-05,
+      "loss": 1.3556,
+      "step": 291
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953680453258637e-05,
+      "loss": 1.3275,
+      "step": 292
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995303986212131e-05,
+      "loss": 1.2972,
+      "step": 293
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9952394882145718e-05,
+      "loss": 1.2017,
+      "step": 294
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951745513616267e-05,
+      "loss": 1.3712,
+      "step": 295
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951091756819298e-05,
+      "loss": 1.2625,
+      "step": 296
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9950433612043092e-05,
+      "loss": 1.1302,
+      "step": 297
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994977107957786e-05,
+      "loss": 1.3726,
+      "step": 298
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949104159715746e-05,
+      "loss": 1.2161,
+      "step": 299
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994843285275083e-05,
+      "loss": 1.2361,
+      "step": 300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9947757158979134e-05,
+      "loss": 1.3262,
+      "step": 301
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9947077078698608e-05,
+      "loss": 1.3648,
+      "step": 302
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994639261220913e-05,
+      "loss": 1.4277,
+      "step": 303
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994570375981252e-05,
+      "loss": 1.2515,
+      "step": 304
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9945010521812537e-05,
+      "loss": 1.2247,
+      "step": 305
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944312898514862e-05,
+      "loss": 1.4765,
+      "step": 306
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943610890227115e-05,
+      "loss": 1.4142,
+      "step": 307
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994290449725885e-05,
+      "loss": 1.3603,
+      "step": 308
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942193719921557e-05,
+      "loss": 1.4625,
+      "step": 309
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9941478558528655e-05,
+      "loss": 1.2238,
+      "step": 310
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99407590133955e-05,
+      "loss": 1.2972,
+      "step": 311
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994003508483937e-05,
+      "loss": 1.274,
+      "step": 312
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939306773179498e-05,
+      "loss": 1.3408,
+      "step": 313
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938574078737023e-05,
+      "loss": 1.2927,
+      "step": 314
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937837001835038e-05,
+      "loss": 1.3302,
+      "step": 315
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937095542798563e-05,
+      "loss": 1.2541,
+      "step": 316
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936349701954535e-05,
+      "loss": 1.2527,
+      "step": 317
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993559947963185e-05,
+      "loss": 1.3932,
+      "step": 318
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934844876161317e-05,
+      "loss": 1.4356,
+      "step": 319
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993408589187568e-05,
+      "loss": 1.3972,
+      "step": 320
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933322527109614e-05,
+      "loss": 1.3374,
+      "step": 321
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993255478219973e-05,
+      "loss": 1.4003,
+      "step": 322
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931782657484577e-05,
+      "loss": 1.3478,
+      "step": 323
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931006153304618e-05,
+      "loss": 1.3069,
+      "step": 324
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930225270002258e-05,
+      "loss": 1.2592,
+      "step": 325
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992944000792183e-05,
+      "loss": 1.2298,
+      "step": 326
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9928650367409602e-05,
+      "loss": 1.2996,
+      "step": 327
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9927856348813764e-05,
+      "loss": 1.3204,
+      "step": 328
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992705795248445e-05,
+      "loss": 1.2558,
+      "step": 329
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9926255178773713e-05,
+      "loss": 1.3122,
+      "step": 330
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9925448028035536e-05,
+      "loss": 1.4741,
+      "step": 331
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992463650062584e-05,
+      "loss": 1.3816,
+      "step": 332
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9923820596902473e-05,
+      "loss": 1.3561,
+      "step": 333
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9923000317225205e-05,
+      "loss": 1.2778,
+      "step": 334
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992217566195575e-05,
+      "loss": 1.4077,
+      "step": 335
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9921346631457737e-05,
+      "loss": 1.319,
+      "step": 336
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9920513226096735e-05,
+      "loss": 1.3585,
+      "step": 337
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9919675446240236e-05,
+      "loss": 1.4324,
+      "step": 338
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9918833292257663e-05,
+      "loss": 1.2455,
+      "step": 339
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9917986764520364e-05,
+      "loss": 1.3333,
+      "step": 340
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9917135863401627e-05,
+      "loss": 1.2818,
+      "step": 341
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991628058927666e-05,
+      "loss": 1.3331,
+      "step": 342
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991542094252259e-05,
+      "loss": 1.362,
+      "step": 343
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9914556923518495e-05,
+      "loss": 1.389,
+      "step": 344
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991368853264536e-05,
+      "loss": 1.3336,
+      "step": 345
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9912815770286105e-05,
+      "loss": 1.356,
+      "step": 346
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9911938636825586e-05,
+      "loss": 1.3678,
+      "step": 347
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991105713265057e-05,
+      "loss": 1.3628,
+      "step": 348
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991017125814977e-05,
+      "loss": 1.31,
+      "step": 349
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9909281013713806e-05,
+      "loss": 1.3063,
+      "step": 350
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9908386399735243e-05,
+      "loss": 1.3182,
+      "step": 351
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9907487416608564e-05,
+      "loss": 1.3165,
+      "step": 352
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9906584064730173e-05,
+      "loss": 1.4263,
+      "step": 353
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990567634449842e-05,
+      "loss": 1.2424,
+      "step": 354
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9904764256313557e-05,
+      "loss": 1.3259,
+      "step": 355
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9903847800577777e-05,
+      "loss": 1.3383,
+      "step": 356
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9902926977695196e-05,
+      "loss": 1.2055,
+      "step": 357
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9902001788071854e-05,
+      "loss": 1.2486,
+      "step": 358
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990107223211572e-05,
+      "loss": 1.2247,
+      "step": 359
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9900138310236683e-05,
+      "loss": 1.2413,
+      "step": 360
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9899200022846563e-05,
+      "loss": 1.2365,
+      "step": 361
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9898257370359096e-05,
+      "loss": 1.4523,
+      "step": 362
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9897310353189958e-05,
+      "loss": 1.3715,
+      "step": 363
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9896358971756733e-05,
+      "loss": 1.3303,
+      "step": 364
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9895403226478942e-05,
+      "loss": 1.3561,
+      "step": 365
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9894443117778022e-05,
+      "loss": 1.3438,
+      "step": 366
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9893478646077337e-05,
+      "loss": 1.4192,
+      "step": 367
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9892509811802178e-05,
+      "loss": 1.4931,
+      "step": 368
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9891536615379756e-05,
+      "loss": 1.2417,
+      "step": 369
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9890559057239204e-05,
+      "loss": 1.2074,
+      "step": 370
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9889577137811583e-05,
+      "loss": 1.2164,
+      "step": 371
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9888590857529878e-05,
+      "loss": 1.3832,
+      "step": 372
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9887600216828985e-05,
+      "loss": 1.4079,
+      "step": 373
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988660521614574e-05,
+      "loss": 1.1575,
+      "step": 374
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9885605855918887e-05,
+      "loss": 1.2907,
+      "step": 375
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9884602136589102e-05,
+      "loss": 1.2755,
+      "step": 376
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988359405859898e-05,
+      "loss": 1.269,
+      "step": 377
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9882581622393036e-05,
+      "loss": 1.3032,
+      "step": 378
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988156482841771e-05,
+      "loss": 1.311,
+      "step": 379
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9880543677121358e-05,
+      "loss": 1.1817,
+      "step": 380
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9879518168954265e-05,
+      "loss": 1.1642,
+      "step": 381
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9878488304368635e-05,
+      "loss": 1.3636,
+      "step": 382
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9877454083818587e-05,
+      "loss": 1.3806,
+      "step": 383
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9876415507760164e-05,
+      "loss": 1.4995,
+      "step": 384
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987537257665134e-05,
+      "loss": 1.2202,
+      "step": 385
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987432529095199e-05,
+      "loss": 1.3787,
+      "step": 386
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9873273651123925e-05,
+      "loss": 1.311,
+      "step": 387
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987221765763087e-05,
+      "loss": 1.3958,
+      "step": 388
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9871157310938466e-05,
+      "loss": 1.4219,
+      "step": 389
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9870092611514287e-05,
+      "loss": 1.0922,
+      "step": 390
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9869023559827807e-05,
+      "loss": 1.3746,
+      "step": 391
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9867950156350434e-05,
+      "loss": 1.3511,
+      "step": 392
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986687240155549e-05,
+      "loss": 1.2806,
+      "step": 393
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9865790295918212e-05,
+      "loss": 1.2993,
+      "step": 394
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9864703839915768e-05,
+      "loss": 1.3929,
+      "step": 395
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9863613034027224e-05,
+      "loss": 1.2349,
+      "step": 396
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9862517878733587e-05,
+      "loss": 1.2937,
+      "step": 397
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9861418374517764e-05,
+      "loss": 1.2348,
+      "step": 398
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.986031452186459e-05,
+      "loss": 1.3392,
+      "step": 399
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985920632126081e-05,
+      "loss": 1.3576,
+      "step": 400
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9858093773195093e-05,
+      "loss": 1.3981,
+      "step": 401
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9856976878158024e-05,
+      "loss": 1.2983,
+      "step": 402
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9855855636642097e-05,
+      "loss": 1.3031,
+      "step": 403
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985473004914173e-05,
+      "loss": 1.2745,
+      "step": 404
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985360011615326e-05,
+      "loss": 1.2125,
+      "step": 405
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9852465838174935e-05,
+      "loss": 1.3689,
+      "step": 406
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9851327215706923e-05,
+      "loss": 1.3366,
+      "step": 407
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9850184249251298e-05,
+      "loss": 1.1641,
+      "step": 408
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9849036939312055e-05,
+      "loss": 1.2753,
+      "step": 409
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9847885286395113e-05,
+      "loss": 1.3238,
+      "step": 410
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9846729291008293e-05,
+      "loss": 1.3409,
+      "step": 411
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9845568953661342e-05,
+      "loss": 1.4463,
+      "step": 412
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984440427486591e-05,
+      "loss": 1.1998,
+      "step": 413
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9843235255135572e-05,
+      "loss": 1.21,
+      "step": 414
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9842061894985808e-05,
+      "loss": 1.3378,
+      "step": 415
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.984088419493402e-05,
+      "loss": 1.4075,
+      "step": 416
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983970215549952e-05,
+      "loss": 1.3568,
+      "step": 417
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983851577720353e-05,
+      "loss": 1.2859,
+      "step": 418
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9837325060569197e-05,
+      "loss": 1.2138,
+      "step": 419
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9836130006121565e-05,
+      "loss": 1.2704,
+      "step": 420
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.98349306143876e-05,
+      "loss": 1.1971,
+      "step": 421
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9833726885896185e-05,
+      "loss": 1.1858,
+      "step": 422
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9832518821178103e-05,
+      "loss": 1.3263,
+      "step": 423
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983130642076606e-05,
+      "loss": 1.2812,
+      "step": 424
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9830089685194663e-05,
+      "loss": 1.3682,
+      "step": 425
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9828868615000447e-05,
+      "loss": 1.1222,
+      "step": 426
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982764321072184e-05,
+      "loss": 1.2187,
+      "step": 427
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9826413472899192e-05,
+      "loss": 1.1914,
+      "step": 428
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982517940207476e-05,
+      "loss": 1.2487,
+      "step": 429
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982394099879272e-05,
+      "loss": 1.2583,
+      "step": 430
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9822698263599144e-05,
+      "loss": 1.1972,
+      "step": 431
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9821451197042028e-05,
+      "loss": 1.3612,
+      "step": 432
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9820199799671267e-05,
+      "loss": 1.1928,
+      "step": 433
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981894407203867e-05,
+      "loss": 1.1519,
+      "step": 434
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981768401469796e-05,
+      "loss": 1.2893,
+      "step": 435
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981641962820476e-05,
+      "loss": 1.3017,
+      "step": 436
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981515091311661e-05,
+      "loss": 1.2228,
+      "step": 437
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9813877869992955e-05,
+      "loss": 1.3039,
+      "step": 438
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9812600499395153e-05,
+      "loss": 1.2868,
+      "step": 439
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9811318801886458e-05,
+      "loss": 1.2772,
+      "step": 440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9810032778032043e-05,
+      "loss": 1.2717,
+      "step": 441
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9808742428398996e-05,
+      "loss": 1.3799,
+      "step": 442
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9807447753556287e-05,
+      "loss": 1.3186,
+      "step": 443
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980614875407482e-05,
+      "loss": 1.3679,
+      "step": 444
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980484543052739e-05,
+      "loss": 1.2933,
+      "step": 445
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9803537783488707e-05,
+      "loss": 1.3903,
+      "step": 446
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9802225813535383e-05,
+      "loss": 1.2843,
+      "step": 447
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9800909521245932e-05,
+      "loss": 1.3779,
+      "step": 448
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979958890720078e-05,
+      "loss": 1.2536,
+      "step": 449
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979826397198227e-05,
+      "loss": 1.2989,
+      "step": 450
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979693471617462e-05,
+      "loss": 1.3518,
+      "step": 451
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9795601140363985e-05,
+      "loss": 1.3693,
+      "step": 452
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9794263245138406e-05,
+      "loss": 1.3377,
+      "step": 453
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979292103108783e-05,
+      "loss": 1.3052,
+      "step": 454
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9791574498804124e-05,
+      "loss": 1.4319,
+      "step": 455
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9790223648881036e-05,
+      "loss": 1.1858,
+      "step": 456
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9788868481914234e-05,
+      "loss": 1.0976,
+      "step": 457
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9787508998501285e-05,
+      "loss": 1.2539,
+      "step": 458
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978614519924166e-05,
+      "loss": 1.3354,
+      "step": 459
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9784777084736733e-05,
+      "loss": 1.2611,
+      "step": 460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9783404655589778e-05,
+      "loss": 1.343,
+      "step": 461
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9782027912405973e-05,
+      "loss": 1.2555,
+      "step": 462
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9780646855792404e-05,
+      "loss": 1.3323,
+      "step": 463
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9779261486358053e-05,
+      "loss": 1.41,
+      "step": 464
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.97778718047138e-05,
+      "loss": 1.273,
+      "step": 465
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9776477811472437e-05,
+      "loss": 1.2696,
+      "step": 466
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9775079507248647e-05,
+      "loss": 1.3091,
+      "step": 467
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9773676892659025e-05,
+      "loss": 1.2212,
+      "step": 468
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977226996832205e-05,
+      "loss": 1.2532,
+      "step": 469
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9770858734858123e-05,
+      "loss": 1.1848,
+      "step": 470
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976944319288953e-05,
+      "loss": 1.1478,
+      "step": 471
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9768023343040457e-05,
+      "loss": 1.2949,
+      "step": 472
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9766599185936995e-05,
+      "loss": 1.3043,
+      "step": 473
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9765170722207135e-05,
+      "loss": 1.2597,
+      "step": 474
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9763737952480763e-05,
+      "loss": 1.2319,
+      "step": 475
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9762300877389665e-05,
+      "loss": 1.1703,
+      "step": 476
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976085949756753e-05,
+      "loss": 1.1467,
+      "step": 477
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9759413813649934e-05,
+      "loss": 1.2557,
+      "step": 478
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9757963826274357e-05,
+      "loss": 1.3335,
+      "step": 479
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9756509536080187e-05,
+      "loss": 1.4156,
+      "step": 480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975505094370869e-05,
+      "loss": 1.3097,
+      "step": 481
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9753588049803048e-05,
+      "loss": 1.2415,
+      "step": 482
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9752120855008326e-05,
+      "loss": 1.1565,
+      "step": 483
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9750649359971487e-05,
+      "loss": 1.3095,
+      "step": 484
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.97491735653414e-05,
+      "loss": 1.3197,
+      "step": 485
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974769347176882e-05,
+      "loss": 1.5948,
+      "step": 486
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.97462090799064e-05,
+      "loss": 1.2156,
+      "step": 487
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974472039040869e-05,
+      "loss": 1.2448,
+      "step": 488
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9743227403932135e-05,
+      "loss": 1.3757,
+      "step": 489
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9741730121135076e-05,
+      "loss": 1.2372,
+      "step": 490
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974022854267774e-05,
+      "loss": 1.2742,
+      "step": 491
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9738722669222268e-05,
+      "loss": 1.2042,
+      "step": 492
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9737212501432667e-05,
+      "loss": 1.3304,
+      "step": 493
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973569803997486e-05,
+      "loss": 1.2688,
+      "step": 494
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9734179285516657e-05,
+      "loss": 1.187,
+      "step": 495
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9732656238727754e-05,
+      "loss": 1.1385,
+      "step": 496
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.973112890027975e-05,
+      "loss": 1.2787,
+      "step": 497
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9729597270846134e-05,
+      "loss": 1.2299,
+      "step": 498
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9728061351102273e-05,
+      "loss": 1.3243,
+      "step": 499
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9726521141725454e-05,
+      "loss": 1.3318,
+      "step": 500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972497664339483e-05,
+      "loss": 1.3425,
+      "step": 501
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9723427856791453e-05,
+      "loss": 1.3175,
+      "step": 502
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9721874782598274e-05,
+      "loss": 1.2496,
+      "step": 503
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9720317421500123e-05,
+      "loss": 1.3312,
+      "step": 504
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9718755774183728e-05,
+      "loss": 1.2349,
+      "step": 505
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9717189841337702e-05,
+      "loss": 1.222,
+      "step": 506
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9715619623652554e-05,
+      "loss": 1.1412,
+      "step": 507
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9714045121820676e-05,
+      "loss": 1.264,
+      "step": 508
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9712466336536353e-05,
+      "loss": 1.1533,
+      "step": 509
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971088326849576e-05,
+      "loss": 1.3338,
+      "step": 510
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.970929591839695e-05,
+      "loss": 1.3665,
+      "step": 511
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.970770428693988e-05,
+      "loss": 1.5099,
+      "step": 512
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9706108374826383e-05,
+      "loss": 1.1882,
+      "step": 513
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9704508182760185e-05,
+      "loss": 1.3452,
+      "step": 514
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9702903711446898e-05,
+      "loss": 1.3224,
+      "step": 515
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9701294961594022e-05,
+      "loss": 1.2438,
+      "step": 516
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969968193391094e-05,
+      "loss": 1.1742,
+      "step": 517
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969806462910893e-05,
+      "loss": 1.2659,
+      "step": 518
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969644304790114e-05,
+      "loss": 1.2785,
+      "step": 519
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9694817191002623e-05,
+      "loss": 1.3755,
+      "step": 520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9693187059130304e-05,
+      "loss": 1.3191,
+      "step": 521
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9691552653002993e-05,
+      "loss": 1.5118,
+      "step": 522
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9689913973341397e-05,
+      "loss": 1.1613,
+      "step": 523
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9688271020868093e-05,
+      "loss": 1.2919,
+      "step": 524
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968662379630755e-05,
+      "loss": 1.2716,
+      "step": 525
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9684972300386122e-05,
+      "loss": 1.3327,
+      "step": 526
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968331653383204e-05,
+      "loss": 1.2249,
+      "step": 527
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9681656497375427e-05,
+      "loss": 1.2693,
+      "step": 528
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9679992191748275e-05,
+      "loss": 1.2976,
+      "step": 529
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9678323617684473e-05,
+      "loss": 1.2077,
+      "step": 530
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967665077591979e-05,
+      "loss": 1.3272,
+      "step": 531
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967497366719186e-05,
+      "loss": 1.2515,
+      "step": 532
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967329229224023e-05,
+      "loss": 1.3169,
+      "step": 533
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9671606651806293e-05,
+      "loss": 1.3118,
+      "step": 534
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966991674663335e-05,
+      "loss": 1.3,
+      "step": 535
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9668222577466568e-05,
+      "loss": 1.297,
+      "step": 536
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9666524145053004e-05,
+      "loss": 1.204,
+      "step": 537
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966482145014158e-05,
+      "loss": 1.317,
+      "step": 538
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9663114493483116e-05,
+      "loss": 1.2996,
+      "step": 539
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.96614032758303e-05,
+      "loss": 1.0686,
+      "step": 540
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9659687797937696e-05,
+      "loss": 1.2577,
+      "step": 541
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9657968060561758e-05,
+      "loss": 1.1256,
+      "step": 542
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965624406446081e-05,
+      "loss": 1.2083,
+      "step": 543
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965451581039506e-05,
+      "loss": 1.1878,
+      "step": 544
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9652783299126578e-05,
+      "loss": 1.3296,
+      "step": 545
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9651046531419335e-05,
+      "loss": 1.2652,
+      "step": 546
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964930550803916e-05,
+      "loss": 1.1391,
+      "step": 547
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964756022975377e-05,
+      "loss": 1.4055,
+      "step": 548
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9645810697332747e-05,
+      "loss": 1.3151,
+      "step": 549
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964405691154756e-05,
+      "loss": 1.264,
+      "step": 550
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9642298873171545e-05,
+      "loss": 1.3329,
+      "step": 551
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9640536582979924e-05,
+      "loss": 1.2639,
+      "step": 552
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963877004174978e-05,
+      "loss": 1.2665,
+      "step": 553
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963699925026008e-05,
+      "loss": 1.3266,
+      "step": 554
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963522420929166e-05,
+      "loss": 1.299,
+      "step": 555
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963344491962724e-05,
+      "loss": 1.3151,
+      "step": 556
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9631661382051396e-05,
+      "loss": 1.2952,
+      "step": 557
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.96298735973506e-05,
+      "loss": 1.3092,
+      "step": 558
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9628081566313165e-05,
+      "loss": 1.2224,
+      "step": 559
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962628528972931e-05,
+      "loss": 1.3079,
+      "step": 560
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9624484768391106e-05,
+      "loss": 1.1844,
+      "step": 561
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9622680003092503e-05,
+      "loss": 1.2185,
+      "step": 562
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.962087099462932e-05,
+      "loss": 1.2537,
+      "step": 563
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961905774379925e-05,
+      "loss": 1.2955,
+      "step": 564
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961724025140185e-05,
+      "loss": 1.2276,
+      "step": 565
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9615418518238553e-05,
+      "loss": 1.2248,
+      "step": 566
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9613592545112657e-05,
+      "loss": 1.2336,
+      "step": 567
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9611762332829342e-05,
+      "loss": 1.2544,
+      "step": 568
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9609927882195637e-05,
+      "loss": 1.3216,
+      "step": 569
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960808919402046e-05,
+      "loss": 1.4293,
+      "step": 570
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960624626911459e-05,
+      "loss": 1.2691,
+      "step": 571
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9604399108290667e-05,
+      "loss": 1.2707,
+      "step": 572
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9602547712363205e-05,
+      "loss": 1.3425,
+      "step": 573
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960069208214859e-05,
+      "loss": 1.1966,
+      "step": 574
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9598832218465068e-05,
+      "loss": 1.1882,
+      "step": 575
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9596968122132757e-05,
+      "loss": 1.2799,
+      "step": 576
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9595099793973637e-05,
+      "loss": 1.2507,
+      "step": 577
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.959322723481155e-05,
+      "loss": 1.2863,
+      "step": 578
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9591350445472218e-05,
+      "loss": 1.2989,
+      "step": 579
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958946942678322e-05,
+      "loss": 1.2229,
+      "step": 580
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9587584179573995e-05,
+      "loss": 1.4161,
+      "step": 581
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958569470467585e-05,
+      "loss": 1.3351,
+      "step": 582
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9583801002921965e-05,
+      "loss": 1.2184,
+      "step": 583
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9581903075147372e-05,
+      "loss": 1.3426,
+      "step": 584
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9580000922188967e-05,
+      "loss": 1.2418,
+      "step": 585
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957809454488552e-05,
+      "loss": 1.4127,
+      "step": 586
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957618394407765e-05,
+      "loss": 1.1855,
+      "step": 587
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957426912060785e-05,
+      "loss": 1.3518,
+      "step": 588
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957235007532047e-05,
+      "loss": 1.2847,
+      "step": 589
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957042680906172e-05,
+      "loss": 1.214,
+      "step": 590
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9568499322679676e-05,
+      "loss": 1.1127,
+      "step": 591
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9566567617024264e-05,
+      "loss": 1.2976,
+      "step": 592
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956463169294729e-05,
+      "loss": 1.3401,
+      "step": 593
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.95626915513024e-05,
+      "loss": 1.1024,
+      "step": 594
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.956074719294511e-05,
+      "loss": 1.1685,
+      "step": 595
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955879861873279e-05,
+      "loss": 1.3357,
+      "step": 596
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9556845829524683e-05,
+      "loss": 1.286,
+      "step": 597
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9554888826181875e-05,
+      "loss": 1.1725,
+      "step": 598
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9552927609567312e-05,
+      "loss": 1.2905,
+      "step": 599
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9550962180545807e-05,
+      "loss": 1.3029,
+      "step": 600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9548992539984022e-05,
+      "loss": 1.1993,
+      "step": 601
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9547018688750476e-05,
+      "loss": 1.3171,
+      "step": 602
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9545040627715554e-05,
+      "loss": 1.2889,
+      "step": 603
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9543058357751485e-05,
+      "loss": 1.3819,
+      "step": 604
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954107187973237e-05,
+      "loss": 1.3516,
+      "step": 605
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953908119453414e-05,
+      "loss": 1.1809,
+      "step": 606
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953708630303461e-05,
+      "loss": 1.455,
+      "step": 607
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9535087206113426e-05,
+      "loss": 1.2556,
+      "step": 608
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953308390465211e-05,
+      "loss": 1.4086,
+      "step": 609
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.953107639953402e-05,
+      "loss": 1.2076,
+      "step": 610
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9529064691644378e-05,
+      "loss": 1.304,
+      "step": 611
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9527048781870248e-05,
+      "loss": 1.2177,
+      "step": 612
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9525028671100565e-05,
+      "loss": 1.353,
+      "step": 613
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.95230043602261e-05,
+      "loss": 1.1943,
+      "step": 614
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9520975850139485e-05,
+      "loss": 1.2325,
+      "step": 615
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9518943141735197e-05,
+      "loss": 1.3351,
+      "step": 616
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951690623590957e-05,
+      "loss": 1.3052,
+      "step": 617
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9514865133560788e-05,
+      "loss": 1.4012,
+      "step": 618
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9512819835588884e-05,
+      "loss": 1.4172,
+      "step": 619
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9510770342895743e-05,
+      "loss": 1.1872,
+      "step": 620
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950871665638509e-05,
+      "loss": 1.2326,
+      "step": 621
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9506658776962522e-05,
+      "loss": 1.2717,
+      "step": 622
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950459670553546e-05,
+      "loss": 1.1905,
+      "step": 623
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950253044301318e-05,
+      "loss": 1.3779,
+      "step": 624
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9500459990306817e-05,
+      "loss": 1.2169,
+      "step": 625
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.949838534832935e-05,
+      "loss": 1.1622,
+      "step": 626
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949630651799559e-05,
+      "loss": 1.1821,
+      "step": 627
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949422350022222e-05,
+      "loss": 1.2861,
+      "step": 628
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9492136295927743e-05,
+      "loss": 1.1972,
+      "step": 629
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949004490603253e-05,
+      "loss": 1.3694,
+      "step": 630
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9487949331458793e-05,
+      "loss": 1.2334,
+      "step": 631
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9485849573130573e-05,
+      "loss": 1.217,
+      "step": 632
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9483745631973777e-05,
+      "loss": 1.2454,
+      "step": 633
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948163750891614e-05,
+      "loss": 1.4178,
+      "step": 634
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947952520488726e-05,
+      "loss": 1.3406,
+      "step": 635
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9477408720818553e-05,
+      "loss": 1.2682,
+      "step": 636
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9475288057643306e-05,
+      "loss": 1.2777,
+      "step": 637
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9473163216296623e-05,
+      "loss": 1.2345,
+      "step": 638
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947103419771547e-05,
+      "loss": 1.3093,
+      "step": 639
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9468901002838647e-05,
+      "loss": 1.1988,
+      "step": 640
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946676363260679e-05,
+      "loss": 1.3089,
+      "step": 641
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946462208796239e-05,
+      "loss": 1.2137,
+      "step": 642
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9462476369849767e-05,
+      "loss": 1.3574,
+      "step": 643
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9460326479215085e-05,
+      "loss": 1.3093,
+      "step": 644
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9458172417006347e-05,
+      "loss": 1.2575,
+      "step": 645
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.94560141841734e-05,
+      "loss": 1.4229,
+      "step": 646
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9453851781667926e-05,
+      "loss": 1.2174,
+      "step": 647
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9451685210443443e-05,
+      "loss": 1.0319,
+      "step": 648
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9449514471455312e-05,
+      "loss": 1.3197,
+      "step": 649
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944733956566073e-05,
+      "loss": 1.2852,
+      "step": 650
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9445160494018734e-05,
+      "loss": 1.3261,
+      "step": 651
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9442977257490193e-05,
+      "loss": 1.3305,
+      "step": 652
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944078985703782e-05,
+      "loss": 1.2851,
+      "step": 653
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9438598293626148e-05,
+      "loss": 1.2109,
+      "step": 654
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943640256822157e-05,
+      "loss": 1.2841,
+      "step": 655
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9434202681792293e-05,
+      "loss": 1.1969,
+      "step": 656
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9431998635308372e-05,
+      "loss": 1.3162,
+      "step": 657
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9429790429741682e-05,
+      "loss": 1.3378,
+      "step": 658
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9427578066065954e-05,
+      "loss": 1.3083,
+      "step": 659
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942536154525673e-05,
+      "loss": 1.0642,
+      "step": 660
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9423140868291397e-05,
+      "loss": 1.3075,
+      "step": 661
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9420916036149177e-05,
+      "loss": 1.2463,
+      "step": 662
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9418687049811116e-05,
+      "loss": 1.3446,
+      "step": 663
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.94164539102601e-05,
+      "loss": 1.1994,
+      "step": 664
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9414216618480834e-05,
+      "loss": 1.2535,
+      "step": 665
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9411975175459864e-05,
+      "loss": 1.255,
+      "step": 666
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9409729582185576e-05,
+      "loss": 1.2369,
+      "step": 667
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940747983964816e-05,
+      "loss": 1.1912,
+      "step": 668
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9405225948839657e-05,
+      "loss": 1.3812,
+      "step": 669
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9402967910753928e-05,
+      "loss": 1.253,
+      "step": 670
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940070572638667e-05,
+      "loss": 1.2724,
+      "step": 671
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9398439396735396e-05,
+      "loss": 1.3032,
+      "step": 672
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9396168922799463e-05,
+      "loss": 1.2574,
+      "step": 673
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9393894305580042e-05,
+      "loss": 1.2042,
+      "step": 674
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9391615546080132e-05,
+      "loss": 1.338,
+      "step": 675
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938933264530457e-05,
+      "loss": 1.3259,
+      "step": 676
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938704560426001e-05,
+      "loss": 1.2627,
+      "step": 677
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9384754423954927e-05,
+      "loss": 1.1552,
+      "step": 678
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9382459105399634e-05,
+      "loss": 1.2369,
+      "step": 679
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938015964960626e-05,
+      "loss": 1.071,
+      "step": 680
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9377856057588756e-05,
+      "loss": 1.2965,
+      "step": 681
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9375548330362906e-05,
+      "loss": 1.2588,
+      "step": 682
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9373236468946318e-05,
+      "loss": 1.3699,
+      "step": 683
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937092047435841e-05,
+      "loss": 1.3029,
+      "step": 684
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936860034762043e-05,
+      "loss": 1.2985,
+      "step": 685
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9366276089755452e-05,
+      "loss": 1.1404,
+      "step": 686
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9363947701788374e-05,
+      "loss": 1.3703,
+      "step": 687
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9361615184745894e-05,
+      "loss": 1.2048,
+      "step": 688
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.935927853965656e-05,
+      "loss": 1.1683,
+      "step": 689
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9356937767550715e-05,
+      "loss": 1.2682,
+      "step": 690
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9354592869460544e-05,
+      "loss": 1.2895,
+      "step": 691
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9352243846420033e-05,
+      "loss": 1.0943,
+      "step": 692
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9349890699464997e-05,
+      "loss": 1.2881,
+      "step": 693
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934753342963307e-05,
+      "loss": 1.2054,
+      "step": 694
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9345172037963694e-05,
+      "loss": 1.2015,
+      "step": 695
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934280652549814e-05,
+      "loss": 1.1412,
+      "step": 696
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934043689327949e-05,
+      "loss": 1.2159,
+      "step": 697
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9338063142352644e-05,
+      "loss": 1.243,
+      "step": 698
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9335685273764322e-05,
+      "loss": 1.2194,
+      "step": 699
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9333303288563053e-05,
+      "loss": 1.2734,
+      "step": 700
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933091718779918e-05,
+      "loss": 1.2416,
+      "step": 701
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932852697252487e-05,
+      "loss": 1.368,
+      "step": 702
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.93261326437941e-05,
+      "loss": 1.2133,
+      "step": 703
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932373420266266e-05,
+      "loss": 1.2235,
+      "step": 704
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932133165018815e-05,
+      "loss": 1.2041,
+      "step": 705
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.931892498742999e-05,
+      "loss": 1.2285,
+      "step": 706
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9316514215449405e-05,
+      "loss": 1.2559,
+      "step": 707
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.931409933530944e-05,
+      "loss": 1.2379,
+      "step": 708
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9311680348074948e-05,
+      "loss": 1.3007,
+      "step": 709
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9309257254812587e-05,
+      "loss": 1.3369,
+      "step": 710
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9306830056590832e-05,
+      "loss": 1.3211,
+      "step": 711
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9304398754479977e-05,
+      "loss": 1.2535,
+      "step": 712
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.93019633495521e-05,
+      "loss": 1.1954,
+      "step": 713
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9299523842881117e-05,
+      "loss": 1.2121,
+      "step": 714
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9297080235542733e-05,
+      "loss": 1.2806,
+      "step": 715
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929463252861447e-05,
+      "loss": 1.3923,
+      "step": 716
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9292180723175656e-05,
+      "loss": 1.3361,
+      "step": 717
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928972482030742e-05,
+      "loss": 1.2523,
+      "step": 718
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928726482109271e-05,
+      "loss": 1.2542,
+      "step": 719
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9284800726616276e-05,
+      "loss": 1.2115,
+      "step": 720
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9282332537964663e-05,
+      "loss": 1.265,
+      "step": 721
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.927986025622624e-05,
+      "loss": 1.2717,
+      "step": 722
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9277383882491162e-05,
+      "loss": 1.2433,
+      "step": 723
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.92749034178514e-05,
+      "loss": 1.2879,
+      "step": 724
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.927241886340073e-05,
+      "loss": 1.3527,
+      "step": 725
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926993022023472e-05,
+      "loss": 1.2951,
+      "step": 726
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926743748945076e-05,
+      "loss": 1.3446,
+      "step": 727
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9264940672148018e-05,
+      "loss": 1.3691,
+      "step": 728
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9262439769427488e-05,
+      "loss": 1.2513,
+      "step": 729
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9259934782391948e-05,
+      "loss": 1.4462,
+      "step": 730
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9257425712145984e-05,
+      "loss": 1.336,
+      "step": 731
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9254912559795984e-05,
+      "loss": 1.254,
+      "step": 732
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9252395326450132e-05,
+      "loss": 1.2911,
+      "step": 733
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9249874013218416e-05,
+      "loss": 1.2905,
+      "step": 734
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924734862121262e-05,
+      "loss": 1.3016,
+      "step": 735
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9244819151546325e-05,
+      "loss": 1.187,
+      "step": 736
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924228560533491e-05,
+      "loss": 1.3786,
+      "step": 737
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9239747983695564e-05,
+      "loss": 1.3028,
+      "step": 738
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9237206287747252e-05,
+      "loss": 1.224,
+      "step": 739
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.923466051861075e-05,
+      "loss": 1.3477,
+      "step": 740
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9232110677408625e-05,
+      "loss": 1.3178,
+      "step": 741
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9229556765265248e-05,
+      "loss": 1.3637,
+      "step": 742
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922699878330677e-05,
+      "loss": 1.4178,
+      "step": 743
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9224436732661148e-05,
+      "loss": 1.215,
+      "step": 744
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922187061445813e-05,
+      "loss": 1.2684,
+      "step": 745
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921930042982926e-05,
+      "loss": 1.2131,
+      "step": 746
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9216726179907866e-05,
+      "loss": 1.4188,
+      "step": 747
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9214147865829084e-05,
+      "loss": 1.2906,
+      "step": 748
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9211565488729827e-05,
+      "loss": 1.2497,
+      "step": 749
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9208979049748808e-05,
+      "loss": 1.2564,
+      "step": 750
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9206388550026523e-05,
+      "loss": 1.3523,
+      "step": 751
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9203793990705275e-05,
+      "loss": 1.1271,
+      "step": 752
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920119537292914e-05,
+      "loss": 1.2798,
+      "step": 753
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.919859269784399e-05,
+      "loss": 1.223,
+      "step": 754
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9195985966597495e-05,
+      "loss": 1.3012,
+      "step": 755
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9193375180339094e-05,
+      "loss": 1.2564,
+      "step": 756
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9190760340220033e-05,
+      "loss": 1.1614,
+      "step": 757
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9188141447393335e-05,
+      "loss": 1.2689,
+      "step": 758
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918551850301381e-05,
+      "loss": 1.2139,
+      "step": 759
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918289150823806e-05,
+      "loss": 1.2756,
+      "step": 760
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9180260464224468e-05,
+      "loss": 1.1731,
+      "step": 761
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9177625372133208e-05,
+      "loss": 1.2268,
+      "step": 762
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9174986233126236e-05,
+      "loss": 1.4057,
+      "step": 763
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917234304836729e-05,
+      "loss": 1.2525,
+      "step": 764
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9169695819021893e-05,
+      "loss": 1.1821,
+      "step": 765
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9167044546257357e-05,
+      "loss": 1.387,
+      "step": 766
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9164389231242772e-05,
+      "loss": 1.254,
+      "step": 767
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9161729875149006e-05,
+      "loss": 1.2328,
+      "step": 768
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915906647914872e-05,
+      "loss": 1.1351,
+      "step": 769
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915639904441635e-05,
+      "loss": 1.2521,
+      "step": 770
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9153727572128113e-05,
+      "loss": 1.2962,
+      "step": 771
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9151052063462008e-05,
+      "loss": 1.2482,
+      "step": 772
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9148372519597807e-05,
+      "loss": 1.3951,
+      "step": 773
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9145688941717074e-05,
+      "loss": 1.1146,
+      "step": 774
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914300133100314e-05,
+      "loss": 1.2025,
+      "step": 775
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9140309688641122e-05,
+      "loss": 1.1529,
+      "step": 776
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9137614015817913e-05,
+      "loss": 1.1571,
+      "step": 777
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9134914313722176e-05,
+      "loss": 1.2051,
+      "step": 778
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9132210583544366e-05,
+      "loss": 1.335,
+      "step": 779
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9129502826476697e-05,
+      "loss": 1.1533,
+      "step": 780
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9126791043713175e-05,
+      "loss": 1.245,
+      "step": 781
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912407523644956e-05,
+      "loss": 1.2996,
+      "step": 782
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9121355405883416e-05,
+      "loss": 1.1436,
+      "step": 783
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9118631553214052e-05,
+      "loss": 1.292,
+      "step": 784
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9115903679642567e-05,
+      "loss": 1.2447,
+      "step": 785
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911317178637183e-05,
+      "loss": 1.1622,
+      "step": 786
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9110435874606487e-05,
+      "loss": 1.3115,
+      "step": 787
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.910769594555294e-05,
+      "loss": 1.1748,
+      "step": 788
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.910495200041938e-05,
+      "loss": 1.4203,
+      "step": 789
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9102204040415763e-05,
+      "loss": 1.4214,
+      "step": 790
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909945206675381e-05,
+      "loss": 1.2037,
+      "step": 791
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909669608064702e-05,
+      "loss": 1.2288,
+      "step": 792
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9093936083310653e-05,
+      "loss": 1.4023,
+      "step": 793
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9091172075961746e-05,
+      "loss": 1.3801,
+      "step": 794
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9088404059819106e-05,
+      "loss": 1.3145,
+      "step": 795
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9085632036103292e-05,
+      "loss": 1.2093,
+      "step": 796
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9082856006036644e-05,
+      "loss": 1.2396,
+      "step": 797
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908007597084327e-05,
+      "loss": 1.3575,
+      "step": 798
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9077291931749032e-05,
+      "loss": 1.3067,
+      "step": 799
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907450388998157e-05,
+      "loss": 1.2861,
+      "step": 800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907171184677028e-05,
+      "loss": 1.194,
+      "step": 801
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906891580334633e-05,
+      "loss": 1.3615,
+      "step": 802
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9066115760942648e-05,
+      "loss": 1.2235,
+      "step": 803
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906331172079392e-05,
+      "loss": 1.3965,
+      "step": 804
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9060503684136602e-05,
+      "loss": 1.1772,
+      "step": 805
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9057691652208916e-05,
+      "loss": 1.1761,
+      "step": 806
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9054875626250836e-05,
+      "loss": 1.1822,
+      "step": 807
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9052055607504102e-05,
+      "loss": 1.1966,
+      "step": 808
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9049231597212213e-05,
+      "loss": 1.0136,
+      "step": 809
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9046403596620432e-05,
+      "loss": 1.1737,
+      "step": 810
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9043571606975776e-05,
+      "loss": 1.1958,
+      "step": 811
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9040735629527027e-05,
+      "loss": 1.3651,
+      "step": 812
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9037895665524716e-05,
+      "loss": 1.2767,
+      "step": 813
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9035051716221146e-05,
+      "loss": 1.3415,
+      "step": 814
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9032203782870363e-05,
+      "loss": 1.2727,
+      "step": 815
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902935186672818e-05,
+      "loss": 1.3425,
+      "step": 816
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9026495969052164e-05,
+      "loss": 1.3308,
+      "step": 817
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9023636091101626e-05,
+      "loss": 1.2498,
+      "step": 818
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9020772234137656e-05,
+      "loss": 1.2477,
+      "step": 819
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9017904399423076e-05,
+      "loss": 1.1727,
+      "step": 820
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.901503258822247e-05,
+      "loss": 1.3663,
+      "step": 821
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9012156801802184e-05,
+      "loss": 1.2419,
+      "step": 822
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9009277041430304e-05,
+      "loss": 1.2511,
+      "step": 823
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9006393308376672e-05,
+      "loss": 1.3444,
+      "step": 824
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9003505603912884e-05,
+      "loss": 1.2776,
+      "step": 825
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900061392931229e-05,
+      "loss": 1.2385,
+      "step": 826
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8997718285849983e-05,
+      "loss": 1.2936,
+      "step": 827
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899481867480281e-05,
+      "loss": 1.277,
+      "step": 828
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8991915097449373e-05,
+      "loss": 1.3344,
+      "step": 829
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898900755507001e-05,
+      "loss": 1.1917,
+      "step": 830
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8986096048946826e-05,
+      "loss": 1.2962,
+      "step": 831
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8983180580363648e-05,
+      "loss": 1.1753,
+      "step": 832
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8980261150606075e-05,
+      "loss": 1.1093,
+      "step": 833
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8977337760961444e-05,
+      "loss": 1.3571,
+      "step": 834
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8974410412718836e-05,
+      "loss": 1.3907,
+      "step": 835
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897147910716907e-05,
+      "loss": 1.1492,
+      "step": 836
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.896854384560473e-05,
+      "loss": 1.203,
+      "step": 837
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8965604629320124e-05,
+      "loss": 1.3394,
+      "step": 838
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.896266145961132e-05,
+      "loss": 1.3626,
+      "step": 839
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895971433777612e-05,
+      "loss": 1.2191,
+      "step": 840
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8956763265114065e-05,
+      "loss": 1.2602,
+      "step": 841
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8953808242926454e-05,
+      "loss": 1.4312,
+      "step": 842
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895084927251631e-05,
+      "loss": 1.2532,
+      "step": 843
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8947886355188406e-05,
+      "loss": 1.3748,
+      "step": 844
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8944919492249254e-05,
+      "loss": 1.3496,
+      "step": 845
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894194868500711e-05,
+      "loss": 1.2452,
+      "step": 846
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8938973934771956e-05,
+      "loss": 1.2914,
+      "step": 847
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893599524285553e-05,
+      "loss": 1.237,
+      "step": 848
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8933012610571295e-05,
+      "loss": 1.2573,
+      "step": 849
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893002603923446e-05,
+      "loss": 1.2609,
+      "step": 850
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8927035530161963e-05,
+      "loss": 1.1312,
+      "step": 851
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8924041084672486e-05,
+      "loss": 1.2525,
+      "step": 852
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8921042704086443e-05,
+      "loss": 1.4048,
+      "step": 853
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.891804038972598e-05,
+      "loss": 1.3098,
+      "step": 854
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8915034142914988e-05,
+      "loss": 1.3778,
+      "step": 855
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8912023964979077e-05,
+      "loss": 1.1837,
+      "step": 856
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8909009857245606e-05,
+      "loss": 1.2027,
+      "step": 857
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8905991821043656e-05,
+      "loss": 1.4035,
+      "step": 858
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8902969857704045e-05,
+      "loss": 1.2739,
+      "step": 859
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8899943968559318e-05,
+      "loss": 1.2731,
+      "step": 860
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8896914154943756e-05,
+      "loss": 1.3024,
+      "step": 861
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8893880418193377e-05,
+      "loss": 1.149,
+      "step": 862
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.889084275964591e-05,
+      "loss": 1.0789,
+      "step": 863
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8887801180640826e-05,
+      "loss": 1.2362,
+      "step": 864
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8884755682519328e-05,
+      "loss": 1.1866,
+      "step": 865
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8881706266624344e-05,
+      "loss": 1.2682,
+      "step": 866
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887865293430052e-05,
+      "loss": 1.2847,
+      "step": 867
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8875595686894242e-05,
+      "loss": 1.2285,
+      "step": 868
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8872534525753617e-05,
+      "loss": 1.3632,
+      "step": 869
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8869469452228476e-05,
+      "loss": 1.3218,
+      "step": 870
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.886640046767038e-05,
+      "loss": 1.3538,
+      "step": 871
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8863327573432605e-05,
+      "loss": 1.1445,
+      "step": 872
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8860250770870167e-05,
+      "loss": 1.1636,
+      "step": 873
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8857170061339793e-05,
+      "loss": 1.2624,
+      "step": 874
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885408544619993e-05,
+      "loss": 1.4176,
+      "step": 875
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885099692681076e-05,
+      "loss": 1.3536,
+      "step": 876
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8847904504534176e-05,
+      "loss": 1.2788,
+      "step": 877
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8844808180733797e-05,
+      "loss": 1.3104,
+      "step": 878
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8841707956774965e-05,
+      "loss": 1.3141,
+      "step": 879
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8838603834024728e-05,
+      "loss": 1.2813,
+      "step": 880
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883549581385187e-05,
+      "loss": 1.1912,
+      "step": 881
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8832383897626892e-05,
+      "loss": 1.3535,
+      "step": 882
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8829268086721993e-05,
+      "loss": 1.2004,
+      "step": 883
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882614838251112e-05,
+      "loss": 1.4677,
+      "step": 884
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882302478636991e-05,
+      "loss": 1.3602,
+      "step": 885
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8819897299675728e-05,
+      "loss": 1.1969,
+      "step": 886
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8816765923807655e-05,
+      "loss": 1.267,
+      "step": 887
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.881363066014649e-05,
+      "loss": 1.2516,
+      "step": 888
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8810491510074734e-05,
+      "loss": 1.3133,
+      "step": 889
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880734847497662e-05,
+      "loss": 1.3884,
+      "step": 890
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8804201556238068e-05,
+      "loss": 1.5093,
+      "step": 891
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880105075524674e-05,
+      "loss": 1.1596,
+      "step": 892
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8797896073391994e-05,
+      "loss": 1.2629,
+      "step": 893
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879473751206489e-05,
+      "loss": 1.2772,
+      "step": 894
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8791575072658225e-05,
+      "loss": 1.293,
+      "step": 895
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8788408756566484e-05,
+      "loss": 1.1388,
+      "step": 896
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.878523856518587e-05,
+      "loss": 1.2399,
+      "step": 897
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8782064499914288e-05,
+      "loss": 1.0937,
+      "step": 898
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8778886562151365e-05,
+      "loss": 1.3329,
+      "step": 899
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8775704753298423e-05,
+      "loss": 1.3333,
+      "step": 900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.877251907475849e-05,
+      "loss": 1.2266,
+      "step": 901
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8769329527936316e-05,
+      "loss": 1.4206,
+      "step": 902
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.876613611423834e-05,
+      "loss": 1.3477,
+      "step": 903
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8762938835072713e-05,
+      "loss": 1.1246,
+      "step": 904
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875973769184929e-05,
+      "loss": 1.27,
+      "step": 905
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875653268597963e-05,
+      "loss": 1.3132,
+      "step": 906
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875332381887699e-05,
+      "loss": 1.4268,
+      "step": 907
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8750111091956345e-05,
+      "loss": 1.1397,
+      "step": 908
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8746894506634355e-05,
+      "loss": 1.1716,
+      "step": 909
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8743674064329388e-05,
+      "loss": 1.2865,
+      "step": 910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874044976646151e-05,
+      "loss": 1.2755,
+      "step": 911
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8737221614452497e-05,
+      "loss": 1.3135,
+      "step": 912
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8733989609725815e-05,
+      "loss": 1.2491,
+      "step": 913
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.873075375370663e-05,
+      "loss": 1.1245,
+      "step": 914
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.87275140478218e-05,
+      "loss": 1.3267,
+      "step": 915
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8724270493499902e-05,
+      "loss": 1.3853,
+      "step": 916
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8721023092171187e-05,
+      "loss": 1.3053,
+      "step": 917
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8717771845267615e-05,
+      "loss": 1.1977,
+      "step": 918
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8714516754222835e-05,
+      "loss": 1.2452,
+      "step": 919
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8711257820472197e-05,
+      "loss": 1.3364,
+      "step": 920
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8707995045452744e-05,
+      "loss": 1.2296,
+      "step": 921
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8704728430603206e-05,
+      "loss": 1.2295,
+      "step": 922
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8701457977364016e-05,
+      "loss": 1.3374,
+      "step": 923
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8698183687177295e-05,
+      "loss": 1.1612,
+      "step": 924
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8694905561486855e-05,
+      "loss": 1.2117,
+      "step": 925
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.86916236017382e-05,
+      "loss": 1.2032,
+      "step": 926
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868833780937853e-05,
+      "loss": 1.4152,
+      "step": 927
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868504818585672e-05,
+      "loss": 1.322,
+      "step": 928
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8681754732623355e-05,
+      "loss": 1.249,
+      "step": 929
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8678457451130693e-05,
+      "loss": 1.2818,
+      "step": 930
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8675156342832686e-05,
+      "loss": 1.1283,
+      "step": 931
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.867185140918497e-05,
+      "loss": 1.3203,
+      "step": 932
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.866854265164488e-05,
+      "loss": 1.3528,
+      "step": 933
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8665230071671412e-05,
+      "loss": 1.3009,
+      "step": 934
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.866191367072528e-05,
+      "loss": 1.2906,
+      "step": 935
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8658593450268853e-05,
+      "loss": 1.2935,
+      "step": 936
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8655269411766208e-05,
+      "loss": 1.1679,
+      "step": 937
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8651941556683083e-05,
+      "loss": 1.3061,
+      "step": 938
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8648609886486923e-05,
+      "loss": 1.1926,
+      "step": 939
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8645274402646837e-05,
+      "loss": 1.112,
+      "step": 940
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8641935106633622e-05,
+      "loss": 1.2456,
+      "step": 941
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8638591999919755e-05,
+      "loss": 1.1775,
+      "step": 942
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8635245083979394e-05,
+      "loss": 1.2434,
+      "step": 943
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8631894360288382e-05,
+      "loss": 1.3155,
+      "step": 944
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.862853983032423e-05,
+      "loss": 1.1456,
+      "step": 945
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8625181495566137e-05,
+      "loss": 1.2565,
+      "step": 946
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8621819357494972e-05,
+      "loss": 1.506,
+      "step": 947
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8618453417593287e-05,
+      "loss": 1.2745,
+      "step": 948
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861508367734531e-05,
+      "loss": 1.1212,
+      "step": 949
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8611710138236943e-05,
+      "loss": 1.2468,
+      "step": 950
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8608332801755764e-05,
+      "loss": 1.1272,
+      "step": 951
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.860495166939102e-05,
+      "loss": 1.2709,
+      "step": 952
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8601566742633645e-05,
+      "loss": 1.3757,
+      "step": 953
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.859817802297623e-05,
+      "loss": 1.2216,
+      "step": 954
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8594785511913048e-05,
+      "loss": 1.301,
+      "step": 955
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.859138921094005e-05,
+      "loss": 1.3413,
+      "step": 956
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8587989121554837e-05,
+      "loss": 1.2593,
+      "step": 957
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8584585245256707e-05,
+      "loss": 1.2081,
+      "step": 958
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8581177583546606e-05,
+      "loss": 1.1673,
+      "step": 959
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8577766137927163e-05,
+      "loss": 1.3634,
+      "step": 960
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8574350909902663e-05,
+      "loss": 1.2523,
+      "step": 961
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8570931900979077e-05,
+      "loss": 1.3753,
+      "step": 962
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8567509112664024e-05,
+      "loss": 1.2985,
+      "step": 963
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8564082546466804e-05,
+      "loss": 1.4052,
+      "step": 964
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8560652203898373e-05,
+      "loss": 1.161,
+      "step": 965
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8557218086471356e-05,
+      "loss": 1.3841,
+      "step": 966
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8553780195700044e-05,
+      "loss": 1.2488,
+      "step": 967
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8550338533100394e-05,
+      "loss": 1.3141,
+      "step": 968
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854689310019002e-05,
+      "loss": 1.2295,
+      "step": 969
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8543443898488197e-05,
+      "loss": 1.1333,
+      "step": 970
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.853999092951587e-05,
+      "loss": 1.4165,
+      "step": 971
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8536534194795647e-05,
+      "loss": 1.3529,
+      "step": 972
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.853307369585178e-05,
+      "loss": 1.1698,
+      "step": 973
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8529609434210198e-05,
+      "loss": 1.353,
+      "step": 974
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8526141411398482e-05,
+      "loss": 1.2774,
+      "step": 975
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8522669628945877e-05,
+      "loss": 1.2533,
+      "step": 976
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.851919408838327e-05,
+      "loss": 1.2536,
+      "step": 977
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.851571479124323e-05,
+      "loss": 1.2228,
+      "step": 978
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8512231739059957e-05,
+      "loss": 1.1973,
+      "step": 979
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.850874493336933e-05,
+      "loss": 1.2131,
+      "step": 980
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.850525437570886e-05,
+      "loss": 1.2123,
+      "step": 981
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8501760067617732e-05,
+      "loss": 1.2397,
+      "step": 982
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8498262010636777e-05,
+      "loss": 1.2903,
+      "step": 983
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8494760206308476e-05,
+      "loss": 1.2065,
+      "step": 984
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8491254656176967e-05,
+      "loss": 1.1584,
+      "step": 985
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.848774536178804e-05,
+      "loss": 1.1364,
+      "step": 986
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8484232324689127e-05,
+      "loss": 1.1893,
+      "step": 987
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8480715546429325e-05,
+      "loss": 1.2326,
+      "step": 988
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8477195028559376e-05,
+      "loss": 1.2638,
+      "step": 989
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847367077263166e-05,
+      "loss": 1.2415,
+      "step": 990
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847014278020022e-05,
+      "loss": 1.1726,
+      "step": 991
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.846661105282074e-05,
+      "loss": 1.3022,
+      "step": 992
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.846307559205055e-05,
+      "loss": 1.3035,
+      "step": 993
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8459536399448622e-05,
+      "loss": 1.2339,
+      "step": 994
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8455993476575585e-05,
+      "loss": 1.2438,
+      "step": 995
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8452446824993704e-05,
+      "loss": 1.3863,
+      "step": 996
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.84488964462669e-05,
+      "loss": 1.3351,
+      "step": 997
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8445342341960715e-05,
+      "loss": 1.2146,
+      "step": 998
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.844178451364236e-05,
+      "loss": 1.1703,
+      "step": 999
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8438222962880667e-05,
+      "loss": 1.1103,
+      "step": 1000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8434657691246123e-05,
+      "loss": 1.2725,
+      "step": 1001
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8431088700310846e-05,
+      "loss": 1.1289,
+      "step": 1002
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8427515991648604e-05,
+      "loss": 1.1428,
+      "step": 1003
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8423939566834796e-05,
+      "loss": 1.3072,
+      "step": 1004
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.842035942744646e-05,
+      "loss": 1.2262,
+      "step": 1005
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8416775575062284e-05,
+      "loss": 1.0779,
+      "step": 1006
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8413188011262573e-05,
+      "loss": 1.2039,
+      "step": 1007
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.840959673762929e-05,
+      "loss": 1.293,
+      "step": 1008
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8406001755746013e-05,
+      "loss": 1.1731,
+      "step": 1009
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8402403067197973e-05,
+      "loss": 1.2138,
+      "step": 1010
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8398800673572032e-05,
+      "loss": 1.2255,
+      "step": 1011
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8395194576456668e-05,
+      "loss": 1.2055,
+      "step": 1012
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8391584777442016e-05,
+      "loss": 1.25,
+      "step": 1013
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8387971278119834e-05,
+      "loss": 1.402,
+      "step": 1014
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8384354080083506e-05,
+      "loss": 1.2107,
+      "step": 1015
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8380733184928053e-05,
+      "loss": 1.2571,
+      "step": 1016
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.837710859425013e-05,
+      "loss": 1.1725,
+      "step": 1017
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.837348030964801e-05,
+      "loss": 1.3706,
+      "step": 1018
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.836984833272161e-05,
+      "loss": 1.3172,
+      "step": 1019
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8366212665072455e-05,
+      "loss": 1.3635,
+      "step": 1020
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.836257330830372e-05,
+      "loss": 1.1431,
+      "step": 1021
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.835893026402019e-05,
+      "loss": 1.1481,
+      "step": 1022
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8355283533828285e-05,
+      "loss": 1.2331,
+      "step": 1023
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8351633119336045e-05,
+      "loss": 1.2815,
+      "step": 1024
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8347979022153137e-05,
+      "loss": 1.2233,
+      "step": 1025
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8344321243890856e-05,
+      "loss": 1.2385,
+      "step": 1026
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8340659786162107e-05,
+      "loss": 1.3233,
+      "step": 1027
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833699465058144e-05,
+      "loss": 1.2515,
+      "step": 1028
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8333325838765e-05,
+      "loss": 1.3213,
+      "step": 1029
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832965335233057e-05,
+      "loss": 1.3432,
+      "step": 1030
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8325977192897555e-05,
+      "loss": 1.2301,
+      "step": 1031
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8322297362086972e-05,
+      "loss": 1.3068,
+      "step": 1032
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8318613861521456e-05,
+      "loss": 1.3421,
+      "step": 1033
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8314926692825264e-05,
+      "loss": 1.3102,
+      "step": 1034
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.831123585762427e-05,
+      "loss": 1.2488,
+      "step": 1035
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8307541357545966e-05,
+      "loss": 1.1844,
+      "step": 1036
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.830384319421946e-05,
+      "loss": 1.2557,
+      "step": 1037
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8300141369275472e-05,
+      "loss": 1.3052,
+      "step": 1038
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8296435884346337e-05,
+      "loss": 1.3491,
+      "step": 1039
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8292726741066008e-05,
+      "loss": 1.2583,
+      "step": 1040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8289013941070046e-05,
+      "loss": 1.2532,
+      "step": 1041
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828529748599563e-05,
+      "loss": 1.2769,
+      "step": 1042
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8281577377481546e-05,
+      "loss": 1.3971,
+      "step": 1043
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8277853617168197e-05,
+      "loss": 1.3062,
+      "step": 1044
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8274126206697584e-05,
+      "loss": 1.3041,
+      "step": 1045
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8270395147713334e-05,
+      "loss": 1.3453,
+      "step": 1046
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8266660441860666e-05,
+      "loss": 1.256,
+      "step": 1047
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8262922090786422e-05,
+      "loss": 1.2604,
+      "step": 1048
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8259180096139047e-05,
+      "loss": 1.2167,
+      "step": 1049
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.825543445956858e-05,
+      "loss": 1.2755,
+      "step": 1050
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8251685182726686e-05,
+      "loss": 1.0997,
+      "step": 1051
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8247932267266626e-05,
+      "loss": 1.2128,
+      "step": 1052
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8244175714843256e-05,
+      "loss": 1.2719,
+      "step": 1053
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8240415527113055e-05,
+      "loss": 1.1964,
+      "step": 1054
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.823665170573409e-05,
+      "loss": 1.4168,
+      "step": 1055
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8232884252366036e-05,
+      "loss": 1.2522,
+      "step": 1056
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8229113168670172e-05,
+      "loss": 1.2806,
+      "step": 1057
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8225338456309365e-05,
+      "loss": 1.3415,
+      "step": 1058
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8221560116948103e-05,
+      "loss": 1.1688,
+      "step": 1059
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.821777815225245e-05,
+      "loss": 1.1823,
+      "step": 1060
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8213992563890094e-05,
+      "loss": 1.2792,
+      "step": 1061
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.82102033535303e-05,
+      "loss": 1.2418,
+      "step": 1062
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8206410522843934e-05,
+      "loss": 1.2202,
+      "step": 1063
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8202614073503472e-05,
+      "loss": 1.0525,
+      "step": 1064
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.819881400718297e-05,
+      "loss": 1.1332,
+      "step": 1065
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.819501032555808e-05,
+      "loss": 1.2854,
+      "step": 1066
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.819120303030606e-05,
+      "loss": 1.1196,
+      "step": 1067
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8187392123105752e-05,
+      "loss": 1.2886,
+      "step": 1068
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.818357760563759e-05,
+      "loss": 1.1101,
+      "step": 1069
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8179759479583607e-05,
+      "loss": 1.147,
+      "step": 1070
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.817593774662742e-05,
+      "loss": 1.1052,
+      "step": 1071
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.817211240845424e-05,
+      "loss": 1.3067,
+      "step": 1072
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8168283466750872e-05,
+      "loss": 1.2147,
+      "step": 1073
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8164450923205697e-05,
+      "loss": 1.3917,
+      "step": 1074
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8160614779508702e-05,
+      "loss": 1.2066,
+      "step": 1075
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8156775037351443e-05,
+      "loss": 1.2592,
+      "step": 1076
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8152931698427076e-05,
+      "loss": 1.3777,
+      "step": 1077
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.814908476443034e-05,
+      "loss": 1.271,
+      "step": 1078
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8145234237057556e-05,
+      "loss": 1.2988,
+      "step": 1079
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8141380118006632e-05,
+      "loss": 1.3793,
+      "step": 1080
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.813752240897706e-05,
+      "loss": 1.2792,
+      "step": 1081
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8133661111669916e-05,
+      "loss": 1.4162,
+      "step": 1082
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.812979622778785e-05,
+      "loss": 1.3528,
+      "step": 1083
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.812592775903511e-05,
+      "loss": 1.2111,
+      "step": 1084
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8122055707117505e-05,
+      "loss": 1.3651,
+      "step": 1085
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8118180073742442e-05,
+      "loss": 1.5163,
+      "step": 1086
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8114300860618892e-05,
+      "loss": 1.2125,
+      "step": 1087
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8110418069457418e-05,
+      "loss": 1.1985,
+      "step": 1088
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.810653170197015e-05,
+      "loss": 1.2017,
+      "step": 1089
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8102641759870805e-05,
+      "loss": 1.2396,
+      "step": 1090
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8098748244874665e-05,
+      "loss": 1.1221,
+      "step": 1091
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8094851158698597e-05,
+      "loss": 1.1907,
+      "step": 1092
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8090950503061036e-05,
+      "loss": 1.2027,
+      "step": 1093
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8087046279681992e-05,
+      "loss": 1.1988,
+      "step": 1094
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8083138490283057e-05,
+      "loss": 1.1468,
+      "step": 1095
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.807922713658738e-05,
+      "loss": 1.2817,
+      "step": 1096
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.80753122203197e-05,
+      "loss": 1.2056,
+      "step": 1097
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.807139374320631e-05,
+      "loss": 1.1878,
+      "step": 1098
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8067471706975084e-05,
+      "loss": 1.153,
+      "step": 1099
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8063546113355458e-05,
+      "loss": 1.0988,
+      "step": 1100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8059616964078443e-05,
+      "loss": 1.2412,
+      "step": 1101
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8055684260876614e-05,
+      "loss": 1.237,
+      "step": 1102
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8051748005484118e-05,
+      "loss": 1.3164,
+      "step": 1103
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.804780819963666e-05,
+      "loss": 1.3637,
+      "step": 1104
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8043864845071514e-05,
+      "loss": 1.3529,
+      "step": 1105
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.803991794352753e-05,
+      "loss": 1.2715,
+      "step": 1106
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.80359674967451e-05,
+      "loss": 1.1334,
+      "step": 1107
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.80320135064662e-05,
+      "loss": 1.3755,
+      "step": 1108
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8028055974434354e-05,
+      "loss": 1.2286,
+      "step": 1109
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.802409490239466e-05,
+      "loss": 1.1384,
+      "step": 1110
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8020130292093768e-05,
+      "loss": 1.1891,
+      "step": 1111
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.801616214527989e-05,
+      "loss": 1.2222,
+      "step": 1112
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.80121904637028e-05,
+      "loss": 1.2596,
+      "step": 1113
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8008215249113826e-05,
+      "loss": 1.3661,
+      "step": 1114
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.800423650326586e-05,
+      "loss": 1.2206,
+      "step": 1115
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8000254227913346e-05,
+      "loss": 1.2147,
+      "step": 1116
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.799626842481229e-05,
+      "loss": 1.243,
+      "step": 1117
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7992279095720245e-05,
+      "loss": 1.2162,
+      "step": 1118
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.798828624239633e-05,
+      "loss": 1.1072,
+      "step": 1119
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7984289866601204e-05,
+      "loss": 1.2211,
+      "step": 1120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7980289970097095e-05,
+      "loss": 1.33,
+      "step": 1121
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7976286554647773e-05,
+      "loss": 1.2091,
+      "step": 1122
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7972279622018557e-05,
+      "loss": 1.3098,
+      "step": 1123
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796826917397633e-05,
+      "loss": 1.3476,
+      "step": 1124
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7964255212289513e-05,
+      "loss": 1.2005,
+      "step": 1125
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796023773872808e-05,
+      "loss": 1.2009,
+      "step": 1126
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.795621675506356e-05,
+      "loss": 1.4053,
+      "step": 1127
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7952192263069018e-05,
+      "loss": 1.1585,
+      "step": 1128
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7948164264519072e-05,
+      "loss": 1.2645,
+      "step": 1129
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.794413276118989e-05,
+      "loss": 1.3151,
+      "step": 1130
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7940097754859177e-05,
+      "loss": 1.3107,
+      "step": 1131
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7936059247306198e-05,
+      "loss": 1.4253,
+      "step": 1132
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7932017240311738e-05,
+      "loss": 1.28,
+      "step": 1133
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7927971735658143e-05,
+      "loss": 1.2183,
+      "step": 1134
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7923922735129303e-05,
+      "loss": 1.238,
+      "step": 1135
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7919870240510634e-05,
+      "loss": 1.2932,
+      "step": 1136
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.791581425358911e-05,
+      "loss": 1.275,
+      "step": 1137
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7911754776153224e-05,
+      "loss": 1.2728,
+      "step": 1138
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.790769180999304e-05,
+      "loss": 1.183,
+      "step": 1139
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.790362535690013e-05,
+      "loss": 1.2565,
+      "step": 1140
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7899555418667614e-05,
+      "loss": 1.1282,
+      "step": 1141
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7895481997090153e-05,
+      "loss": 1.3149,
+      "step": 1142
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.789140509396394e-05,
+      "loss": 1.3334,
+      "step": 1143
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.78873247110867e-05,
+      "loss": 1.2437,
+      "step": 1144
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7883240850257705e-05,
+      "loss": 1.2967,
+      "step": 1145
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.787915351327775e-05,
+      "loss": 1.2552,
+      "step": 1146
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.787506270194916e-05,
+      "loss": 1.3162,
+      "step": 1147
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.78709684180758e-05,
+      "loss": 1.2882,
+      "step": 1148
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7866870663463057e-05,
+      "loss": 1.3458,
+      "step": 1149
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7862769439917867e-05,
+      "loss": 1.4414,
+      "step": 1150
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7858664749248672e-05,
+      "loss": 1.1967,
+      "step": 1151
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785455659326546e-05,
+      "loss": 1.1798,
+      "step": 1152
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785044497377974e-05,
+      "loss": 1.2101,
+      "step": 1153
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7846329892604548e-05,
+      "loss": 1.1152,
+      "step": 1154
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.784221135155445e-05,
+      "loss": 1.4206,
+      "step": 1155
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783808935244553e-05,
+      "loss": 1.2696,
+      "step": 1156
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7833963897095408e-05,
+      "loss": 1.1239,
+      "step": 1157
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.782983498732322e-05,
+      "loss": 1.1932,
+      "step": 1158
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.782570262494963e-05,
+      "loss": 1.2108,
+      "step": 1159
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.782156681179682e-05,
+      "loss": 1.2514,
+      "step": 1160
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7817427549688493e-05,
+      "loss": 1.3212,
+      "step": 1161
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.781328484044988e-05,
+      "loss": 1.2501,
+      "step": 1162
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7809138685907727e-05,
+      "loss": 1.2945,
+      "step": 1163
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.78049890878903e-05,
+      "loss": 1.397,
+      "step": 1164
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7800836048227375e-05,
+      "loss": 1.3768,
+      "step": 1165
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.779667956875027e-05,
+      "loss": 1.2636,
+      "step": 1166
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7792519651291783e-05,
+      "loss": 1.2315,
+      "step": 1167
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7788356297686267e-05,
+      "loss": 1.2228,
+      "step": 1168
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7784189509769562e-05,
+      "loss": 1.246,
+      "step": 1169
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7780019289379033e-05,
+      "loss": 1.3493,
+      "step": 1170
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7775845638353563e-05,
+      "loss": 1.2918,
+      "step": 1171
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7771668558533536e-05,
+      "loss": 1.4218,
+      "step": 1172
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7767488051760858e-05,
+      "loss": 1.229,
+      "step": 1173
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.776330411987894e-05,
+      "loss": 1.2475,
+      "step": 1174
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7759116764732707e-05,
+      "loss": 1.351,
+      "step": 1175
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7754925988168592e-05,
+      "loss": 1.1466,
+      "step": 1176
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.775073179203454e-05,
+      "loss": 1.1176,
+      "step": 1177
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7746534178179997e-05,
+      "loss": 1.3978,
+      "step": 1178
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7742333148455922e-05,
+      "loss": 1.2342,
+      "step": 1179
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7738128704714775e-05,
+      "loss": 1.1547,
+      "step": 1180
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.773392084881053e-05,
+      "loss": 1.197,
+      "step": 1181
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7729709582598655e-05,
+      "loss": 1.3136,
+      "step": 1182
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7725494907936133e-05,
+      "loss": 1.2446,
+      "step": 1183
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7721276826681436e-05,
+      "loss": 1.1431,
+      "step": 1184
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7717055340694555e-05,
+      "loss": 1.2346,
+      "step": 1185
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7712830451836966e-05,
+      "loss": 1.2558,
+      "step": 1186
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7708602161971657e-05,
+      "loss": 1.3096,
+      "step": 1187
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7704370472963107e-05,
+      "loss": 1.304,
+      "step": 1188
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7700135386677306e-05,
+      "loss": 1.2217,
+      "step": 1189
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769589690498173e-05,
+      "loss": 1.2846,
+      "step": 1190
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769165502974536e-05,
+      "loss": 1.3475,
+      "step": 1191
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7687409762838666e-05,
+      "loss": 1.3418,
+      "step": 1192
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7683161106133618e-05,
+      "loss": 1.4199,
+      "step": 1193
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7678909061503685e-05,
+      "loss": 1.2544,
+      "step": 1194
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.767465363082382e-05,
+      "loss": 1.16,
+      "step": 1195
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7670394815970477e-05,
+      "loss": 1.1724,
+      "step": 1196
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7666132618821605e-05,
+      "loss": 1.2067,
+      "step": 1197
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7661867041256628e-05,
+      "loss": 1.1705,
+      "step": 1198
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.765759808515648e-05,
+      "loss": 1.2108,
+      "step": 1199
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7653325752403576e-05,
+      "loss": 1.244,
+      "step": 1200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.764905004488182e-05,
+      "loss": 1.1272,
+      "step": 1201
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7644770964476602e-05,
+      "loss": 1.3263,
+      "step": 1202
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7640488513074804e-05,
+      "loss": 1.3171,
+      "step": 1203
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7636202692564795e-05,
+      "loss": 1.2528,
+      "step": 1204
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.763191350483642e-05,
+      "loss": 1.2339,
+      "step": 1205
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7627620951781024e-05,
+      "loss": 1.3418,
+      "step": 1206
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7623325035291424e-05,
+      "loss": 1.3372,
+      "step": 1207
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7619025757261922e-05,
+      "loss": 1.2942,
+      "step": 1208
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7614723119588305e-05,
+      "loss": 1.1979,
+      "step": 1209
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7610417124167847e-05,
+      "loss": 1.164,
+      "step": 1210
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.760610777289929e-05,
+      "loss": 1.4324,
+      "step": 1211
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.760179506768286e-05,
+      "loss": 1.3036,
+      "step": 1212
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.759747901042027e-05,
+      "loss": 1.4394,
+      "step": 1213
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7593159603014706e-05,
+      "loss": 1.3688,
+      "step": 1214
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7588836847370817e-05,
+      "loss": 1.2283,
+      "step": 1215
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.758451074539476e-05,
+      "loss": 1.1634,
+      "step": 1216
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.758018129899414e-05,
+      "loss": 1.2289,
+      "step": 1217
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7575848510078047e-05,
+      "loss": 1.258,
+      "step": 1218
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7571512380557042e-05,
+      "loss": 1.2453,
+      "step": 1219
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7567172912343165e-05,
+      "loss": 1.395,
+      "step": 1220
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7562830107349923e-05,
+      "loss": 1.3022,
+      "step": 1221
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7558483967492296e-05,
+      "loss": 1.3968,
+      "step": 1222
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.755413449468673e-05,
+      "loss": 1.0975,
+      "step": 1223
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7549781690851148e-05,
+      "loss": 1.3449,
+      "step": 1224
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.754542555790494e-05,
+      "loss": 1.282,
+      "step": 1225
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7541066097768965e-05,
+      "loss": 1.1561,
+      "step": 1226
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.753670331236554e-05,
+      "loss": 1.2498,
+      "step": 1227
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.753233720361846e-05,
+      "loss": 1.1675,
+      "step": 1228
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7527967773452976e-05,
+      "loss": 1.195,
+      "step": 1229
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7523595023795814e-05,
+      "loss": 1.2882,
+      "step": 1230
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7519218956575154e-05,
+      "loss": 1.0588,
+      "step": 1231
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7514839573720644e-05,
+      "loss": 1.3484,
+      "step": 1232
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7510456877163394e-05,
+      "loss": 1.1949,
+      "step": 1233
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7506070868835968e-05,
+      "loss": 1.1581,
+      "step": 1234
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7501681550672406e-05,
+      "loss": 1.4466,
+      "step": 1235
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7497288924608192e-05,
+      "loss": 1.2577,
+      "step": 1236
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7492892992580273e-05,
+      "loss": 1.1783,
+      "step": 1237
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7488493756527058e-05,
+      "loss": 1.1573,
+      "step": 1238
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7484091218388414e-05,
+      "loss": 1.1963,
+      "step": 1239
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.747968538010565e-05,
+      "loss": 1.1939,
+      "step": 1240
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.747527624362155e-05,
+      "loss": 1.1825,
+      "step": 1241
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7470863810880335e-05,
+      "loss": 1.3248,
+      "step": 1242
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7466448083827697e-05,
+      "loss": 1.2497,
+      "step": 1243
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7462029064410763e-05,
+      "loss": 1.2188,
+      "step": 1244
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7457606754578122e-05,
+      "loss": 1.2991,
+      "step": 1245
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7453181156279814e-05,
+      "loss": 1.3713,
+      "step": 1246
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.744875227146732e-05,
+      "loss": 1.3347,
+      "step": 1247
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7444320102093586e-05,
+      "loss": 1.1949,
+      "step": 1248
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.743988465011299e-05,
+      "loss": 1.2162,
+      "step": 1249
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7435445917481368e-05,
+      "loss": 1.372,
+      "step": 1250
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7431003906156e-05,
+      "loss": 1.4101,
+      "step": 1251
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.742655861809561e-05,
+      "loss": 1.258,
+      "step": 1252
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.742211005526037e-05,
+      "loss": 1.2107,
+      "step": 1253
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7417658219611892e-05,
+      "loss": 1.2713,
+      "step": 1254
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.741320311311323e-05,
+      "loss": 1.322,
+      "step": 1255
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7408744737728887e-05,
+      "loss": 1.2039,
+      "step": 1256
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7404283095424803e-05,
+      "loss": 1.1666,
+      "step": 1257
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7399818188168364e-05,
+      "loss": 1.1836,
+      "step": 1258
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7395350017928387e-05,
+      "loss": 1.2688,
+      "step": 1259
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7390878586675127e-05,
+      "loss": 1.1971,
+      "step": 1260
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7386403896380296e-05,
+      "loss": 1.2036,
+      "step": 1261
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7381925949017014e-05,
+      "loss": 1.2798,
+      "step": 1262
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7377444746559865e-05,
+      "loss": 1.2939,
+      "step": 1263
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7372960290984845e-05,
+      "loss": 1.2221,
+      "step": 1264
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.73684725842694e-05,
+      "loss": 1.2502,
+      "step": 1265
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7363981628392405e-05,
+      "loss": 1.0911,
+      "step": 1266
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7359487425334166e-05,
+      "loss": 1.2723,
+      "step": 1267
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.735498997707642e-05,
+      "loss": 1.2206,
+      "step": 1268
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7350489285602345e-05,
+      "loss": 1.2461,
+      "step": 1269
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7345985352896537e-05,
+      "loss": 1.3063,
+      "step": 1270
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7341478180945024e-05,
+      "loss": 1.4491,
+      "step": 1271
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7336967771735266e-05,
+      "loss": 1.371,
+      "step": 1272
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7332454127256148e-05,
+      "loss": 1.1619,
+      "step": 1273
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.732793724949798e-05,
+      "loss": 1.3218,
+      "step": 1274
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7323417140452507e-05,
+      "loss": 1.3033,
+      "step": 1275
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.731889380211288e-05,
+      "loss": 1.097,
+      "step": 1276
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7314367236473697e-05,
+      "loss": 1.185,
+      "step": 1277
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7309837445530958e-05,
+      "loss": 1.3681,
+      "step": 1278
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7305304431282105e-05,
+      "loss": 1.2733,
+      "step": 1279
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7300768195725985e-05,
+      "loss": 1.1696,
+      "step": 1280
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7296228740862872e-05,
+      "loss": 1.0882,
+      "step": 1281
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.729168606869446e-05,
+      "loss": 1.2375,
+      "step": 1282
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.728714018122386e-05,
+      "loss": 1.3172,
+      "step": 1283
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7282591080455602e-05,
+      "loss": 1.4065,
+      "step": 1284
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7278038768395635e-05,
+      "loss": 1.1092,
+      "step": 1285
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7273483247051325e-05,
+      "loss": 1.3763,
+      "step": 1286
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7268924518431437e-05,
+      "loss": 1.1876,
+      "step": 1287
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7264362584546178e-05,
+      "loss": 1.4033,
+      "step": 1288
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7259797447407142e-05,
+      "loss": 1.1305,
+      "step": 1289
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7255229109027357e-05,
+      "loss": 1.3166,
+      "step": 1290
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7250657571421247e-05,
+      "loss": 1.1437,
+      "step": 1291
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.724608283660465e-05,
+      "loss": 1.1972,
+      "step": 1292
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.724150490659482e-05,
+      "loss": 1.2241,
+      "step": 1293
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7236923783410413e-05,
+      "loss": 1.3215,
+      "step": 1294
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7232339469071502e-05,
+      "loss": 1.2481,
+      "step": 1295
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7227751965599556e-05,
+      "loss": 1.2034,
+      "step": 1296
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7223161275017454e-05,
+      "loss": 1.2116,
+      "step": 1297
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.721856739934949e-05,
+      "loss": 1.3471,
+      "step": 1298
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7213970340621346e-05,
+      "loss": 1.1566,
+      "step": 1299
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7209370100860122e-05,
+      "loss": 1.2132,
+      "step": 1300
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7204766682094313e-05,
+      "loss": 1.3215,
+      "step": 1301
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7200160086353815e-05,
+      "loss": 1.2456,
+      "step": 1302
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.719555031566993e-05,
+      "loss": 1.2039,
+      "step": 1303
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7190937372075353e-05,
+      "loss": 1.2767,
+      "step": 1304
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7186321257604186e-05,
+      "loss": 1.2455,
+      "step": 1305
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7181701974291927e-05,
+      "loss": 1.242,
+      "step": 1306
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.717707952417547e-05,
+      "loss": 1.2303,
+      "step": 1307
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7172453909293105e-05,
+      "loss": 1.1022,
+      "step": 1308
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7167825131684516e-05,
+      "loss": 1.2235,
+      "step": 1309
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7163193193390784e-05,
+      "loss": 1.1993,
+      "step": 1310
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.715855809645438e-05,
+      "loss": 1.4159,
+      "step": 1311
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7153919842919176e-05,
+      "loss": 1.2871,
+      "step": 1312
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7149278434830433e-05,
+      "loss": 1.3624,
+      "step": 1313
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.71446338742348e-05,
+      "loss": 1.3527,
+      "step": 1314
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.713998616318031e-05,
+      "loss": 1.2978,
+      "step": 1315
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.71353353037164e-05,
+      "loss": 1.178,
+      "step": 1316
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7130681297893884e-05,
+      "loss": 1.2798,
+      "step": 1317
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7126024147764973e-05,
+      "loss": 1.3131,
+      "step": 1318
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.712136385538325e-05,
+      "loss": 1.2302,
+      "step": 1319
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7116700422803694e-05,
+      "loss": 1.1644,
+      "step": 1320
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.711203385208267e-05,
+      "loss": 1.2062,
+      "step": 1321
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7107364145277924e-05,
+      "loss": 1.5022,
+      "step": 1322
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.710269130444858e-05,
+      "loss": 1.3049,
+      "step": 1323
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7098015331655157e-05,
+      "loss": 1.2515,
+      "step": 1324
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7093336228959538e-05,
+      "loss": 1.2317,
+      "step": 1325
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7088653998424995e-05,
+      "loss": 1.1171,
+      "step": 1326
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7083968642116182e-05,
+      "loss": 1.2671,
+      "step": 1327
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7079280162099127e-05,
+      "loss": 1.2238,
+      "step": 1328
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.707458856044124e-05,
+      "loss": 1.2158,
+      "step": 1329
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.70698938392113e-05,
+      "loss": 1.2222,
+      "step": 1330
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7065196000479468e-05,
+      "loss": 1.2775,
+      "step": 1331
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7060495046317273e-05,
+      "loss": 1.1427,
+      "step": 1332
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.705579097879763e-05,
+      "loss": 1.2428,
+      "step": 1333
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7051083799994813e-05,
+      "loss": 1.2995,
+      "step": 1334
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7046373511984477e-05,
+      "loss": 1.1091,
+      "step": 1335
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7041660116843642e-05,
+      "loss": 1.2485,
+      "step": 1336
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.70369436166507e-05,
+      "loss": 1.2818,
+      "step": 1337
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7032224013485417e-05,
+      "loss": 1.1589,
+      "step": 1338
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7027501309428924e-05,
+      "loss": 1.2742,
+      "step": 1339
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7022775506563715e-05,
+      "loss": 1.2724,
+      "step": 1340
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.701804660697366e-05,
+      "loss": 1.1823,
+      "step": 1341
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.701331461274398e-05,
+      "loss": 1.1463,
+      "step": 1342
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.700857952596128e-05,
+      "loss": 1.1693,
+      "step": 1343
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.700384134871351e-05,
+      "loss": 1.2262,
+      "step": 1344
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6999100083089993e-05,
+      "loss": 1.3528,
+      "step": 1345
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6994355731181413e-05,
+      "loss": 1.3984,
+      "step": 1346
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6989608295079817e-05,
+      "loss": 1.1029,
+      "step": 1347
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.69848577768786e-05,
+      "loss": 1.3276,
+      "step": 1348
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6980104178672528e-05,
+      "loss": 1.3683,
+      "step": 1349
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6975347502557724e-05,
+      "loss": 1.116,
+      "step": 1350
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6970587750631664e-05,
+      "loss": 1.3106,
+      "step": 1351
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6965824924993187e-05,
+      "loss": 1.1846,
+      "step": 1352
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6961059027742474e-05,
+      "loss": 1.1476,
+      "step": 1353
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6956290060981078e-05,
+      "loss": 1.3736,
+      "step": 1354
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.695151802681189e-05,
+      "loss": 1.3091,
+      "step": 1355
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6946742927339165e-05,
+      "loss": 1.3417,
+      "step": 1356
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.69419647646685e-05,
+      "loss": 1.2088,
+      "step": 1357
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.693718354090685e-05,
+      "loss": 1.2333,
+      "step": 1358
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6932399258162518e-05,
+      "loss": 1.2432,
+      "step": 1359
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6927611918545158e-05,
+      "loss": 1.3092,
+      "step": 1360
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.692282152416576e-05,
+      "loss": 1.3019,
+      "step": 1361
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.691802807713668e-05,
+      "loss": 1.2809,
+      "step": 1362
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.691323157957161e-05,
+      "loss": 1.1627,
+      "step": 1363
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6908432033585578e-05,
+      "loss": 1.1423,
+      "step": 1364
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6903629441294978e-05,
+      "loss": 1.3245,
+      "step": 1365
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6898823804817524e-05,
+      "loss": 1.2468,
+      "step": 1366
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6894015126272286e-05,
+      "loss": 1.1985,
+      "step": 1367
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6889203407779678e-05,
+      "loss": 1.2498,
+      "step": 1368
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6884388651461447e-05,
+      "loss": 1.211,
+      "step": 1369
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.687957085944068e-05,
+      "loss": 1.2679,
+      "step": 1370
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6874750033841802e-05,
+      "loss": 1.1873,
+      "step": 1371
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6869926176790583e-05,
+      "loss": 1.4001,
+      "step": 1372
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6865099290414126e-05,
+      "loss": 1.2905,
+      "step": 1373
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6860269376840856e-05,
+      "loss": 1.248,
+      "step": 1374
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6855436438200562e-05,
+      "loss": 1.1956,
+      "step": 1375
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6850600476624343e-05,
+      "loss": 1.2655,
+      "step": 1376
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6845761494244633e-05,
+      "loss": 1.2793,
+      "step": 1377
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6840919493195215e-05,
+      "loss": 1.3864,
+      "step": 1378
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.683607447561118e-05,
+      "loss": 1.3722,
+      "step": 1379
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6831226443628966e-05,
+      "loss": 1.1092,
+      "step": 1380
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6826375399386338e-05,
+      "loss": 1.3173,
+      "step": 1381
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6821521345022377e-05,
+      "loss": 1.3252,
+      "step": 1382
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.681666428267751e-05,
+      "loss": 1.2707,
+      "step": 1383
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6811804214493477e-05,
+      "loss": 1.2008,
+      "step": 1384
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6806941142613344e-05,
+      "loss": 1.3404,
+      "step": 1385
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.680207506918151e-05,
+      "loss": 1.2909,
+      "step": 1386
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6797205996343686e-05,
+      "loss": 1.3875,
+      "step": 1387
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.679233392624692e-05,
+      "loss": 1.2081,
+      "step": 1388
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6787458861039562e-05,
+      "loss": 1.4628,
+      "step": 1389
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6782580802871305e-05,
+      "loss": 1.3938,
+      "step": 1390
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6777699753893148e-05,
+      "loss": 1.1989,
+      "step": 1391
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6772815716257414e-05,
+      "loss": 1.3166,
+      "step": 1392
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6767928692117732e-05,
+      "loss": 1.2398,
+      "step": 1393
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.676303868362907e-05,
+      "loss": 1.188,
+      "step": 1394
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.675814569294769e-05,
+      "loss": 1.1662,
+      "step": 1395
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6753249722231185e-05,
+      "loss": 1.1391,
+      "step": 1396
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.674835077363845e-05,
+      "loss": 1.2803,
+      "step": 1397
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6743448849329702e-05,
+      "loss": 1.3141,
+      "step": 1398
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6738543951466465e-05,
+      "loss": 1.2964,
+      "step": 1399
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.673363608221158e-05,
+      "loss": 1.1799,
+      "step": 1400
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.672872524372919e-05,
+      "loss": 1.1216,
+      "step": 1401
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6723811438184757e-05,
+      "loss": 1.1746,
+      "step": 1402
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6718894667745038e-05,
+      "loss": 1.4241,
+      "step": 1403
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.671397493457811e-05,
+      "loss": 1.3513,
+      "step": 1404
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6709052240853352e-05,
+      "loss": 1.2872,
+      "step": 1405
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.670412658874145e-05,
+      "loss": 1.242,
+      "step": 1406
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6699197980414385e-05,
+      "loss": 1.2643,
+      "step": 1407
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6694266418045458e-05,
+      "loss": 1.257,
+      "step": 1408
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.668933190380926e-05,
+      "loss": 1.3014,
+      "step": 1409
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6684394439881688e-05,
+      "loss": 1.2942,
+      "step": 1410
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.667945402843994e-05,
+      "loss": 1.173,
+      "step": 1411
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6674510671662512e-05,
+      "loss": 1.1785,
+      "step": 1412
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6669564371729196e-05,
+      "loss": 1.2164,
+      "step": 1413
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6664615130821092e-05,
+      "loss": 1.2065,
+      "step": 1414
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.665966295112059e-05,
+      "loss": 1.2364,
+      "step": 1415
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.665470783481137e-05,
+      "loss": 1.2818,
+      "step": 1416
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.664974978407842e-05,
+      "loss": 1.2057,
+      "step": 1417
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.664478880110801e-05,
+      "loss": 1.3148,
+      "step": 1418
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.663982488808771e-05,
+      "loss": 1.3526,
+      "step": 1419
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.663485804720638e-05,
+      "loss": 1.3626,
+      "step": 1420
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6629888280654174e-05,
+      "loss": 1.2505,
+      "step": 1421
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6624915590622527e-05,
+      "loss": 1.2598,
+      "step": 1422
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6619939979304172e-05,
+      "loss": 1.2324,
+      "step": 1423
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6614961448893133e-05,
+      "loss": 1.1618,
+      "step": 1424
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6609980001584708e-05,
+      "loss": 1.1745,
+      "step": 1425
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.660499563957549e-05,
+      "loss": 1.2647,
+      "step": 1426
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.660000836506336e-05,
+      "loss": 1.2389,
+      "step": 1427
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6595018180247478e-05,
+      "loss": 1.0912,
+      "step": 1428
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6590025087328284e-05,
+      "loss": 1.2432,
+      "step": 1429
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6585029088507513e-05,
+      "loss": 1.3731,
+      "step": 1430
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6580030185988167e-05,
+      "loss": 1.3082,
+      "step": 1431
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6575028381974542e-05,
+      "loss": 1.3448,
+      "step": 1432
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6570023678672195e-05,
+      "loss": 1.226,
+      "step": 1433
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6565016078287984e-05,
+      "loss": 1.2947,
+      "step": 1434
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6560005583030028e-05,
+      "loss": 1.1516,
+      "step": 1435
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6554992195107725e-05,
+      "loss": 1.316,
+      "step": 1436
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.654997591673176e-05,
+      "loss": 1.3311,
+      "step": 1437
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6544956750114073e-05,
+      "loss": 1.341,
+      "step": 1438
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6539934697467895e-05,
+      "loss": 1.2004,
+      "step": 1439
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6534909761007723e-05,
+      "loss": 1.1808,
+      "step": 1440
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6529881942949322e-05,
+      "loss": 1.117,
+      "step": 1441
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6524851245509736e-05,
+      "loss": 1.3625,
+      "step": 1442
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.651981767090727e-05,
+      "loss": 1.2426,
+      "step": 1443
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.65147812213615e-05,
+      "loss": 1.3973,
+      "step": 1444
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.650974189909328e-05,
+      "loss": 1.2355,
+      "step": 1445
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6504699706324717e-05,
+      "loss": 1.3985,
+      "step": 1446
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6499654645279184e-05,
+      "loss": 1.1854,
+      "step": 1447
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6494606718181332e-05,
+      "loss": 1.336,
+      "step": 1448
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6489555927257063e-05,
+      "loss": 1.2698,
+      "step": 1449
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6484502274733544e-05,
+      "loss": 1.2386,
+      "step": 1450
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6479445762839214e-05,
+      "loss": 1.0984,
+      "step": 1451
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.647438639380375e-05,
+      "loss": 1.3083,
+      "step": 1452
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6469324169858123e-05,
+      "loss": 1.3035,
+      "step": 1453
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6464259093234535e-05,
+      "loss": 1.185,
+      "step": 1454
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.645919116616645e-05,
+      "loss": 1.2115,
+      "step": 1455
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.64541203908886e-05,
+      "loss": 1.0564,
+      "step": 1456
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.644904676963696e-05,
+      "loss": 1.2826,
+      "step": 1457
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.644397030464877e-05,
+      "loss": 1.3325,
+      "step": 1458
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6438890998162523e-05,
+      "loss": 1.2129,
+      "step": 1459
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6433808852417963e-05,
+      "loss": 1.2721,
+      "step": 1460
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.642872386965608e-05,
+      "loss": 1.1953,
+      "step": 1461
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.642363605211912e-05,
+      "loss": 1.2643,
+      "step": 1462
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6418545402050587e-05,
+      "loss": 1.2125,
+      "step": 1463
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.641345192169522e-05,
+      "loss": 1.2296,
+      "step": 1464
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6408355613299014e-05,
+      "loss": 1.276,
+      "step": 1465
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.640325647910921e-05,
+      "loss": 1.2328,
+      "step": 1466
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.639815452137429e-05,
+      "loss": 1.3783,
+      "step": 1467
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.639304974234399e-05,
+      "loss": 1.1771,
+      "step": 1468
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6387942144269285e-05,
+      "loss": 1.4057,
+      "step": 1469
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6382831729402395e-05,
+      "loss": 1.2946,
+      "step": 1470
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6377718499996776e-05,
+      "loss": 1.3098,
+      "step": 1471
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6372602458307138e-05,
+      "loss": 1.2303,
+      "step": 1472
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6367483606589413e-05,
+      "loss": 1.3207,
+      "step": 1473
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.636236194710079e-05,
+      "loss": 1.2577,
+      "step": 1474
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6357237482099682e-05,
+      "loss": 1.2955,
+      "step": 1475
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6352110213845746e-05,
+      "loss": 1.2906,
+      "step": 1476
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.634698014459988e-05,
+      "loss": 1.2512,
+      "step": 1477
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6341847276624208e-05,
+      "loss": 1.2514,
+      "step": 1478
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.633671161218209e-05,
+      "loss": 1.181,
+      "step": 1479
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6331573153538123e-05,
+      "loss": 1.2836,
+      "step": 1480
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.632643190295813e-05,
+      "loss": 1.0722,
+      "step": 1481
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6321287862709173e-05,
+      "loss": 1.2785,
+      "step": 1482
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.631614103505954e-05,
+      "loss": 1.3254,
+      "step": 1483
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6310991422278744e-05,
+      "loss": 1.2897,
+      "step": 1484
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6305839026637532e-05,
+      "loss": 1.2895,
+      "step": 1485
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.630068385040788e-05,
+      "loss": 1.1154,
+      "step": 1486
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.629552589586298e-05,
+      "loss": 1.1799,
+      "step": 1487
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6290365165277263e-05,
+      "loss": 1.2964,
+      "step": 1488
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6285201660926372e-05,
+      "loss": 1.3743,
+      "step": 1489
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6280035385087177e-05,
+      "loss": 1.3995,
+      "step": 1490
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.627486634003777e-05,
+      "loss": 1.1567,
+      "step": 1491
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.626969452805747e-05,
+      "loss": 1.1801,
+      "step": 1492
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6264519951426805e-05,
+      "loss": 1.2875,
+      "step": 1493
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.625934261242754e-05,
+      "loss": 1.369,
+      "step": 1494
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.625416251334263e-05,
+      "loss": 1.156,
+      "step": 1495
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6248979656456273e-05,
+      "loss": 1.2528,
+      "step": 1496
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.624379404405387e-05,
+      "loss": 1.1661,
+      "step": 1497
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6238605678422045e-05,
+      "loss": 1.2211,
+      "step": 1498
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6233414561848627e-05,
+      "loss": 1.3975,
+      "step": 1499
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.622822069662266e-05,
+      "loss": 1.2514,
+      "step": 1500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6223024085034413e-05,
+      "loss": 1.308,
+      "step": 1501
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6217824729375346e-05,
+      "loss": 1.2657,
+      "step": 1502
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6212622631938137e-05,
+      "loss": 1.2712,
+      "step": 1503
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6207417795016684e-05,
+      "loss": 1.2113,
+      "step": 1504
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6202210220906077e-05,
+      "loss": 1.145,
+      "step": 1505
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.619699991190262e-05,
+      "loss": 1.2772,
+      "step": 1506
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6191786870303824e-05,
+      "loss": 1.2803,
+      "step": 1507
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6186571098408402e-05,
+      "loss": 1.3046,
+      "step": 1508
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6181352598516275e-05,
+      "loss": 1.2812,
+      "step": 1509
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6176131372928563e-05,
+      "loss": 1.2917,
+      "step": 1510
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6170907423947587e-05,
+      "loss": 1.4167,
+      "step": 1511
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6165680753876872e-05,
+      "loss": 1.1596,
+      "step": 1512
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6160451365021146e-05,
+      "loss": 1.3529,
+      "step": 1513
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.615521925968633e-05,
+      "loss": 1.3029,
+      "step": 1514
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.614998444017954e-05,
+      "loss": 1.156,
+      "step": 1515
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6144746908809094e-05,
+      "loss": 1.3841,
+      "step": 1516
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6139506667884513e-05,
+      "loss": 1.2027,
+      "step": 1517
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.61342637197165e-05,
+      "loss": 1.2574,
+      "step": 1518
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6129018066616953e-05,
+      "loss": 1.0603,
+      "step": 1519
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6123769710898976e-05,
+      "loss": 1.1356,
+      "step": 1520
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.611851865487685e-05,
+      "loss": 1.2619,
+      "step": 1521
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6113264900866047e-05,
+      "loss": 1.1479,
+      "step": 1522
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.610800845118324e-05,
+      "loss": 1.3405,
+      "step": 1523
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6102749308146285e-05,
+      "loss": 1.1566,
+      "step": 1524
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6097487474074228e-05,
+      "loss": 1.1651,
+      "step": 1525
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6092222951287292e-05,
+      "loss": 1.1449,
+      "step": 1526
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.608695574210689e-05,
+      "loss": 1.2765,
+      "step": 1527
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6081685848855628e-05,
+      "loss": 1.1197,
+      "step": 1528
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6076413273857288e-05,
+      "loss": 1.2179,
+      "step": 1529
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.607113801943684e-05,
+      "loss": 1.3313,
+      "step": 1530
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6065860087920423e-05,
+      "loss": 1.1923,
+      "step": 1531
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.606057948163537e-05,
+      "loss": 1.1719,
+      "step": 1532
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.605529620291019e-05,
+      "loss": 1.2012,
+      "step": 1533
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6050010254074564e-05,
+      "loss": 1.2692,
+      "step": 1534
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6044721637459354e-05,
+      "loss": 1.2246,
+      "step": 1535
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.603943035539661e-05,
+      "loss": 1.2254,
+      "step": 1536
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.603413641021954e-05,
+      "loss": 1.2103,
+      "step": 1537
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6028839804262528e-05,
+      "loss": 1.244,
+      "step": 1538
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6023540539861144e-05,
+      "loss": 1.1795,
+      "step": 1539
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6018238619352123e-05,
+      "loss": 1.32,
+      "step": 1540
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6012934045073368e-05,
+      "loss": 1.1198,
+      "step": 1541
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6007626819363955e-05,
+      "loss": 1.311,
+      "step": 1542
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.600231694456413e-05,
+      "loss": 1.3614,
+      "step": 1543
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5997004423015305e-05,
+      "loss": 1.2885,
+      "step": 1544
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5991689257060067e-05,
+      "loss": 1.3057,
+      "step": 1545
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.598637144904215e-05,
+      "loss": 1.2142,
+      "step": 1546
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5981051001306482e-05,
+      "loss": 1.2458,
+      "step": 1547
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5975727916199128e-05,
+      "loss": 1.3971,
+      "step": 1548
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.597040219606733e-05,
+      "loss": 1.3148,
+      "step": 1549
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5965073843259483e-05,
+      "loss": 1.2823,
+      "step": 1550
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5959742860125153e-05,
+      "loss": 1.2317,
+      "step": 1551
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5954409249015065e-05,
+      "loss": 1.376,
+      "step": 1552
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5949073012281092e-05,
+      "loss": 1.1889,
+      "step": 1553
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5943734152276277e-05,
+      "loss": 1.2476,
+      "step": 1554
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5938392671354815e-05,
+      "loss": 1.304,
+      "step": 1555
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.593304857187205e-05,
+      "loss": 1.1791,
+      "step": 1556
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5927701856184494e-05,
+      "loss": 1.1954,
+      "step": 1557
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5922352526649803e-05,
+      "loss": 1.2342,
+      "step": 1558
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5917000585626786e-05,
+      "loss": 1.3376,
+      "step": 1559
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.591164603547541e-05,
+      "loss": 1.2126,
+      "step": 1560
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5906288878556784e-05,
+      "loss": 1.2203,
+      "step": 1561
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5900929117233173e-05,
+      "loss": 1.1681,
+      "step": 1562
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.589556675386799e-05,
+      "loss": 1.1899,
+      "step": 1563
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5890201790825788e-05,
+      "loss": 1.3145,
+      "step": 1564
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5884834230472278e-05,
+      "loss": 1.3303,
+      "step": 1565
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.587946407517431e-05,
+      "loss": 1.3382,
+      "step": 1566
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5874091327299873e-05,
+      "loss": 1.2927,
+      "step": 1567
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5868715989218107e-05,
+      "loss": 1.1636,
+      "step": 1568
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5863338063299296e-05,
+      "loss": 1.1421,
+      "step": 1569
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5857957551914853e-05,
+      "loss": 1.25,
+      "step": 1570
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5852574457437345e-05,
+      "loss": 1.139,
+      "step": 1571
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5847188782240473e-05,
+      "loss": 1.286,
+      "step": 1572
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.584180052869907e-05,
+      "loss": 1.1735,
+      "step": 1573
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5836409699189114e-05,
+      "loss": 1.3144,
+      "step": 1574
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5831016296087714e-05,
+      "loss": 1.3825,
+      "step": 1575
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.582562032177312e-05,
+      "loss": 1.3622,
+      "step": 1576
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.582022177862471e-05,
+      "loss": 1.2535,
+      "step": 1577
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5814820669022988e-05,
+      "loss": 1.1628,
+      "step": 1578
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5809416995349607e-05,
+      "loss": 1.2009,
+      "step": 1579
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5804010759987343e-05,
+      "loss": 1.1693,
+      "step": 1580
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5798601965320096e-05,
+      "loss": 1.1548,
+      "step": 1581
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5793190613732893e-05,
+      "loss": 1.3024,
+      "step": 1582
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5787776707611903e-05,
+      "loss": 1.2648,
+      "step": 1583
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.578236024934441e-05,
+      "loss": 1.3231,
+      "step": 1584
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5776941241318823e-05,
+      "loss": 1.226,
+      "step": 1585
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5771519685924684e-05,
+      "loss": 1.1604,
+      "step": 1586
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5766095585552648e-05,
+      "loss": 1.2642,
+      "step": 1587
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.57606689425945e-05,
+      "loss": 1.2761,
+      "step": 1588
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5755239759443135e-05,
+      "loss": 1.3093,
+      "step": 1589
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5749808038492587e-05,
+      "loss": 1.1468,
+      "step": 1590
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5744373782137993e-05,
+      "loss": 1.283,
+      "step": 1591
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.573893699277561e-05,
+      "loss": 1.2408,
+      "step": 1592
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.573349767280282e-05,
+      "loss": 1.2787,
+      "step": 1593
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5728055824618114e-05,
+      "loss": 1.2527,
+      "step": 1594
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5722611450621102e-05,
+      "loss": 1.3031,
+      "step": 1595
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.57171645532125e-05,
+      "loss": 1.3462,
+      "step": 1596
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.571171513479415e-05,
+      "loss": 1.1631,
+      "step": 1597
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5706263197768987e-05,
+      "loss": 1.2179,
+      "step": 1598
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.570080874454108e-05,
+      "loss": 1.2422,
+      "step": 1599
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5695351777515583e-05,
+      "loss": 1.3516,
+      "step": 1600
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.568989229909878e-05,
+      "loss": 1.275,
+      "step": 1601
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.568443031169805e-05,
+      "loss": 1.1432,
+      "step": 1602
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.567896581772188e-05,
+      "loss": 1.1667,
+      "step": 1603
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5673498819579864e-05,
+      "loss": 1.3064,
+      "step": 1604
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5668029319682698e-05,
+      "loss": 1.3033,
+      "step": 1605
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5662557320442187e-05,
+      "loss": 1.1833,
+      "step": 1606
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.565708282427123e-05,
+      "loss": 1.1699,
+      "step": 1607
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5651605833583834e-05,
+      "loss": 1.2384,
+      "step": 1608
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5646126350795104e-05,
+      "loss": 1.177,
+      "step": 1609
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5640644378321236e-05,
+      "loss": 1.3168,
+      "step": 1610
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5635159918579537e-05,
+      "loss": 1.1793,
+      "step": 1611
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5629672973988404e-05,
+      "loss": 1.4475,
+      "step": 1612
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5624183546967323e-05,
+      "loss": 1.3527,
+      "step": 1613
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5618691639936897e-05,
+      "loss": 1.2169,
+      "step": 1614
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.561319725531879e-05,
+      "loss": 1.3323,
+      "step": 1615
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.560770039553579e-05,
+      "loss": 1.2015,
+      "step": 1616
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5602201063011754e-05,
+      "loss": 1.318,
+      "step": 1617
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5596699260171638e-05,
+      "loss": 1.2448,
+      "step": 1618
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5591194989441494e-05,
+      "loss": 1.2983,
+      "step": 1619
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.558568825324845e-05,
+      "loss": 1.1493,
+      "step": 1620
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5580179054020725e-05,
+      "loss": 1.2575,
+      "step": 1621
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5574667394187627e-05,
+      "loss": 1.2342,
+      "step": 1622
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5569153276179546e-05,
+      "loss": 1.2202,
+      "step": 1623
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5563636702427967e-05,
+      "loss": 1.2874,
+      "step": 1624
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5558117675365436e-05,
+      "loss": 1.2245,
+      "step": 1625
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5552596197425596e-05,
+      "loss": 0.8736,
+      "step": 1626
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5547072271043173e-05,
+      "loss": 1.2331,
+      "step": 1627
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5541545898653964e-05,
+      "loss": 1.2597,
+      "step": 1628
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5536017082694846e-05,
+      "loss": 1.1166,
+      "step": 1629
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.553048582560378e-05,
+      "loss": 1.1265,
+      "step": 1630
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5524952129819796e-05,
+      "loss": 1.2837,
+      "step": 1631
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5519415997783003e-05,
+      "loss": 1.0215,
+      "step": 1632
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5513877431934586e-05,
+      "loss": 1.3268,
+      "step": 1633
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5508336434716794e-05,
+      "loss": 1.2241,
+      "step": 1634
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5502793008572964e-05,
+      "loss": 1.1426,
+      "step": 1635
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5497247155947493e-05,
+      "loss": 1.3368,
+      "step": 1636
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.549169887928584e-05,
+      "loss": 1.3124,
+      "step": 1637
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5486148181034555e-05,
+      "loss": 1.1711,
+      "step": 1638
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.548059506364124e-05,
+      "loss": 1.2251,
+      "step": 1639
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5475039529554567e-05,
+      "loss": 1.3264,
+      "step": 1640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5469481581224274e-05,
+      "loss": 1.3101,
+      "step": 1641
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5463921221101156e-05,
+      "loss": 1.179,
+      "step": 1642
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5458358451637093e-05,
+      "loss": 1.203,
+      "step": 1643
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5452793275285006e-05,
+      "loss": 1.3132,
+      "step": 1644
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5447225694498887e-05,
+      "loss": 1.2375,
+      "step": 1645
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5441655711733784e-05,
+      "loss": 1.2359,
+      "step": 1646
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5436083329445807e-05,
+      "loss": 1.176,
+      "step": 1647
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5430508550092123e-05,
+      "loss": 1.2559,
+      "step": 1648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5424931376130957e-05,
+      "loss": 1.2041,
+      "step": 1649
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.541935181002159e-05,
+      "loss": 1.3534,
+      "step": 1650
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.541376985422436e-05,
+      "loss": 1.3674,
+      "step": 1651
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5408185511200645e-05,
+      "loss": 1.3244,
+      "step": 1652
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5402598783412897e-05,
+      "loss": 1.0735,
+      "step": 1653
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.539700967332461e-05,
+      "loss": 1.1969,
+      "step": 1654
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.539141818340031e-05,
+      "loss": 1.3274,
+      "step": 1655
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5385824316105615e-05,
+      "loss": 1.4759,
+      "step": 1656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5380228073907154e-05,
+      "loss": 1.1413,
+      "step": 1657
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.537462945927261e-05,
+      "loss": 1.1673,
+      "step": 1658
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.536902847467073e-05,
+      "loss": 1.2314,
+      "step": 1659
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5363425122571286e-05,
+      "loss": 1.1877,
+      "step": 1660
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.53578194054451e-05,
+      "loss": 1.3135,
+      "step": 1661
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5352211325764045e-05,
+      "loss": 1.2493,
+      "step": 1662
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.534660088600102e-05,
+      "loss": 1.1711,
+      "step": 1663
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5340988088629982e-05,
+      "loss": 1.1755,
+      "step": 1664
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.533537293612592e-05,
+      "loss": 1.2345,
+      "step": 1665
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5329755430964858e-05,
+      "loss": 1.2362,
+      "step": 1666
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.532413557562386e-05,
+      "loss": 1.2395,
+      "step": 1667
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5318513372581026e-05,
+      "loss": 1.1305,
+      "step": 1668
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5312888824315494e-05,
+      "loss": 1.3167,
+      "step": 1669
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5307261933307432e-05,
+      "loss": 1.1739,
+      "step": 1670
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5301632702038047e-05,
+      "loss": 1.2394,
+      "step": 1671
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5296001132989573e-05,
+      "loss": 1.1875,
+      "step": 1672
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5290367228645277e-05,
+      "loss": 1.3243,
+      "step": 1673
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5284730991489448e-05,
+      "loss": 1.2906,
+      "step": 1674
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5279092424007416e-05,
+      "loss": 1.1163,
+      "step": 1675
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.527345152868554e-05,
+      "loss": 1.2642,
+      "step": 1676
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5267808308011183e-05,
+      "loss": 1.252,
+      "step": 1677
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5262162764472758e-05,
+      "loss": 1.1907,
+      "step": 1678
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5256514900559694e-05,
+      "loss": 1.2792,
+      "step": 1679
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.525086471876244e-05,
+      "loss": 1.184,
+      "step": 1680
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.524521222157247e-05,
+      "loss": 1.0359,
+      "step": 1681
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5239557411482278e-05,
+      "loss": 1.2416,
+      "step": 1682
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5233900290985375e-05,
+      "loss": 1.2159,
+      "step": 1683
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5228240862576302e-05,
+      "loss": 1.3218,
+      "step": 1684
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5222579128750604e-05,
+      "loss": 1.2311,
+      "step": 1685
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5216915092004847e-05,
+      "loss": 1.2768,
+      "step": 1686
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5211248754836617e-05,
+      "loss": 1.2902,
+      "step": 1687
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5205580119744512e-05,
+      "loss": 1.2417,
+      "step": 1688
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5199909189228139e-05,
+      "loss": 1.2184,
+      "step": 1689
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5194235965788125e-05,
+      "loss": 1.363,
+      "step": 1690
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.51885604519261e-05,
+      "loss": 1.2324,
+      "step": 1691
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.518288265014471e-05,
+      "loss": 1.2832,
+      "step": 1692
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5177202562947604e-05,
+      "loss": 1.249,
+      "step": 1693
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5171520192839446e-05,
+      "loss": 1.2695,
+      "step": 1694
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5165835542325901e-05,
+      "loss": 1.2904,
+      "step": 1695
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5160148613913643e-05,
+      "loss": 1.2477,
+      "step": 1696
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5154459410110351e-05,
+      "loss": 1.2395,
+      "step": 1697
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5148767933424697e-05,
+      "loss": 1.3,
+      "step": 1698
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5143074186366376e-05,
+      "loss": 1.3208,
+      "step": 1699
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.513737817144606e-05,
+      "loss": 1.3076,
+      "step": 1700
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.513167989117544e-05,
+      "loss": 1.1535,
+      "step": 1701
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5125979348067197e-05,
+      "loss": 1.3448,
+      "step": 1702
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5120276544635008e-05,
+      "loss": 1.2648,
+      "step": 1703
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5114571483393554e-05,
+      "loss": 1.1004,
+      "step": 1704
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5108864166858506e-05,
+      "loss": 1.2429,
+      "step": 1705
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5103154597546532e-05,
+      "loss": 1.2843,
+      "step": 1706
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5097442777975295e-05,
+      "loss": 1.4537,
+      "step": 1707
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5091728710663445e-05,
+      "loss": 1.22,
+      "step": 1708
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5086012398130622e-05,
+      "loss": 1.3561,
+      "step": 1709
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.508029384289747e-05,
+      "loss": 1.2633,
+      "step": 1710
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5074573047485603e-05,
+      "loss": 1.2298,
+      "step": 1711
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5068850014417635e-05,
+      "loss": 1.1871,
+      "step": 1712
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5063124746217165e-05,
+      "loss": 1.2498,
+      "step": 1713
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5057397245408773e-05,
+      "loss": 1.1082,
+      "step": 1714
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5051667514518031e-05,
+      "loss": 1.1745,
+      "step": 1715
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5045935556071487e-05,
+      "loss": 1.3047,
+      "step": 1716
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.504020137259667e-05,
+      "loss": 1.1516,
+      "step": 1717
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5034464966622105e-05,
+      "loss": 1.2164,
+      "step": 1718
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5028726340677278e-05,
+      "loss": 1.2524,
+      "step": 1719
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5022985497292661e-05,
+      "loss": 1.126,
+      "step": 1720
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5017242438999713e-05,
+      "loss": 1.2411,
+      "step": 1721
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5011497168330853e-05,
+      "loss": 1.2603,
+      "step": 1722
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5005749687819488e-05,
+      "loss": 1.2695,
+      "step": 1723
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 1.154,
+      "step": 1724
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4994248107407736e-05,
+      "loss": 1.1598,
+      "step": 1725
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4988494012579018e-05,
+      "loss": 1.0572,
+      "step": 1726
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4982737718051143e-05,
+      "loss": 1.1626,
+      "step": 1727
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4976979226362372e-05,
+      "loss": 1.2144,
+      "step": 1728
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4971218540051942e-05,
+      "loss": 1.0926,
+      "step": 1729
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.496545566166005e-05,
+      "loss": 1.2194,
+      "step": 1730
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4959690593727867e-05,
+      "loss": 1.1498,
+      "step": 1731
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4953923338797525e-05,
+      "loss": 1.275,
+      "step": 1732
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4948153899412116e-05,
+      "loss": 1.255,
+      "step": 1733
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4942382278115713e-05,
+      "loss": 1.2826,
+      "step": 1734
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4936608477453327e-05,
+      "loss": 1.2368,
+      "step": 1735
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4930832499970943e-05,
+      "loss": 1.3384,
+      "step": 1736
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4925054348215514e-05,
+      "loss": 1.2064,
+      "step": 1737
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4919274024734931e-05,
+      "loss": 1.2317,
+      "step": 1738
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.491349153207806e-05,
+      "loss": 1.181,
+      "step": 1739
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4907706872794717e-05,
+      "loss": 1.3355,
+      "step": 1740
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.490192004943567e-05,
+      "loss": 1.2087,
+      "step": 1741
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.489613106455265e-05,
+      "loss": 1.4038,
+      "step": 1742
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4890339920698334e-05,
+      "loss": 1.2942,
+      "step": 1743
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4884546620426354e-05,
+      "loss": 1.1957,
+      "step": 1744
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4878751166291294e-05,
+      "loss": 1.2351,
+      "step": 1745
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4872953560848678e-05,
+      "loss": 1.3788,
+      "step": 1746
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4867153806654997e-05,
+      "loss": 1.2687,
+      "step": 1747
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4861351906267673e-05,
+      "loss": 1.1089,
+      "step": 1748
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4855547862245082e-05,
+      "loss": 1.193,
+      "step": 1749
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4849741677146542e-05,
+      "loss": 1.3947,
+      "step": 1750
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4843933353532318e-05,
+      "loss": 1.1461,
+      "step": 1751
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4838122893963619e-05,
+      "loss": 1.2113,
+      "step": 1752
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4832310301002588e-05,
+      "loss": 1.3801,
+      "step": 1753
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.482649557721232e-05,
+      "loss": 1.3092,
+      "step": 1754
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4820678725156844e-05,
+      "loss": 1.3396,
+      "step": 1755
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4814859747401123e-05,
+      "loss": 1.2184,
+      "step": 1756
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4809038646511062e-05,
+      "loss": 1.2697,
+      "step": 1757
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4803215425053505e-05,
+      "loss": 1.1407,
+      "step": 1758
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4797390085596228e-05,
+      "loss": 1.1809,
+      "step": 1759
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4791562630707941e-05,
+      "loss": 1.2628,
+      "step": 1760
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4785733062958284e-05,
+      "loss": 1.1348,
+      "step": 1761
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4779901384917833e-05,
+      "loss": 1.2753,
+      "step": 1762
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4774067599158092e-05,
+      "loss": 1.2712,
+      "step": 1763
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4768231708251497e-05,
+      "loss": 1.1387,
+      "step": 1764
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4762393714771411e-05,
+      "loss": 1.2668,
+      "step": 1765
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4756553621292123e-05,
+      "loss": 1.2856,
+      "step": 1766
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4750711430388846e-05,
+      "loss": 1.2121,
+      "step": 1767
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4744867144637726e-05,
+      "loss": 1.1659,
+      "step": 1768
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4739020766615826e-05,
+      "loss": 1.2081,
+      "step": 1769
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.473317229890113e-05,
+      "loss": 1.2629,
+      "step": 1770
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4727321744072546e-05,
+      "loss": 1.235,
+      "step": 1771
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4721469104709911e-05,
+      "loss": 1.2068,
+      "step": 1772
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4715614383393963e-05,
+      "loss": 1.2348,
+      "step": 1773
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4709757582706374e-05,
+      "loss": 1.2159,
+      "step": 1774
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4703898705229726e-05,
+      "loss": 1.2464,
+      "step": 1775
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4698037753547513e-05,
+      "loss": 1.1564,
+      "step": 1776
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4692174730244158e-05,
+      "loss": 1.1725,
+      "step": 1777
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4686309637904976e-05,
+      "loss": 1.2378,
+      "step": 1778
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4680442479116215e-05,
+      "loss": 1.2958,
+      "step": 1779
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4674573256465023e-05,
+      "loss": 1.2311,
+      "step": 1780
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4668701972539459e-05,
+      "loss": 1.2208,
+      "step": 1781
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4662828629928496e-05,
+      "loss": 1.2332,
+      "step": 1782
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4656953231222007e-05,
+      "loss": 1.3474,
+      "step": 1783
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4651075779010775e-05,
+      "loss": 1.2013,
+      "step": 1784
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4645196275886497e-05,
+      "loss": 1.1851,
+      "step": 1785
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4639314724441756e-05,
+      "loss": 1.1379,
+      "step": 1786
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4633431127270057e-05,
+      "loss": 1.2975,
+      "step": 1787
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4627545486965799e-05,
+      "loss": 1.1753,
+      "step": 1788
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4621657806124275e-05,
+      "loss": 1.2611,
+      "step": 1789
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4615768087341691e-05,
+      "loss": 1.0825,
+      "step": 1790
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4609876333215143e-05,
+      "loss": 1.2796,
+      "step": 1791
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4603982546342626e-05,
+      "loss": 1.1916,
+      "step": 1792
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4598086729323035e-05,
+      "loss": 1.1297,
+      "step": 1793
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4592188884756155e-05,
+      "loss": 1.3306,
+      "step": 1794
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4586289015242667e-05,
+      "loss": 1.2841,
+      "step": 1795
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4580387123384148e-05,
+      "loss": 1.1954,
+      "step": 1796
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4574483211783061e-05,
+      "loss": 1.1479,
+      "step": 1797
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4568577283042766e-05,
+      "loss": 1.2726,
+      "step": 1798
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4562669339767506e-05,
+      "loss": 1.2302,
+      "step": 1799
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4556759384562418e-05,
+      "loss": 1.0719,
+      "step": 1800
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.455084742003352e-05,
+      "loss": 1.2169,
+      "step": 1801
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4544933448787726e-05,
+      "loss": 1.1918,
+      "step": 1802
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4539017473432822e-05,
+      "loss": 1.2105,
+      "step": 1803
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4533099496577487e-05,
+      "loss": 1.2254,
+      "step": 1804
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4527179520831282e-05,
+      "loss": 1.1565,
+      "step": 1805
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4521257548804644e-05,
+      "loss": 1.1532,
+      "step": 1806
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4515333583108896e-05,
+      "loss": 1.2036,
+      "step": 1807
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4509407626356234e-05,
+      "loss": 1.194,
+      "step": 1808
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.450347968115974e-05,
+      "loss": 1.1375,
+      "step": 1809
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4497549750133366e-05,
+      "loss": 1.2647,
+      "step": 1810
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.449161783589194e-05,
+      "loss": 1.2435,
+      "step": 1811
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4485683941051174e-05,
+      "loss": 1.2581,
+      "step": 1812
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4479748068227637e-05,
+      "loss": 1.2366,
+      "step": 1813
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4473810220038786e-05,
+      "loss": 1.2214,
+      "step": 1814
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.446787039910294e-05,
+      "loss": 1.1971,
+      "step": 1815
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4461928608039285e-05,
+      "loss": 1.2202,
+      "step": 1816
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.445598484946789e-05,
+      "loss": 1.1491,
+      "step": 1817
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4450039126009679e-05,
+      "loss": 1.1063,
+      "step": 1818
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.444409144028644e-05,
+      "loss": 1.189,
+      "step": 1819
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.443814179492084e-05,
+      "loss": 1.2174,
+      "step": 1820
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4432190192536398e-05,
+      "loss": 1.3334,
+      "step": 1821
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.44262366357575e-05,
+      "loss": 1.3194,
+      "step": 1822
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.44202811272094e-05,
+      "loss": 1.176,
+      "step": 1823
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4414323669518194e-05,
+      "loss": 1.2612,
+      "step": 1824
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4408364265310864e-05,
+      "loss": 1.1873,
+      "step": 1825
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4402402917215228e-05,
+      "loss": 1.296,
+      "step": 1826
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.439643962785997e-05,
+      "loss": 1.2333,
+      "step": 1827
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4390474399874637e-05,
+      "loss": 1.2295,
+      "step": 1828
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4384507235889613e-05,
+      "loss": 1.2034,
+      "step": 1829
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4378538138536154e-05,
+      "loss": 1.0904,
+      "step": 1830
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4372567110446359e-05,
+      "loss": 1.0623,
+      "step": 1831
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4366594154253174e-05,
+      "loss": 1.3064,
+      "step": 1832
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4360619272590413e-05,
+      "loss": 1.1948,
+      "step": 1833
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4354642468092714e-05,
+      "loss": 1.2975,
+      "step": 1834
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4348663743395585e-05,
+      "loss": 1.2235,
+      "step": 1835
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.434268310113537e-05,
+      "loss": 1.258,
+      "step": 1836
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4336700543949257e-05,
+      "loss": 1.3102,
+      "step": 1837
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4330716074475287e-05,
+      "loss": 1.2545,
+      "step": 1838
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4324729695352338e-05,
+      "loss": 1.1817,
+      "step": 1839
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4318741409220129e-05,
+      "loss": 1.2568,
+      "step": 1840
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4312751218719224e-05,
+      "loss": 1.2317,
+      "step": 1841
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4306759126491021e-05,
+      "loss": 1.222,
+      "step": 1842
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4300765135177766e-05,
+      "loss": 1.2003,
+      "step": 1843
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4294769247422531e-05,
+      "loss": 1.1973,
+      "step": 1844
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4288771465869236e-05,
+      "loss": 1.272,
+      "step": 1845
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4282771793162626e-05,
+      "loss": 1.2483,
+      "step": 1846
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4276770231948285e-05,
+      "loss": 1.3267,
+      "step": 1847
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4270766784872627e-05,
+      "loss": 1.2739,
+      "step": 1848
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4264761454582902e-05,
+      "loss": 1.2198,
+      "step": 1849
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4258754243727189e-05,
+      "loss": 1.1403,
+      "step": 1850
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4252745154954393e-05,
+      "loss": 1.2264,
+      "step": 1851
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4246734190914247e-05,
+      "loss": 1.3277,
+      "step": 1852
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4240721354257313e-05,
+      "loss": 1.1832,
+      "step": 1853
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4234706647634982e-05,
+      "loss": 1.3564,
+      "step": 1854
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4228690073699467e-05,
+      "loss": 1.3683,
+      "step": 1855
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4222671635103802e-05,
+      "loss": 1.235,
+      "step": 1856
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.421665133450184e-05,
+      "loss": 1.3064,
+      "step": 1857
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.421062917454827e-05,
+      "loss": 1.1837,
+      "step": 1858
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4204605157898581e-05,
+      "loss": 1.2454,
+      "step": 1859
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4198579287209098e-05,
+      "loss": 1.3177,
+      "step": 1860
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4192551565136951e-05,
+      "loss": 1.1169,
+      "step": 1861
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4186521994340094e-05,
+      "loss": 1.1357,
+      "step": 1862
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4180490577477295e-05,
+      "loss": 1.4032,
+      "step": 1863
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4174457317208132e-05,
+      "loss": 1.2988,
+      "step": 1864
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4168422216192998e-05,
+      "loss": 1.2467,
+      "step": 1865
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4162385277093102e-05,
+      "loss": 1.2174,
+      "step": 1866
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4156346502570454e-05,
+      "loss": 1.2236,
+      "step": 1867
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4150305895287888e-05,
+      "loss": 1.2383,
+      "step": 1868
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4144263457909028e-05,
+      "loss": 1.2531,
+      "step": 1869
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4138219193098321e-05,
+      "loss": 1.2424,
+      "step": 1870
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4132173103521012e-05,
+      "loss": 1.2856,
+      "step": 1871
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4126125191843146e-05,
+      "loss": 1.1487,
+      "step": 1872
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4120075460731583e-05,
+      "loss": 1.3326,
+      "step": 1873
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4114023912853977e-05,
+      "loss": 1.1981,
+      "step": 1874
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.410797055087879e-05,
+      "loss": 1.1944,
+      "step": 1875
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4101915377475275e-05,
+      "loss": 1.3124,
+      "step": 1876
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4095858395313483e-05,
+      "loss": 1.1915,
+      "step": 1877
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4089799607064282e-05,
+      "loss": 1.1857,
+      "step": 1878
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4083739015399313e-05,
+      "loss": 1.2132,
+      "step": 1879
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4077676622991022e-05,
+      "loss": 1.1164,
+      "step": 1880
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4071612432512652e-05,
+      "loss": 1.289,
+      "step": 1881
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.406554644663823e-05,
+      "loss": 1.2528,
+      "step": 1882
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4059478668042583e-05,
+      "loss": 1.2295,
+      "step": 1883
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4053409099401325e-05,
+      "loss": 1.3471,
+      "step": 1884
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4047337743390865e-05,
+      "loss": 1.2165,
+      "step": 1885
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4041264602688388e-05,
+      "loss": 1.0489,
+      "step": 1886
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4035189679971874e-05,
+      "loss": 1.1982,
+      "step": 1887
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.402911297792009e-05,
+      "loss": 1.3086,
+      "step": 1888
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4023034499212588e-05,
+      "loss": 1.1943,
+      "step": 1889
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4016954246529697e-05,
+      "loss": 1.2757,
+      "step": 1890
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4010872222552532e-05,
+      "loss": 1.2802,
+      "step": 1891
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.400478842996299e-05,
+      "loss": 1.1986,
+      "step": 1892
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.399870287144375e-05,
+      "loss": 1.203,
+      "step": 1893
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3992615549678262e-05,
+      "loss": 1.1846,
+      "step": 1894
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.398652646735076e-05,
+      "loss": 1.2982,
+      "step": 1895
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3980435627146252e-05,
+      "loss": 1.4168,
+      "step": 1896
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3974343031750524e-05,
+      "loss": 1.2687,
+      "step": 1897
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3968248683850135e-05,
+      "loss": 1.2784,
+      "step": 1898
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3962152586132409e-05,
+      "loss": 1.2999,
+      "step": 1899
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3956054741285452e-05,
+      "loss": 1.2426,
+      "step": 1900
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3949955151998138e-05,
+      "loss": 1.3278,
+      "step": 1901
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3943853820960104e-05,
+      "loss": 1.1362,
+      "step": 1902
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3937750750861767e-05,
+      "loss": 1.0947,
+      "step": 1903
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3931645944394296e-05,
+      "loss": 1.3227,
+      "step": 1904
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3925539404249638e-05,
+      "loss": 1.2466,
+      "step": 1905
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.39194311331205e-05,
+      "loss": 1.2103,
+      "step": 1906
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3913321133700346e-05,
+      "loss": 1.33,
+      "step": 1907
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3907209408683415e-05,
+      "loss": 1.2618,
+      "step": 1908
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3901095960764697e-05,
+      "loss": 1.187,
+      "step": 1909
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3894980792639944e-05,
+      "loss": 1.0519,
+      "step": 1910
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3888863907005669e-05,
+      "loss": 1.319,
+      "step": 1911
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.388274530655914e-05,
+      "loss": 1.3661,
+      "step": 1912
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3876624993998382e-05,
+      "loss": 1.1867,
+      "step": 1913
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3870502972022175e-05,
+      "loss": 1.2603,
+      "step": 1914
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3864379243330046e-05,
+      "loss": 1.2279,
+      "step": 1915
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3858253810622293e-05,
+      "loss": 1.0667,
+      "step": 1916
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3852126676599944e-05,
+      "loss": 1.3578,
+      "step": 1917
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3845997843964792e-05,
+      "loss": 1.0855,
+      "step": 1918
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3839867315419371e-05,
+      "loss": 1.1772,
+      "step": 1919
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3833735093666962e-05,
+      "loss": 1.2188,
+      "step": 1920
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3827601181411605e-05,
+      "loss": 1.2555,
+      "step": 1921
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.382146558135807e-05,
+      "loss": 1.365,
+      "step": 1922
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3815328296211878e-05,
+      "loss": 1.1822,
+      "step": 1923
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.38091893286793e-05,
+      "loss": 1.1166,
+      "step": 1924
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.380304868146733e-05,
+      "loss": 1.2509,
+      "step": 1925
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3796906357283723e-05,
+      "loss": 1.1772,
+      "step": 1926
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.379076235883696e-05,
+      "loss": 1.2723,
+      "step": 1927
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.378461668883627e-05,
+      "loss": 1.1356,
+      "step": 1928
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3778469349991612e-05,
+      "loss": 1.3048,
+      "step": 1929
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3772320345013679e-05,
+      "loss": 1.2439,
+      "step": 1930
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3766169676613907e-05,
+      "loss": 1.2071,
+      "step": 1931
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3760017347504462e-05,
+      "loss": 1.267,
+      "step": 1932
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3753863360398243e-05,
+      "loss": 1.1273,
+      "step": 1933
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3747707718008872e-05,
+      "loss": 1.2479,
+      "step": 1934
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3741550423050711e-05,
+      "loss": 1.1514,
+      "step": 1935
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3735391478238848e-05,
+      "loss": 1.2371,
+      "step": 1936
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3729230886289098e-05,
+      "loss": 1.1888,
+      "step": 1937
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3723068649918e-05,
+      "loss": 1.1876,
+      "step": 1938
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3716904771842825e-05,
+      "loss": 1.2345,
+      "step": 1939
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3710739254781555e-05,
+      "loss": 1.2015,
+      "step": 1940
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3704572101452911e-05,
+      "loss": 1.2902,
+      "step": 1941
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3698403314576325e-05,
+      "loss": 1.1777,
+      "step": 1942
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3692232896871946e-05,
+      "loss": 1.193,
+      "step": 1943
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3686060851060657e-05,
+      "loss": 1.3016,
+      "step": 1944
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3679887179864042e-05,
+      "loss": 1.274,
+      "step": 1945
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3673711886004414e-05,
+      "loss": 1.1774,
+      "step": 1946
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3667534972204795e-05,
+      "loss": 1.2634,
+      "step": 1947
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3661356441188922e-05,
+      "loss": 1.2882,
+      "step": 1948
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3655176295681252e-05,
+      "loss": 1.4004,
+      "step": 1949
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.364899453840694e-05,
+      "loss": 1.2741,
+      "step": 1950
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.364281117209187e-05,
+      "loss": 1.1435,
+      "step": 1951
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3636626199462615e-05,
+      "loss": 1.3145,
+      "step": 1952
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3630439623246474e-05,
+      "loss": 1.3211,
+      "step": 1953
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3624251446171446e-05,
+      "loss": 1.2214,
+      "step": 1954
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3618061670966227e-05,
+      "loss": 1.3427,
+      "step": 1955
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.361187030036024e-05,
+      "loss": 1.2472,
+      "step": 1956
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3605677337083586e-05,
+      "loss": 1.1897,
+      "step": 1957
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3599482783867091e-05,
+      "loss": 1.0197,
+      "step": 1958
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3593286643442265e-05,
+      "loss": 1.2049,
+      "step": 1959
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3587088918541323e-05,
+      "loss": 1.4511,
+      "step": 1960
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3580889611897186e-05,
+      "loss": 1.1889,
+      "step": 1961
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3574688726243462e-05,
+      "loss": 1.1765,
+      "step": 1962
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3568486264314458e-05,
+      "loss": 1.2811,
+      "step": 1963
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3562282228845183e-05,
+      "loss": 1.2324,
+      "step": 1964
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3556076622571332e-05,
+      "loss": 1.2105,
+      "step": 1965
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3549869448229296e-05,
+      "loss": 1.1048,
+      "step": 1966
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3543660708556158e-05,
+      "loss": 1.2027,
+      "step": 1967
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3537450406289686e-05,
+      "loss": 1.2264,
+      "step": 1968
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3531238544168344e-05,
+      "loss": 1.256,
+      "step": 1969
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3525025124931282e-05,
+      "loss": 1.2373,
+      "step": 1970
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.351881015131833e-05,
+      "loss": 1.1611,
+      "step": 1971
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3512593626070021e-05,
+      "loss": 1.2244,
+      "step": 1972
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3506375551927546e-05,
+      "loss": 1.1229,
+      "step": 1973
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3500155931632799e-05,
+      "loss": 1.2716,
+      "step": 1974
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3493934767928352e-05,
+      "loss": 1.1627,
+      "step": 1975
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3487712063557452e-05,
+      "loss": 1.213,
+      "step": 1976
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3481487821264034e-05,
+      "loss": 1.119,
+      "step": 1977
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.34752620437927e-05,
+      "loss": 1.2521,
+      "step": 1978
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3469034733888738e-05,
+      "loss": 1.254,
+      "step": 1979
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3462805894298105e-05,
+      "loss": 1.2861,
+      "step": 1980
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3456575527767446e-05,
+      "loss": 1.2956,
+      "step": 1981
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3450343637044059e-05,
+      "loss": 1.2946,
+      "step": 1982
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3444110224875926e-05,
+      "loss": 1.29,
+      "step": 1983
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3437875294011704e-05,
+      "loss": 1.0777,
+      "step": 1984
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3431638847200708e-05,
+      "loss": 1.2166,
+      "step": 1985
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3425400887192934e-05,
+      "loss": 1.3022,
+      "step": 1986
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3419161416739033e-05,
+      "loss": 1.3143,
+      "step": 1987
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3412920438590331e-05,
+      "loss": 1.1797,
+      "step": 1988
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3406677955498817e-05,
+      "loss": 1.2235,
+      "step": 1989
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3400433970217137e-05,
+      "loss": 1.1779,
+      "step": 1990
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3394188485498611e-05,
+      "loss": 1.2423,
+      "step": 1991
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3387941504097213e-05,
+      "loss": 1.3349,
+      "step": 1992
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3381693028767573e-05,
+      "loss": 1.1783,
+      "step": 1993
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3375443062264987e-05,
+      "loss": 1.2164,
+      "step": 1994
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.336919160734541e-05,
+      "loss": 1.104,
+      "step": 1995
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3362938666765443e-05,
+      "loss": 1.315,
+      "step": 1996
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3356684243282357e-05,
+      "loss": 1.1404,
+      "step": 1997
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3350428339654057e-05,
+      "loss": 1.2083,
+      "step": 1998
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3344170958639124e-05,
+      "loss": 1.2158,
+      "step": 1999
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3337912102996773e-05,
+      "loss": 1.2113,
+      "step": 2000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3331651775486874e-05,
+      "loss": 1.2493,
+      "step": 2001
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3325389978869949e-05,
+      "loss": 1.2183,
+      "step": 2002
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3319126715907165e-05,
+      "loss": 1.2125,
+      "step": 2003
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3312861989360336e-05,
+      "loss": 1.026,
+      "step": 2004
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3306595801991921e-05,
+      "loss": 1.1761,
+      "step": 2005
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3300328156565028e-05,
+      "loss": 1.2136,
+      "step": 2006
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.32940590558434e-05,
+      "loss": 1.297,
+      "step": 2007
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3287788502591425e-05,
+      "loss": 1.3666,
+      "step": 2008
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3281516499574134e-05,
+      "loss": 1.2885,
+      "step": 2009
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3275243049557195e-05,
+      "loss": 1.2828,
+      "step": 2010
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3268968155306913e-05,
+      "loss": 1.1525,
+      "step": 2011
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3262691819590235e-05,
+      "loss": 1.1799,
+      "step": 2012
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3256414045174734e-05,
+      "loss": 1.3087,
+      "step": 2013
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3250134834828627e-05,
+      "loss": 1.2542,
+      "step": 2014
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.324385419132076e-05,
+      "loss": 1.3681,
+      "step": 2015
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3237572117420612e-05,
+      "loss": 1.189,
+      "step": 2016
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.323128861589829e-05,
+      "loss": 1.2312,
+      "step": 2017
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3225003689524534e-05,
+      "loss": 1.1936,
+      "step": 2018
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3218717341070709e-05,
+      "loss": 1.2751,
+      "step": 2019
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3212429573308814e-05,
+      "loss": 1.2024,
+      "step": 2020
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3206140389011463e-05,
+      "loss": 1.2457,
+      "step": 2021
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3199849790951901e-05,
+      "loss": 1.1,
+      "step": 2022
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3193557781904e-05,
+      "loss": 1.2485,
+      "step": 2023
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3187264364642252e-05,
+      "loss": 1.1547,
+      "step": 2024
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.318096954194176e-05,
+      "loss": 1.363,
+      "step": 2025
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3174673316578256e-05,
+      "loss": 1.2887,
+      "step": 2026
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3168375691328094e-05,
+      "loss": 1.1348,
+      "step": 2027
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.316207666896824e-05,
+      "loss": 1.339,
+      "step": 2028
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3155776252276276e-05,
+      "loss": 1.2353,
+      "step": 2029
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3149474444030394e-05,
+      "loss": 1.2197,
+      "step": 2030
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3143171247009414e-05,
+      "loss": 1.106,
+      "step": 2031
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3136866663992753e-05,
+      "loss": 1.1775,
+      "step": 2032
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3130560697760446e-05,
+      "loss": 1.2334,
+      "step": 2033
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3124253351093141e-05,
+      "loss": 1.1932,
+      "step": 2034
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.311794462677209e-05,
+      "loss": 1.2437,
+      "step": 2035
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3111634527579152e-05,
+      "loss": 1.2632,
+      "step": 2036
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3105323056296797e-05,
+      "loss": 1.1803,
+      "step": 2037
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3099010215708087e-05,
+      "loss": 1.2162,
+      "step": 2038
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3092696008596715e-05,
+      "loss": 1.3584,
+      "step": 2039
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3086380437746948e-05,
+      "loss": 1.3145,
+      "step": 2040
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3080063505943665e-05,
+      "loss": 1.2469,
+      "step": 2041
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3073745215972352e-05,
+      "loss": 1.0871,
+      "step": 2042
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3067425570619083e-05,
+      "loss": 1.2558,
+      "step": 2043
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3061104572670538e-05,
+      "loss": 1.2765,
+      "step": 2044
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3054782224913988e-05,
+      "loss": 1.2954,
+      "step": 2045
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3048458530137299e-05,
+      "loss": 1.1148,
+      "step": 2046
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3042133491128934e-05,
+      "loss": 1.1459,
+      "step": 2047
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3035807110677949e-05,
+      "loss": 1.1715,
+      "step": 2048
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3029479391573991e-05,
+      "loss": 1.4073,
+      "step": 2049
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3023150336607298e-05,
+      "loss": 1.1093,
+      "step": 2050
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3016819948568686e-05,
+      "loss": 1.1928,
+      "step": 2051
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3010488230249583e-05,
+      "loss": 1.1906,
+      "step": 2052
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3004155184441978e-05,
+      "loss": 1.1994,
+      "step": 2053
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.299782081393846e-05,
+      "loss": 1.2477,
+      "step": 2054
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.29914851215322e-05,
+      "loss": 1.3453,
+      "step": 2055
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2985148110016947e-05,
+      "loss": 1.2048,
+      "step": 2056
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2978809782187039e-05,
+      "loss": 1.2095,
+      "step": 2057
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2972470140837386e-05,
+      "loss": 1.2725,
+      "step": 2058
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2966129188763485e-05,
+      "loss": 1.1552,
+      "step": 2059
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2959786928761409e-05,
+      "loss": 1.2941,
+      "step": 2060
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2953443363627805e-05,
+      "loss": 1.4365,
+      "step": 2061
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2947098496159894e-05,
+      "loss": 1.2101,
+      "step": 2062
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2940752329155475e-05,
+      "loss": 1.2185,
+      "step": 2063
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2934404865412924e-05,
+      "loss": 1.2173,
+      "step": 2064
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.292805610773118e-05,
+      "loss": 1.1782,
+      "step": 2065
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2921706058909757e-05,
+      "loss": 1.2954,
+      "step": 2066
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2915354721748739e-05,
+      "loss": 1.2999,
+      "step": 2067
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2909002099048775e-05,
+      "loss": 1.2909,
+      "step": 2068
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2902648193611087e-05,
+      "loss": 1.2438,
+      "step": 2069
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2896293008237454e-05,
+      "loss": 1.3469,
+      "step": 2070
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2889936545730226e-05,
+      "loss": 1.3423,
+      "step": 2071
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2883578808892319e-05,
+      "loss": 1.1828,
+      "step": 2072
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2877219800527194e-05,
+      "loss": 1.1383,
+      "step": 2073
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2870859523438894e-05,
+      "loss": 1.164,
+      "step": 2074
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.286449798043201e-05,
+      "loss": 1.1198,
+      "step": 2075
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2858135174311692e-05,
+      "loss": 1.3606,
+      "step": 2076
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2851771107883654e-05,
+      "loss": 1.3283,
+      "step": 2077
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2845405783954152e-05,
+      "loss": 1.2526,
+      "step": 2078
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.283903920533001e-05,
+      "loss": 1.3906,
+      "step": 2079
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2832671374818597e-05,
+      "loss": 1.3495,
+      "step": 2080
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2826302295227836e-05,
+      "loss": 1.2022,
+      "step": 2081
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2819931969366208e-05,
+      "loss": 1.1959,
+      "step": 2082
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.281356040004273e-05,
+      "loss": 1.3407,
+      "step": 2083
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.280718759006698e-05,
+      "loss": 1.2268,
+      "step": 2084
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2800813542249073e-05,
+      "loss": 1.2572,
+      "step": 2085
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2794438259399672e-05,
+      "loss": 1.353,
+      "step": 2086
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2788061744329998e-05,
+      "loss": 1.146,
+      "step": 2087
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2781683999851794e-05,
+      "loss": 1.2713,
+      "step": 2088
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.277530502877736e-05,
+      "loss": 1.327,
+      "step": 2089
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2768924833919533e-05,
+      "loss": 1.2217,
+      "step": 2090
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2762543418091688e-05,
+      "loss": 1.2418,
+      "step": 2091
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.275616078410774e-05,
+      "loss": 1.2374,
+      "step": 2092
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2749776934782134e-05,
+      "loss": 1.3223,
+      "step": 2093
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2743391872929864e-05,
+      "loss": 1.0999,
+      "step": 2094
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2737005601366457e-05,
+      "loss": 1.1242,
+      "step": 2095
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.273061812290796e-05,
+      "loss": 1.1875,
+      "step": 2096
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2724229440370959e-05,
+      "loss": 1.2473,
+      "step": 2097
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.271783955657258e-05,
+      "loss": 1.244,
+      "step": 2098
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.271144847433047e-05,
+      "loss": 1.2817,
+      "step": 2099
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2705056196462801e-05,
+      "loss": 1.1977,
+      "step": 2100
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.269866272578828e-05,
+      "loss": 1.2671,
+      "step": 2101
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2692268065126139e-05,
+      "loss": 1.3965,
+      "step": 2102
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2685872217296131e-05,
+      "loss": 1.1736,
+      "step": 2103
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2679475185118535e-05,
+      "loss": 1.3302,
+      "step": 2104
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2673076971414149e-05,
+      "loss": 1.229,
+      "step": 2105
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2666677579004296e-05,
+      "loss": 1.2037,
+      "step": 2106
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.266027701071082e-05,
+      "loss": 1.1195,
+      "step": 2107
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2653875269356075e-05,
+      "loss": 1.248,
+      "step": 2108
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.264747235776294e-05,
+      "loss": 1.1352,
+      "step": 2109
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2641068278754812e-05,
+      "loss": 1.0824,
+      "step": 2110
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2634663035155594e-05,
+      "loss": 1.2748,
+      "step": 2111
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2628256629789712e-05,
+      "loss": 1.2228,
+      "step": 2112
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2621849065482095e-05,
+      "loss": 1.2161,
+      "step": 2113
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2615440345058189e-05,
+      "loss": 1.2041,
+      "step": 2114
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2609030471343953e-05,
+      "loss": 1.1688,
+      "step": 2115
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.260261944716584e-05,
+      "loss": 1.1402,
+      "step": 2116
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2596207275350833e-05,
+      "loss": 1.3138,
+      "step": 2117
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2589793958726398e-05,
+      "loss": 1.2642,
+      "step": 2118
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2583379500120519e-05,
+      "loss": 1.0298,
+      "step": 2119
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2576963902361685e-05,
+      "loss": 1.4068,
+      "step": 2120
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2570547168278875e-05,
+      "loss": 1.2445,
+      "step": 2121
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2564129300701587e-05,
+      "loss": 1.2844,
+      "step": 2122
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2557710302459803e-05,
+      "loss": 1.0283,
+      "step": 2123
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2551290176384005e-05,
+      "loss": 1.2732,
+      "step": 2124
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.254486892530519e-05,
+      "loss": 1.1732,
+      "step": 2125
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2538446552054824e-05,
+      "loss": 1.2616,
+      "step": 2126
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.253202305946489e-05,
+      "loss": 1.3322,
+      "step": 2127
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2525598450367855e-05,
+      "loss": 1.3484,
+      "step": 2128
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2519172727596676e-05,
+      "loss": 1.2348,
+      "step": 2129
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.251274589398481e-05,
+      "loss": 1.2434,
+      "step": 2130
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2506317952366196e-05,
+      "loss": 1.3359,
+      "step": 2131
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2499888905575261e-05,
+      "loss": 1.0748,
+      "step": 2132
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.249345875644693e-05,
+      "loss": 1.253,
+      "step": 2133
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2487027507816597e-05,
+      "loss": 1.2887,
+      "step": 2134
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.248059516252016e-05,
+      "loss": 1.1531,
+      "step": 2135
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2474161723393987e-05,
+      "loss": 1.164,
+      "step": 2136
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.246772719327493e-05,
+      "loss": 1.1374,
+      "step": 2137
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.246129157500033e-05,
+      "loss": 1.1758,
+      "step": 2138
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2454854871407993e-05,
+      "loss": 1.1748,
+      "step": 2139
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2448417085336219e-05,
+      "loss": 1.2213,
+      "step": 2140
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.244197821962378e-05,
+      "loss": 1.3643,
+      "step": 2141
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2435538277109919e-05,
+      "loss": 1.2794,
+      "step": 2142
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2429097260634367e-05,
+      "loss": 1.1602,
+      "step": 2143
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2422655173037305e-05,
+      "loss": 1.2253,
+      "step": 2144
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2416212017159413e-05,
+      "loss": 1.1774,
+      "step": 2145
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2409767795841826e-05,
+      "loss": 1.2564,
+      "step": 2146
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2403322511926153e-05,
+      "loss": 1.0907,
+      "step": 2147
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2396876168254468e-05,
+      "loss": 1.3112,
+      "step": 2148
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2390428767669318e-05,
+      "loss": 1.142,
+      "step": 2149
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2383980313013715e-05,
+      "loss": 1.2264,
+      "step": 2150
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2377530807131138e-05,
+      "loss": 1.2654,
+      "step": 2151
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2371080252865514e-05,
+      "loss": 1.212,
+      "step": 2152
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2364628653061257e-05,
+      "loss": 1.2384,
+      "step": 2153
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2358176010563223e-05,
+      "loss": 1.2909,
+      "step": 2154
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2351722328216735e-05,
+      "loss": 1.1288,
+      "step": 2155
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2345267608867576e-05,
+      "loss": 1.3779,
+      "step": 2156
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2338811855361978e-05,
+      "loss": 1.1561,
+      "step": 2157
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.233235507054664e-05,
+      "loss": 1.2337,
+      "step": 2158
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2325897257268709e-05,
+      "loss": 1.207,
+      "step": 2159
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2319438418375792e-05,
+      "loss": 1.3053,
+      "step": 2160
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2312978556715934e-05,
+      "loss": 1.2074,
+      "step": 2161
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2306517675137646e-05,
+      "loss": 1.1205,
+      "step": 2162
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2300055776489883e-05,
+      "loss": 1.0768,
+      "step": 2163
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2293592863622045e-05,
+      "loss": 1.28,
+      "step": 2164
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2287128939383992e-05,
+      "loss": 1.2283,
+      "step": 2165
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2280664006626012e-05,
+      "loss": 1.2548,
+      "step": 2166
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.227419806819885e-05,
+      "loss": 1.3491,
+      "step": 2167
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2267731126953692e-05,
+      "loss": 1.2299,
+      "step": 2168
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.226126318574216e-05,
+      "loss": 1.112,
+      "step": 2169
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.225479424741633e-05,
+      "loss": 1.3544,
+      "step": 2170
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.22483243148287e-05,
+      "loss": 1.2072,
+      "step": 2171
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2241853390832229e-05,
+      "loss": 1.266,
+      "step": 2172
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.223538147828029e-05,
+      "loss": 1.1407,
+      "step": 2173
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2228908580026701e-05,
+      "loss": 1.2531,
+      "step": 2174
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2222434698925728e-05,
+      "loss": 1.2394,
+      "step": 2175
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.221595983783205e-05,
+      "loss": 1.0896,
+      "step": 2176
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2209483999600782e-05,
+      "loss": 1.1895,
+      "step": 2177
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2203007187087485e-05,
+      "loss": 1.2664,
+      "step": 2178
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2196529403148132e-05,
+      "loss": 1.1402,
+      "step": 2179
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2190050650639131e-05,
+      "loss": 1.2148,
+      "step": 2180
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2183570932417324e-05,
+      "loss": 1.1809,
+      "step": 2181
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2177090251339964e-05,
+      "loss": 1.2303,
+      "step": 2182
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2170608610264743e-05,
+      "loss": 1.2093,
+      "step": 2183
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2164126012049766e-05,
+      "loss": 1.147,
+      "step": 2184
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2157642459553565e-05,
+      "loss": 1.1723,
+      "step": 2185
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2151157955635098e-05,
+      "loss": 1.3066,
+      "step": 2186
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2144672503153728e-05,
+      "loss": 1.3848,
+      "step": 2187
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2138186104969248e-05,
+      "loss": 1.2017,
+      "step": 2188
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2131698763941863e-05,
+      "loss": 1.2466,
+      "step": 2189
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2125210482932204e-05,
+      "loss": 1.1954,
+      "step": 2190
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2118721264801299e-05,
+      "loss": 1.1702,
+      "step": 2191
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.21122311124106e-05,
+      "loss": 1.3583,
+      "step": 2192
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2105740028621971e-05,
+      "loss": 1.264,
+      "step": 2193
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2099248016297682e-05,
+      "loss": 1.2011,
+      "step": 2194
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2092755078300423e-05,
+      "loss": 1.2766,
+      "step": 2195
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2086261217493276e-05,
+      "loss": 1.3269,
+      "step": 2196
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2079766436739742e-05,
+      "loss": 1.1157,
+      "step": 2197
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2073270738903726e-05,
+      "loss": 1.1149,
+      "step": 2198
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.206677412684953e-05,
+      "loss": 1.3246,
+      "step": 2199
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2060276603441872e-05,
+      "loss": 1.4043,
+      "step": 2200
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2053778171545859e-05,
+      "loss": 1.1794,
+      "step": 2201
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2047278834027006e-05,
+      "loss": 1.3312,
+      "step": 2202
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2040778593751227e-05,
+      "loss": 1.0945,
+      "step": 2203
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2034277453584828e-05,
+      "loss": 1.1828,
+      "step": 2204
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2027775416394523e-05,
+      "loss": 1.1495,
+      "step": 2205
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.202127248504741e-05,
+      "loss": 1.191,
+      "step": 2206
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2014768662410984e-05,
+      "loss": 1.1785,
+      "step": 2207
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2008263951353143e-05,
+      "loss": 1.1198,
+      "step": 2208
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2001758354742163e-05,
+      "loss": 1.2585,
+      "step": 2209
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1995251875446716e-05,
+      "loss": 1.1766,
+      "step": 2210
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.198874451633587e-05,
+      "loss": 1.2352,
+      "step": 2211
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1982236280279067e-05,
+      "loss": 1.2388,
+      "step": 2212
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.197572717014615e-05,
+      "loss": 1.2522,
+      "step": 2213
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1969217188807334e-05,
+      "loss": 1.2445,
+      "step": 2214
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.196270633913323e-05,
+      "loss": 1.1337,
+      "step": 2215
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1956194623994828e-05,
+      "loss": 1.1672,
+      "step": 2216
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.194968204626349e-05,
+      "loss": 1.1682,
+      "step": 2217
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1943168608810977e-05,
+      "loss": 1.2777,
+      "step": 2218
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1936654314509415e-05,
+      "loss": 1.2682,
+      "step": 2219
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.193013916623131e-05,
+      "loss": 1.1838,
+      "step": 2220
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1923623166849548e-05,
+      "loss": 1.1499,
+      "step": 2221
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1917106319237386e-05,
+      "loss": 1.2573,
+      "step": 2222
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.191058862626846e-05,
+      "loss": 1.1919,
+      "step": 2223
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1904070090816777e-05,
+      "loss": 1.2868,
+      "step": 2224
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1897550715756714e-05,
+      "loss": 1.1732,
+      "step": 2225
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1891030503963022e-05,
+      "loss": 1.3387,
+      "step": 2226
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.188450945831081e-05,
+      "loss": 1.2035,
+      "step": 2227
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1877987581675572e-05,
+      "loss": 1.1946,
+      "step": 2228
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1871464876933156e-05,
+      "loss": 1.2168,
+      "step": 2229
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1864941346959775e-05,
+      "loss": 1.3371,
+      "step": 2230
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1858416994632013e-05,
+      "loss": 1.1599,
+      "step": 2231
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.185189182282681e-05,
+      "loss": 1.2923,
+      "step": 2232
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1845365834421473e-05,
+      "loss": 1.2731,
+      "step": 2233
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1838839032293669e-05,
+      "loss": 1.0936,
+      "step": 2234
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1832311419321414e-05,
+      "loss": 1.2151,
+      "step": 2235
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1825782998383093e-05,
+      "loss": 1.2021,
+      "step": 2236
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1819253772357442e-05,
+      "loss": 1.1418,
+      "step": 2237
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1812723744123553e-05,
+      "loss": 1.2267,
+      "step": 2238
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1806192916560873e-05,
+      "loss": 1.1674,
+      "step": 2239
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1799661292549196e-05,
+      "loss": 1.2251,
+      "step": 2240
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1793128874968675e-05,
+      "loss": 1.1546,
+      "step": 2241
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1786595666699809e-05,
+      "loss": 1.2256,
+      "step": 2242
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1780061670623448e-05,
+      "loss": 1.2103,
+      "step": 2243
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1773526889620782e-05,
+      "loss": 1.23,
+      "step": 2244
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1766991326573356e-05,
+      "loss": 1.2375,
+      "step": 2245
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1760454984363059e-05,
+      "loss": 1.1686,
+      "step": 2246
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1753917865872111e-05,
+      "loss": 1.2204,
+      "step": 2247
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1747379973983096e-05,
+      "loss": 1.1298,
+      "step": 2248
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1740841311578918e-05,
+      "loss": 1.0505,
+      "step": 2249
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1734301881542836e-05,
+      "loss": 1.272,
+      "step": 2250
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1727761686758438e-05,
+      "loss": 1.2947,
+      "step": 2251
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1721220730109654e-05,
+      "loss": 1.2322,
+      "step": 2252
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1714679014480752e-05,
+      "loss": 1.1337,
+      "step": 2253
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1708136542756325e-05,
+      "loss": 1.2433,
+      "step": 2254
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1701593317821308e-05,
+      "loss": 1.0971,
+      "step": 2255
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1695049342560969e-05,
+      "loss": 1.1378,
+      "step": 2256
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1688504619860899e-05,
+      "loss": 1.1557,
+      "step": 2257
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1681959152607025e-05,
+      "loss": 1.3098,
+      "step": 2258
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1675412943685605e-05,
+      "loss": 1.2297,
+      "step": 2259
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.166886599598321e-05,
+      "loss": 1.2626,
+      "step": 2260
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1662318312386755e-05,
+      "loss": 1.2943,
+      "step": 2261
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1655769895783469e-05,
+      "loss": 1.3109,
+      "step": 2262
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1649220749060903e-05,
+      "loss": 1.1585,
+      "step": 2263
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1642670875106939e-05,
+      "loss": 1.2352,
+      "step": 2264
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1636120276809764e-05,
+      "loss": 1.2147,
+      "step": 2265
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1629568957057904e-05,
+      "loss": 1.2102,
+      "step": 2266
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.162301691874019e-05,
+      "loss": 1.2682,
+      "step": 2267
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1616464164745768e-05,
+      "loss": 1.2274,
+      "step": 2268
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1609910697964115e-05,
+      "loss": 1.2437,
+      "step": 2269
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1603356521285005e-05,
+      "loss": 1.229,
+      "step": 2270
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1596801637598532e-05,
+      "loss": 1.1402,
+      "step": 2271
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1590246049795102e-05,
+      "loss": 1.3038,
+      "step": 2272
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1583689760765436e-05,
+      "loss": 1.0937,
+      "step": 2273
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1577132773400552e-05,
+      "loss": 1.2083,
+      "step": 2274
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.157057509059179e-05,
+      "loss": 1.3304,
+      "step": 2275
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1564016715230789e-05,
+      "loss": 1.2919,
+      "step": 2276
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.155745765020949e-05,
+      "loss": 1.1469,
+      "step": 2277
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1550897898420147e-05,
+      "loss": 1.2177,
+      "step": 2278
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.154433746275531e-05,
+      "loss": 1.19,
+      "step": 2279
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1537776346107834e-05,
+      "loss": 1.2771,
+      "step": 2280
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1531214551370877e-05,
+      "loss": 1.2228,
+      "step": 2281
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1524652081437886e-05,
+      "loss": 1.2578,
+      "step": 2282
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1518088939202613e-05,
+      "loss": 1.1693,
+      "step": 2283
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.151152512755911e-05,
+      "loss": 1.2235,
+      "step": 2284
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1504960649401714e-05,
+      "loss": 1.253,
+      "step": 2285
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1498395507625066e-05,
+      "loss": 1.3779,
+      "step": 2286
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1491829705124093e-05,
+      "loss": 1.3469,
+      "step": 2287
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1485263244794016e-05,
+      "loss": 1.186,
+      "step": 2288
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1478696129530347e-05,
+      "loss": 1.1237,
+      "step": 2289
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1472128362228881e-05,
+      "loss": 1.288,
+      "step": 2290
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1465559945785712e-05,
+      "loss": 1.2214,
+      "step": 2291
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1458990883097205e-05,
+      "loss": 1.2791,
+      "step": 2292
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1452421177060023e-05,
+      "loss": 1.2401,
+      "step": 2293
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.144585083057111e-05,
+      "loss": 1.2221,
+      "step": 2294
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1439279846527682e-05,
+      "loss": 1.2635,
+      "step": 2295
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1432708227827253e-05,
+      "loss": 1.2602,
+      "step": 2296
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1426135977367606e-05,
+      "loss": 1.1589,
+      "step": 2297
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.14195630980468e-05,
+      "loss": 1.2373,
+      "step": 2298
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1412989592763181e-05,
+      "loss": 1.2847,
+      "step": 2299
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1406415464415364e-05,
+      "loss": 1.1671,
+      "step": 2300
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1399840715902243e-05,
+      "loss": 1.2465,
+      "step": 2301
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1393265350122982e-05,
+      "loss": 1.2811,
+      "step": 2302
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1386689369977016e-05,
+      "loss": 1.21,
+      "step": 2303
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1380112778364058e-05,
+      "loss": 1.2623,
+      "step": 2304
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1373535578184083e-05,
+      "loss": 1.2327,
+      "step": 2305
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1366957772337337e-05,
+      "loss": 1.247,
+      "step": 2306
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1360379363724338e-05,
+      "loss": 1.2205,
+      "step": 2307
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1353800355245857e-05,
+      "loss": 1.1606,
+      "step": 2308
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1347220749802945e-05,
+      "loss": 1.1243,
+      "step": 2309
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1340640550296906e-05,
+      "loss": 1.1129,
+      "step": 2310
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.133405975962931e-05,
+      "loss": 1.2248,
+      "step": 2311
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1327478380701988e-05,
+      "loss": 1.3316,
+      "step": 2312
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1320896416417026e-05,
+      "loss": 1.2513,
+      "step": 2313
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1314313869676769e-05,
+      "loss": 1.0329,
+      "step": 2314
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1307730743383827e-05,
+      "loss": 1.3499,
+      "step": 2315
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1301147040441056e-05,
+      "loss": 1.1978,
+      "step": 2316
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1294562763751573e-05,
+      "loss": 1.2836,
+      "step": 2317
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.128797791621874e-05,
+      "loss": 1.2086,
+      "step": 2318
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1281392500746178e-05,
+      "loss": 1.3395,
+      "step": 2319
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1274806520237755e-05,
+      "loss": 1.1592,
+      "step": 2320
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1268219977597594e-05,
+      "loss": 1.2109,
+      "step": 2321
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1261632875730052e-05,
+      "loss": 1.1889,
+      "step": 2322
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1255045217539748e-05,
+      "loss": 1.252,
+      "step": 2323
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.124845700593154e-05,
+      "loss": 1.1332,
+      "step": 2324
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1241868243810525e-05,
+      "loss": 1.3159,
+      "step": 2325
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1235278934082056e-05,
+      "loss": 1.2525,
+      "step": 2326
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1228689079651712e-05,
+      "loss": 1.2487,
+      "step": 2327
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.122209868342532e-05,
+      "loss": 1.2188,
+      "step": 2328
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1215507748308949e-05,
+      "loss": 1.1145,
+      "step": 2329
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1208916277208894e-05,
+      "loss": 1.3592,
+      "step": 2330
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1202324273031706e-05,
+      "loss": 1.1015,
+      "step": 2331
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.119573173868415e-05,
+      "loss": 1.1642,
+      "step": 2332
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1189138677073237e-05,
+      "loss": 1.0895,
+      "step": 2333
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.118254509110621e-05,
+      "loss": 1.1934,
+      "step": 2334
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1175950983690535e-05,
+      "loss": 1.2069,
+      "step": 2335
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1169356357733921e-05,
+      "loss": 1.1497,
+      "step": 2336
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1162761216144293e-05,
+      "loss": 1.3452,
+      "step": 2337
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1156165561829806e-05,
+      "loss": 1.2745,
+      "step": 2338
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1149569397698854e-05,
+      "loss": 1.2363,
+      "step": 2339
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1142972726660038e-05,
+      "loss": 1.1733,
+      "step": 2340
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.113637555162219e-05,
+      "loss": 1.1773,
+      "step": 2341
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1129777875494367e-05,
+      "loss": 1.1764,
+      "step": 2342
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.112317970118584e-05,
+      "loss": 1.3712,
+      "step": 2343
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1116581031606112e-05,
+      "loss": 1.2424,
+      "step": 2344
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1109981869664892e-05,
+      "loss": 1.1545,
+      "step": 2345
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1103382218272107e-05,
+      "loss": 1.1717,
+      "step": 2346
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1096782080337912e-05,
+      "loss": 1.1105,
+      "step": 2347
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1090181458772657e-05,
+      "loss": 1.0296,
+      "step": 2348
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1083580356486925e-05,
+      "loss": 1.2299,
+      "step": 2349
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1076978776391498e-05,
+      "loss": 1.256,
+      "step": 2350
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1070376721397374e-05,
+      "loss": 1.2183,
+      "step": 2351
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1063774194415762e-05,
+      "loss": 1.3073,
+      "step": 2352
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.105717119835807e-05,
+      "loss": 1.0768,
+      "step": 2353
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1050567736135921e-05,
+      "loss": 1.3293,
+      "step": 2354
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1043963810661144e-05,
+      "loss": 1.081,
+      "step": 2355
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.103735942484577e-05,
+      "loss": 1.1429,
+      "step": 2356
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1030754581602034e-05,
+      "loss": 1.3276,
+      "step": 2357
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.102414928384237e-05,
+      "loss": 1.0582,
+      "step": 2358
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1017543534479411e-05,
+      "loss": 1.3064,
+      "step": 2359
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1010937336425997e-05,
+      "loss": 1.1595,
+      "step": 2360
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1004330692595159e-05,
+      "loss": 1.2286,
+      "step": 2361
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0997723605900127e-05,
+      "loss": 1.1716,
+      "step": 2362
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0991116079254328e-05,
+      "loss": 1.1957,
+      "step": 2363
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0984508115571377e-05,
+      "loss": 1.3615,
+      "step": 2364
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.097789971776509e-05,
+      "loss": 1.3012,
+      "step": 2365
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0971290888749466e-05,
+      "loss": 1.4195,
+      "step": 2366
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0964681631438703e-05,
+      "loss": 1.3805,
+      "step": 2367
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0958071948747176e-05,
+      "loss": 1.2207,
+      "step": 2368
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0951461843589464e-05,
+      "loss": 1.1696,
+      "step": 2369
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0944851318880314e-05,
+      "loss": 1.1782,
+      "step": 2370
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0938240377534673e-05,
+      "loss": 1.259,
+      "step": 2371
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0931629022467664e-05,
+      "loss": 1.2053,
+      "step": 2372
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.092501725659459e-05,
+      "loss": 1.2595,
+      "step": 2373
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0918405082830948e-05,
+      "loss": 1.2139,
+      "step": 2374
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0911792504092398e-05,
+      "loss": 1.2715,
+      "step": 2375
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0905179523294791e-05,
+      "loss": 1.1463,
+      "step": 2376
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0898566143354153e-05,
+      "loss": 1.219,
+      "step": 2377
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0891952367186674e-05,
+      "loss": 1.2941,
+      "step": 2378
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0885338197708741e-05,
+      "loss": 1.1821,
+      "step": 2379
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0878723637836896e-05,
+      "loss": 1.2606,
+      "step": 2380
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.087210869048786e-05,
+      "loss": 1.074,
+      "step": 2381
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0865493358578525e-05,
+      "loss": 1.0998,
+      "step": 2382
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0858877645025948e-05,
+      "loss": 1.0849,
+      "step": 2383
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0852261552747365e-05,
+      "loss": 1.1006,
+      "step": 2384
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0845645084660169e-05,
+      "loss": 1.1622,
+      "step": 2385
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0839028243681913e-05,
+      "loss": 1.2156,
+      "step": 2386
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0832411032730339e-05,
+      "loss": 1.2429,
+      "step": 2387
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0825793454723325e-05,
+      "loss": 1.1937,
+      "step": 2388
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0819175512578925e-05,
+      "loss": 1.2012,
+      "step": 2389
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0812557209215355e-05,
+      "loss": 1.2788,
+      "step": 2390
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0805938547550976e-05,
+      "loss": 1.3965,
+      "step": 2391
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0799319530504328e-05,
+      "loss": 1.2977,
+      "step": 2392
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0792700160994091e-05,
+      "loss": 1.2032,
+      "step": 2393
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0786080441939107e-05,
+      "loss": 1.3107,
+      "step": 2394
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0779460376258372e-05,
+      "loss": 1.3469,
+      "step": 2395
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0772839966871034e-05,
+      "loss": 1.1976,
+      "step": 2396
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.076621921669639e-05,
+      "loss": 1.1754,
+      "step": 2397
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0759598128653892e-05,
+      "loss": 1.1914,
+      "step": 2398
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0752976705663141e-05,
+      "loss": 1.2557,
+      "step": 2399
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0746354950643883e-05,
+      "loss": 1.0626,
+      "step": 2400
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0739732866516007e-05,
+      "loss": 1.3699,
+      "step": 2401
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0733110456199553e-05,
+      "loss": 1.3192,
+      "step": 2402
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0726487722614704e-05,
+      "loss": 1.1705,
+      "step": 2403
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.071986466868179e-05,
+      "loss": 1.3712,
+      "step": 2404
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0713241297321266e-05,
+      "loss": 1.2193,
+      "step": 2405
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0706617611453745e-05,
+      "loss": 1.1625,
+      "step": 2406
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0699993613999965e-05,
+      "loss": 1.1698,
+      "step": 2407
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0693369307880817e-05,
+      "loss": 1.3098,
+      "step": 2408
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0686744696017314e-05,
+      "loss": 1.3151,
+      "step": 2409
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0680119781330609e-05,
+      "loss": 1.2391,
+      "step": 2410
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0673494566741987e-05,
+      "loss": 1.1573,
+      "step": 2411
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.066686905517287e-05,
+      "loss": 1.2953,
+      "step": 2412
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0660243249544804e-05,
+      "loss": 1.264,
+      "step": 2413
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0653617152779469e-05,
+      "loss": 1.2734,
+      "step": 2414
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0646990767798672e-05,
+      "loss": 1.1916,
+      "step": 2415
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0640364097524352e-05,
+      "loss": 1.2545,
+      "step": 2416
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0633737144878566e-05,
+      "loss": 1.2217,
+      "step": 2417
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0627109912783497e-05,
+      "loss": 1.2384,
+      "step": 2418
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0620482404161456e-05,
+      "loss": 1.217,
+      "step": 2419
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0613854621934876e-05,
+      "loss": 1.2713,
+      "step": 2420
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0607226569026297e-05,
+      "loss": 1.2404,
+      "step": 2421
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0600598248358403e-05,
+      "loss": 1.2511,
+      "step": 2422
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0593969662853972e-05,
+      "loss": 1.2636,
+      "step": 2423
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0587340815435913e-05,
+      "loss": 1.1398,
+      "step": 2424
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0580711709027247e-05,
+      "loss": 1.37,
+      "step": 2425
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0574082346551106e-05,
+      "loss": 1.2608,
+      "step": 2426
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0567452730930743e-05,
+      "loss": 1.2304,
+      "step": 2427
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0560822865089509e-05,
+      "loss": 1.0654,
+      "step": 2428
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0554192751950881e-05,
+      "loss": 1.3117,
+      "step": 2429
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0547562394438434e-05,
+      "loss": 1.196,
+      "step": 2430
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.054093179547585e-05,
+      "loss": 1.2926,
+      "step": 2431
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0534300957986933e-05,
+      "loss": 1.1823,
+      "step": 2432
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0527669884895573e-05,
+      "loss": 1.3023,
+      "step": 2433
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0521038579125773e-05,
+      "loss": 1.271,
+      "step": 2434
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.051440704360164e-05,
+      "loss": 1.2218,
+      "step": 2435
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0507775281247376e-05,
+      "loss": 1.1186,
+      "step": 2436
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.050114329498729e-05,
+      "loss": 1.213,
+      "step": 2437
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0494511087745787e-05,
+      "loss": 1.2725,
+      "step": 2438
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0487878662447361e-05,
+      "loss": 1.2641,
+      "step": 2439
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0481246022016622e-05,
+      "loss": 1.2659,
+      "step": 2440
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0474613169378256e-05,
+      "loss": 1.1011,
+      "step": 2441
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.046798010745705e-05,
+      "loss": 1.0395,
+      "step": 2442
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0461346839177886e-05,
+      "loss": 1.3205,
+      "step": 2443
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0454713367465729e-05,
+      "loss": 1.2418,
+      "step": 2444
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0448079695245643e-05,
+      "loss": 1.2225,
+      "step": 2445
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0441445825442773e-05,
+      "loss": 1.1503,
+      "step": 2446
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0434811760982356e-05,
+      "loss": 1.3793,
+      "step": 2447
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0428177504789713e-05,
+      "loss": 1.2553,
+      "step": 2448
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.042154305979025e-05,
+      "loss": 1.1433,
+      "step": 2449
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0414908428909452e-05,
+      "loss": 1.3304,
+      "step": 2450
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0408273615072893e-05,
+      "loss": 1.211,
+      "step": 2451
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0401638621206224e-05,
+      "loss": 1.1625,
+      "step": 2452
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0395003450235177e-05,
+      "loss": 1.1934,
+      "step": 2453
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0388368105085558e-05,
+      "loss": 1.2857,
+      "step": 2454
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0381732588683253e-05,
+      "loss": 1.1756,
+      "step": 2455
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0375096903954224e-05,
+      "loss": 1.2393,
+      "step": 2456
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.036846105382451e-05,
+      "loss": 1.2036,
+      "step": 2457
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0361825041220213e-05,
+      "loss": 1.1805,
+      "step": 2458
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0355188869067516e-05,
+      "loss": 1.176,
+      "step": 2459
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.034855254029267e-05,
+      "loss": 1.3729,
+      "step": 2460
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.034191605782199e-05,
+      "loss": 1.2542,
+      "step": 2461
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0335279424581872e-05,
+      "loss": 1.3823,
+      "step": 2462
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0328642643498763e-05,
+      "loss": 1.2613,
+      "step": 2463
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0322005717499181e-05,
+      "loss": 1.1591,
+      "step": 2464
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0315368649509716e-05,
+      "loss": 1.2767,
+      "step": 2465
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0308731442457004e-05,
+      "loss": 1.2972,
+      "step": 2466
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0302094099267758e-05,
+      "loss": 1.2043,
+      "step": 2467
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0295456622868745e-05,
+      "loss": 1.1309,
+      "step": 2468
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0288819016186782e-05,
+      "loss": 1.2666,
+      "step": 2469
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0282181282148764e-05,
+      "loss": 1.1288,
+      "step": 2470
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0275543423681622e-05,
+      "loss": 1.2417,
+      "step": 2471
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0268905443712351e-05,
+      "loss": 1.1285,
+      "step": 2472
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0262267345168003e-05,
+      "loss": 1.2674,
+      "step": 2473
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0255629130975671e-05,
+      "loss": 1.108,
+      "step": 2474
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.024899080406251e-05,
+      "loss": 1.2013,
+      "step": 2475
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0242352367355722e-05,
+      "loss": 1.3231,
+      "step": 2476
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0235713823782552e-05,
+      "loss": 1.2953,
+      "step": 2477
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0229075176270297e-05,
+      "loss": 1.2002,
+      "step": 2478
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0222436427746301e-05,
+      "loss": 1.1786,
+      "step": 2479
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0215797581137948e-05,
+      "loss": 1.2723,
+      "step": 2480
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.020915863937267e-05,
+      "loss": 1.3138,
+      "step": 2481
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0202519605377933e-05,
+      "loss": 1.1899,
+      "step": 2482
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.019588048208126e-05,
+      "loss": 1.1679,
+      "step": 2483
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0189241272410191e-05,
+      "loss": 1.1077,
+      "step": 2484
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0182601979292323e-05,
+      "loss": 1.0958,
+      "step": 2485
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0175962605655278e-05,
+      "loss": 1.2632,
+      "step": 2486
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0169323154426726e-05,
+      "loss": 1.2769,
+      "step": 2487
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0162683628534353e-05,
+      "loss": 1.3201,
+      "step": 2488
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0156044030905894e-05,
+      "loss": 1.1845,
+      "step": 2489
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0149404364469108e-05,
+      "loss": 1.1835,
+      "step": 2490
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0142764632151792e-05,
+      "loss": 1.242,
+      "step": 2491
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0136124836881755e-05,
+      "loss": 1.326,
+      "step": 2492
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0129484981586853e-05,
+      "loss": 1.278,
+      "step": 2493
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0122845069194957e-05,
+      "loss": 1.2498,
+      "step": 2494
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0116205102633974e-05,
+      "loss": 1.1429,
+      "step": 2495
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0109565084831817e-05,
+      "loss": 1.1721,
+      "step": 2496
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0102925018716436e-05,
+      "loss": 1.2234,
+      "step": 2497
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.00962849072158e-05,
+      "loss": 1.2518,
+      "step": 2498
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0089644753257897e-05,
+      "loss": 1.2793,
+      "step": 2499
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.008300455977073e-05,
+      "loss": 1.179,
+      "step": 2500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0076364329682327e-05,
+      "loss": 1.2614,
+      "step": 2501
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.006972406592072e-05,
+      "loss": 1.234,
+      "step": 2502
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0063083771413975e-05,
+      "loss": 1.2507,
+      "step": 2503
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0056443449090147e-05,
+      "loss": 1.1314,
+      "step": 2504
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.004980310187733e-05,
+      "loss": 1.1602,
+      "step": 2505
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0043162732703601e-05,
+      "loss": 1.1613,
+      "step": 2506
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0036522344497073e-05,
+      "loss": 1.2129,
+      "step": 2507
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0029881940185851e-05,
+      "loss": 1.2206,
+      "step": 2508
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0023241522698048e-05,
+      "loss": 1.2416,
+      "step": 2509
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0016601094961793e-05,
+      "loss": 1.2796,
+      "step": 2510
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0009960659905212e-05,
+      "loss": 1.2782,
+      "step": 2511
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0003320220456424e-05,
+      "loss": 1.2047,
+      "step": 2512
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.996679779543577e-06,
+      "loss": 1.165,
+      "step": 2513
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.990039340094793e-06,
+      "loss": 1.1324,
+      "step": 2514
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.983398905038212e-06,
+      "loss": 1.3029,
+      "step": 2515
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.976758477301952e-06,
+      "loss": 1.2265,
+      "step": 2516
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.97011805981415e-06,
+      "loss": 1.3046,
+      "step": 2517
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.96347765550293e-06,
+      "loss": 1.2222,
+      "step": 2518
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.9568372672964e-06,
+      "loss": 1.1098,
+      "step": 2519
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.950196898122677e-06,
+      "loss": 1.1683,
+      "step": 2520
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.943556550909853e-06,
+      "loss": 1.1924,
+      "step": 2521
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.936916228586027e-06,
+      "loss": 1.177,
+      "step": 2522
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.930275934079281e-06,
+      "loss": 1.4776,
+      "step": 2523
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.923635670317676e-06,
+      "loss": 1.1513,
+      "step": 2524
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.916995440229274e-06,
+      "loss": 1.2512,
+      "step": 2525
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.910355246742105e-06,
+      "loss": 1.2445,
+      "step": 2526
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.903715092784202e-06,
+      "loss": 1.2028,
+      "step": 2527
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.897074981283566e-06,
+      "loss": 1.2481,
+      "step": 2528
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.890434915168187e-06,
+      "loss": 1.1212,
+      "step": 2529
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.88379489736603e-06,
+      "loss": 1.2251,
+      "step": 2530
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.877154930805045e-06,
+      "loss": 1.2642,
+      "step": 2531
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.870515018413147e-06,
+      "loss": 1.3227,
+      "step": 2532
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.863875163118248e-06,
+      "loss": 1.2687,
+      "step": 2533
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.857235367848213e-06,
+      "loss": 1.3137,
+      "step": 2534
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.850595635530894e-06,
+      "loss": 1.262,
+      "step": 2535
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.843955969094111e-06,
+      "loss": 1.2259,
+      "step": 2536
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.837316371465647e-06,
+      "loss": 1.1867,
+      "step": 2537
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.830676845573277e-06,
+      "loss": 1.2331,
+      "step": 2538
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.824037394344724e-06,
+      "loss": 1.2654,
+      "step": 2539
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.81739802070768e-06,
+      "loss": 1.1693,
+      "step": 2540
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.810758727589814e-06,
+      "loss": 1.2297,
+      "step": 2541
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.804119517918741e-06,
+      "loss": 1.1035,
+      "step": 2542
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.797480394622067e-06,
+      "loss": 1.1355,
+      "step": 2543
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.790841360627335e-06,
+      "loss": 1.3353,
+      "step": 2544
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.784202418862055e-06,
+      "loss": 1.2585,
+      "step": 2545
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.777563572253704e-06,
+      "loss": 1.3114,
+      "step": 2546
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.770924823729708e-06,
+      "loss": 1.2018,
+      "step": 2547
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.76428617621745e-06,
+      "loss": 1.2322,
+      "step": 2548
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.757647632644282e-06,
+      "loss": 1.2461,
+      "step": 2549
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.751009195937493e-06,
+      "loss": 1.2064,
+      "step": 2550
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.744370869024334e-06,
+      "loss": 1.2067,
+      "step": 2551
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.737732654832002e-06,
+      "loss": 1.3351,
+      "step": 2552
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.731094556287649e-06,
+      "loss": 1.1365,
+      "step": 2553
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.724456576318383e-06,
+      "loss": 1.323,
+      "step": 2554
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.717818717851239e-06,
+      "loss": 1.3235,
+      "step": 2555
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.711180983813222e-06,
+      "loss": 1.3857,
+      "step": 2556
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.70454337713126e-06,
+      "loss": 1.1636,
+      "step": 2557
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.697905900732244e-06,
+      "loss": 1.2447,
+      "step": 2558
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.691268557542997e-06,
+      "loss": 1.2341,
+      "step": 2559
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.684631350490287e-06,
+      "loss": 1.2639,
+      "step": 2560
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.67799428250082e-06,
+      "loss": 1.2663,
+      "step": 2561
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.67135735650124e-06,
+      "loss": 1.2039,
+      "step": 2562
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.664720575418132e-06,
+      "loss": 1.1046,
+      "step": 2563
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.658083942178009e-06,
+      "loss": 1.2651,
+      "step": 2564
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.651447459707333e-06,
+      "loss": 1.3284,
+      "step": 2565
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.644811130932487e-06,
+      "loss": 1.245,
+      "step": 2566
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.63817495877979e-06,
+      "loss": 1.2037,
+      "step": 2567
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.631538946175496e-06,
+      "loss": 1.3877,
+      "step": 2568
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.624903096045777e-06,
+      "loss": 1.2527,
+      "step": 2569
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.618267411316748e-06,
+      "loss": 1.2192,
+      "step": 2570
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.611631894914445e-06,
+      "loss": 1.2999,
+      "step": 2571
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.604996549764826e-06,
+      "loss": 1.107,
+      "step": 2572
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.598361378793779e-06,
+      "loss": 1.3258,
+      "step": 2573
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.59172638492711e-06,
+      "loss": 1.3608,
+      "step": 2574
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.58509157109055e-06,
+      "loss": 1.2032,
+      "step": 2575
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.578456940209754e-06,
+      "loss": 1.3372,
+      "step": 2576
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.571822495210288e-06,
+      "loss": 1.3823,
+      "step": 2577
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.565188239017646e-06,
+      "loss": 1.2509,
+      "step": 2578
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.55855417455723e-06,
+      "loss": 1.1485,
+      "step": 2579
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.551920304754359e-06,
+      "loss": 1.3012,
+      "step": 2580
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.545286632534272e-06,
+      "loss": 1.2836,
+      "step": 2581
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.538653160822118e-06,
+      "loss": 1.1291,
+      "step": 2582
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.532019892542953e-06,
+      "loss": 1.1914,
+      "step": 2583
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.525386830621749e-06,
+      "loss": 1.2905,
+      "step": 2584
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.518753977983378e-06,
+      "loss": 1.3361,
+      "step": 2585
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.51212133755264e-06,
+      "loss": 1.2822,
+      "step": 2586
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.505488912254218e-06,
+      "loss": 1.1391,
+      "step": 2587
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.498856705012713e-06,
+      "loss": 1.1669,
+      "step": 2588
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.492224718752629e-06,
+      "loss": 1.1932,
+      "step": 2589
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.485592956398364e-06,
+      "loss": 1.2337,
+      "step": 2590
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.478961420874227e-06,
+      "loss": 1.2707,
+      "step": 2591
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.472330115104428e-06,
+      "loss": 1.26,
+      "step": 2592
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.465699042013068e-06,
+      "loss": 1.3814,
+      "step": 2593
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.459068204524152e-06,
+      "loss": 1.172,
+      "step": 2594
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.452437605561571e-06,
+      "loss": 1.2659,
+      "step": 2595
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.44580724804912e-06,
+      "loss": 1.2645,
+      "step": 2596
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.439177134910493e-06,
+      "loss": 1.3958,
+      "step": 2597
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.43254726906926e-06,
+      "loss": 1.1265,
+      "step": 2598
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.425917653448897e-06,
+      "loss": 1.3846,
+      "step": 2599
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.419288290972757e-06,
+      "loss": 1.2495,
+      "step": 2600
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.412659184564088e-06,
+      "loss": 1.3983,
+      "step": 2601
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.40603033714603e-06,
+      "loss": 1.2622,
+      "step": 2602
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.3994017516416e-06,
+      "loss": 1.2247,
+      "step": 2603
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.392773430973705e-06,
+      "loss": 1.1784,
+      "step": 2604
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.38614537806513e-06,
+      "loss": 1.3722,
+      "step": 2605
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.37951759583855e-06,
+      "loss": 1.3337,
+      "step": 2606
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.372890087216505e-06,
+      "loss": 1.2094,
+      "step": 2607
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.366262855121436e-06,
+      "loss": 1.2258,
+      "step": 2608
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.359635902475651e-06,
+      "loss": 1.2965,
+      "step": 2609
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.35300923220133e-06,
+      "loss": 1.2691,
+      "step": 2610
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.346382847220536e-06,
+      "loss": 1.0739,
+      "step": 2611
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.339756750455199e-06,
+      "loss": 1.0944,
+      "step": 2612
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.333130944827132e-06,
+      "loss": 1.2012,
+      "step": 2613
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.326505433258015e-06,
+      "loss": 1.1724,
+      "step": 2614
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.319880218669394e-06,
+      "loss": 1.1716,
+      "step": 2615
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.31325530398269e-06,
+      "loss": 1.2246,
+      "step": 2616
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.306630692119183e-06,
+      "loss": 1.2242,
+      "step": 2617
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.300006386000033e-06,
+      "loss": 1.2317,
+      "step": 2618
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.29338238854626e-06,
+      "loss": 1.2551,
+      "step": 2619
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.286758702678737e-06,
+      "loss": 1.1473,
+      "step": 2620
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.280135331318217e-06,
+      "loss": 1.2079,
+      "step": 2621
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.273512277385297e-06,
+      "loss": 1.2501,
+      "step": 2622
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.266889543800447e-06,
+      "loss": 1.1584,
+      "step": 2623
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.260267133483997e-06,
+      "loss": 1.2199,
+      "step": 2624
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.25364504935612e-06,
+      "loss": 1.1512,
+      "step": 2625
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.247023294336862e-06,
+      "loss": 1.1674,
+      "step": 2626
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.24040187134611e-06,
+      "loss": 1.1834,
+      "step": 2627
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.233780783303611e-06,
+      "loss": 1.1504,
+      "step": 2628
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.22716003312897e-06,
+      "loss": 1.1367,
+      "step": 2629
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.22053962374163e-06,
+      "loss": 1.3201,
+      "step": 2630
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.213919558060897e-06,
+      "loss": 1.1823,
+      "step": 2631
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.207299839005912e-06,
+      "loss": 1.3129,
+      "step": 2632
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.200680469495672e-06,
+      "loss": 1.1964,
+      "step": 2633
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.194061452449024e-06,
+      "loss": 1.2834,
+      "step": 2634
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.187442790784648e-06,
+      "loss": 1.3057,
+      "step": 2635
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.180824487421077e-06,
+      "loss": 1.3055,
+      "step": 2636
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.174206545276678e-06,
+      "loss": 1.2465,
+      "step": 2637
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.167588967269666e-06,
+      "loss": 1.2589,
+      "step": 2638
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.160971756318087e-06,
+      "loss": 1.2508,
+      "step": 2639
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.154354915339836e-06,
+      "loss": 1.4471,
+      "step": 2640
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.147738447252639e-06,
+      "loss": 1.2035,
+      "step": 2641
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.141122354974056e-06,
+      "loss": 1.3267,
+      "step": 2642
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.13450664142148e-06,
+      "loss": 1.0884,
+      "step": 2643
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.12789130951214e-06,
+      "loss": 1.3049,
+      "step": 2644
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.121276362163107e-06,
+      "loss": 1.3042,
+      "step": 2645
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.114661802291262e-06,
+      "loss": 1.2256,
+      "step": 2646
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.108047632813328e-06,
+      "loss": 1.1571,
+      "step": 2647
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.101433856645854e-06,
+      "loss": 1.2211,
+      "step": 2648
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.094820476705209e-06,
+      "loss": 1.3208,
+      "step": 2649
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.088207495907604e-06,
+      "loss": 1.213,
+      "step": 2650
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.081594917169056e-06,
+      "loss": 1.222,
+      "step": 2651
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.074982743405413e-06,
+      "loss": 1.2518,
+      "step": 2652
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.068370977532341e-06,
+      "loss": 1.0378,
+      "step": 2653
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.061759622465332e-06,
+      "loss": 1.2409,
+      "step": 2654
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.055148681119688e-06,
+      "loss": 1.1401,
+      "step": 2655
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.04853815641054e-06,
+      "loss": 1.2274,
+      "step": 2656
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.041928051252826e-06,
+      "loss": 1.0904,
+      "step": 2657
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.035318368561302e-06,
+      "loss": 1.1115,
+      "step": 2658
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.028709111250537e-06,
+      "loss": 1.0813,
+      "step": 2659
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.022100282234913e-06,
+      "loss": 1.1625,
+      "step": 2660
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.015491884428623e-06,
+      "loss": 1.3545,
+      "step": 2661
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.008883920745674e-06,
+      "loss": 1.1536,
+      "step": 2662
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.002276394099874e-06,
+      "loss": 1.3022,
+      "step": 2663
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.995669307404844e-06,
+      "loss": 1.2856,
+      "step": 2664
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.989062663574006e-06,
+      "loss": 1.218,
+      "step": 2665
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.98245646552059e-06,
+      "loss": 1.3126,
+      "step": 2666
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.975850716157634e-06,
+      "loss": 1.2445,
+      "step": 2667
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.969245418397969e-06,
+      "loss": 1.0743,
+      "step": 2668
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.962640575154232e-06,
+      "loss": 1.1898,
+      "step": 2669
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.956036189338857e-06,
+      "loss": 1.2119,
+      "step": 2670
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.94943226386408e-06,
+      "loss": 1.2651,
+      "step": 2671
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.942828801641934e-06,
+      "loss": 1.1159,
+      "step": 2672
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.936225805584241e-06,
+      "loss": 1.4155,
+      "step": 2673
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.929623278602627e-06,
+      "loss": 1.1815,
+      "step": 2674
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.923021223608504e-06,
+      "loss": 1.1751,
+      "step": 2675
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.916419643513075e-06,
+      "loss": 1.2267,
+      "step": 2676
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.909818541227345e-06,
+      "loss": 1.1997,
+      "step": 2677
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.90321791966209e-06,
+      "loss": 1.1493,
+      "step": 2678
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.896617781727895e-06,
+      "loss": 1.0893,
+      "step": 2679
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.890018130335112e-06,
+      "loss": 1.2399,
+      "step": 2680
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.883418968393893e-06,
+      "loss": 1.1968,
+      "step": 2681
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.87682029881416e-06,
+      "loss": 1.2866,
+      "step": 2682
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.870222124505636e-06,
+      "loss": 1.2551,
+      "step": 2683
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.863624448377814e-06,
+      "loss": 1.0656,
+      "step": 2684
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.857027273339967e-06,
+      "loss": 1.2253,
+      "step": 2685
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.85043060230115e-06,
+      "loss": 1.1896,
+      "step": 2686
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.843834438170194e-06,
+      "loss": 1.3757,
+      "step": 2687
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.837238783855709e-06,
+      "loss": 1.2354,
+      "step": 2688
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.830643642266082e-06,
+      "loss": 1.177,
+      "step": 2689
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.824049016309467e-06,
+      "loss": 1.1513,
+      "step": 2690
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.817454908893794e-06,
+      "loss": 1.2406,
+      "step": 2691
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.810861322926763e-06,
+      "loss": 1.1685,
+      "step": 2692
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.80426826131585e-06,
+      "loss": 1.3075,
+      "step": 2693
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.797675726968297e-06,
+      "loss": 1.0326,
+      "step": 2694
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.791083722791108e-06,
+      "loss": 1.276,
+      "step": 2695
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.784492251691056e-06,
+      "loss": 1.1992,
+      "step": 2696
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.777901316574687e-06,
+      "loss": 1.1446,
+      "step": 2697
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.771310920348291e-06,
+      "loss": 1.1161,
+      "step": 2698
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.764721065917947e-06,
+      "loss": 1.1016,
+      "step": 2699
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.758131756189477e-06,
+      "loss": 1.3023,
+      "step": 2700
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.751542994068464e-06,
+      "loss": 1.0426,
+      "step": 2701
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.744954782460255e-06,
+      "loss": 1.2293,
+      "step": 2702
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.73836712426995e-06,
+      "loss": 1.2184,
+      "step": 2703
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.731780022402409e-06,
+      "loss": 1.2475,
+      "step": 2704
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.725193479762248e-06,
+      "loss": 1.1933,
+      "step": 2705
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.718607499253825e-06,
+      "loss": 1.2552,
+      "step": 2706
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.712022083781265e-06,
+      "loss": 1.2375,
+      "step": 2707
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.70543723624843e-06,
+      "loss": 1.1036,
+      "step": 2708
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.698852959558944e-06,
+      "loss": 1.3035,
+      "step": 2709
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.692269256616175e-06,
+      "loss": 1.091,
+      "step": 2710
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.685686130323233e-06,
+      "loss": 1.207,
+      "step": 2711
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.67910358358298e-06,
+      "loss": 1.2527,
+      "step": 2712
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.672521619298017e-06,
+      "loss": 1.2063,
+      "step": 2713
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.66594024037069e-06,
+      "loss": 1.2751,
+      "step": 2714
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.659359449703095e-06,
+      "loss": 1.2442,
+      "step": 2715
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.652779250197057e-06,
+      "loss": 1.1755,
+      "step": 2716
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.646199644754147e-06,
+      "loss": 1.2771,
+      "step": 2717
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.639620636275667e-06,
+      "loss": 1.2254,
+      "step": 2718
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.633042227662663e-06,
+      "loss": 1.2221,
+      "step": 2719
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.626464421815919e-06,
+      "loss": 1.4972,
+      "step": 2720
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.619887221635943e-06,
+      "loss": 1.191,
+      "step": 2721
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.613310630022986e-06,
+      "loss": 1.1439,
+      "step": 2722
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.606734649877023e-06,
+      "loss": 1.219,
+      "step": 2723
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.600159284097756e-06,
+      "loss": 1.1875,
+      "step": 2724
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.593584535584636e-06,
+      "loss": 1.3079,
+      "step": 2725
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.58701040723682e-06,
+      "loss": 1.2893,
+      "step": 2726
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.580436901953203e-06,
+      "loss": 0.9705,
+      "step": 2727
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.5738640226324e-06,
+      "loss": 1.1792,
+      "step": 2728
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.567291772172752e-06,
+      "loss": 1.2923,
+      "step": 2729
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.56072015347232e-06,
+      "loss": 1.4513,
+      "step": 2730
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.554149169428894e-06,
+      "loss": 1.1875,
+      "step": 2731
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.547578822939978e-06,
+      "loss": 1.3248,
+      "step": 2732
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.541009116902798e-06,
+      "loss": 1.2291,
+      "step": 2733
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.534440054214295e-06,
+      "loss": 1.167,
+      "step": 2734
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.52787163777112e-06,
+      "loss": 1.1983,
+      "step": 2735
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.521303870469655e-06,
+      "loss": 1.1493,
+      "step": 2736
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.514736755205987e-06,
+      "loss": 1.2826,
+      "step": 2737
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.508170294875908e-06,
+      "loss": 1.2391,
+      "step": 2738
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.501604492374939e-06,
+      "loss": 1.1215,
+      "step": 2739
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.495039350598288e-06,
+      "loss": 1.1633,
+      "step": 2740
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.488474872440892e-06,
+      "loss": 1.1886,
+      "step": 2741
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.48191106079739e-06,
+      "loss": 1.2522,
+      "step": 2742
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.475347918562119e-06,
+      "loss": 1.1864,
+      "step": 2743
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.468785448629128e-06,
+      "loss": 1.1477,
+      "step": 2744
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.462223653892169e-06,
+      "loss": 1.2879,
+      "step": 2745
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.45566253724469e-06,
+      "loss": 1.1998,
+      "step": 2746
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.449102101579856e-06,
+      "loss": 1.2712,
+      "step": 2747
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.442542349790515e-06,
+      "loss": 1.1476,
+      "step": 2748
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.435983284769216e-06,
+      "loss": 1.1781,
+      "step": 2749
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.429424909408215e-06,
+      "loss": 1.1562,
+      "step": 2750
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.42286722659945e-06,
+      "loss": 1.1828,
+      "step": 2751
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.416310239234566e-06,
+      "loss": 1.2948,
+      "step": 2752
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.409753950204901e-06,
+      "loss": 1.2565,
+      "step": 2753
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.403198362401472e-06,
+      "loss": 1.211,
+      "step": 2754
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.396643478715001e-06,
+      "loss": 1.296,
+      "step": 2755
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.390089302035889e-06,
+      "loss": 1.2305,
+      "step": 2756
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.38353583525423e-06,
+      "loss": 1.3159,
+      "step": 2757
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.376983081259814e-06,
+      "loss": 1.1756,
+      "step": 2758
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.370431042942099e-06,
+      "loss": 1.2035,
+      "step": 2759
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.363879723190241e-06,
+      "loss": 1.2805,
+      "step": 2760
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.357329124893068e-06,
+      "loss": 1.3527,
+      "step": 2761
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.350779250939098e-06,
+      "loss": 1.1448,
+      "step": 2762
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.344230104216536e-06,
+      "loss": 1.2688,
+      "step": 2763
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.337681687613247e-06,
+      "loss": 1.2074,
+      "step": 2764
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.331134004016795e-06,
+      "loss": 1.126,
+      "step": 2765
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.324587056314402e-06,
+      "loss": 1.2987,
+      "step": 2766
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.318040847392975e-06,
+      "loss": 1.2813,
+      "step": 2767
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.311495380139105e-06,
+      "loss": 1.2542,
+      "step": 2768
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.304950657439034e-06,
+      "loss": 1.2167,
+      "step": 2769
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.298406682178694e-06,
+      "loss": 1.2654,
+      "step": 2770
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.291863457243679e-06,
+      "loss": 1.2854,
+      "step": 2771
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.285320985519254e-06,
+      "loss": 1.0825,
+      "step": 2772
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.278779269890347e-06,
+      "loss": 1.2349,
+      "step": 2773
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.272238313241563e-06,
+      "loss": 1.2253,
+      "step": 2774
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.265698118457167e-06,
+      "loss": 1.2107,
+      "step": 2775
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.259158688421085e-06,
+      "loss": 1.2927,
+      "step": 2776
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.25262002601691e-06,
+      "loss": 1.3621,
+      "step": 2777
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.24608213412789e-06,
+      "loss": 1.1464,
+      "step": 2778
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.239545015636944e-06,
+      "loss": 1.2748,
+      "step": 2779
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.233008673426647e-06,
+      "loss": 1.1789,
+      "step": 2780
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.226473110379221e-06,
+      "loss": 1.2972,
+      "step": 2781
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.219938329376557e-06,
+      "loss": 1.2683,
+      "step": 2782
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.213404333300191e-06,
+      "loss": 1.3675,
+      "step": 2783
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.206871125031324e-06,
+      "loss": 1.355,
+      "step": 2784
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.200338707450807e-06,
+      "loss": 1.1427,
+      "step": 2785
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.19380708343913e-06,
+      "loss": 1.3445,
+      "step": 2786
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.18727625587645e-06,
+      "loss": 1.1357,
+      "step": 2787
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.180746227642561e-06,
+      "loss": 1.2913,
+      "step": 2788
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.174217001616908e-06,
+      "loss": 1.1768,
+      "step": 2789
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.167688580678587e-06,
+      "loss": 1.103,
+      "step": 2790
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.161160967706333e-06,
+      "loss": 1.2221,
+      "step": 2791
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.154634165578529e-06,
+      "loss": 1.2335,
+      "step": 2792
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.148108177173191e-06,
+      "loss": 1.191,
+      "step": 2793
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.141583005367989e-06,
+      "loss": 1.1511,
+      "step": 2794
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.135058653040227e-06,
+      "loss": 1.2638,
+      "step": 2795
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.128535123066846e-06,
+      "loss": 1.1555,
+      "step": 2796
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.12201241832443e-06,
+      "loss": 1.3368,
+      "step": 2797
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.115490541689193e-06,
+      "loss": 1.2102,
+      "step": 2798
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.10896949603698e-06,
+      "loss": 1.2,
+      "step": 2799
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.102449284243288e-06,
+      "loss": 1.1622,
+      "step": 2800
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.095929909183225e-06,
+      "loss": 1.3032,
+      "step": 2801
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.089411373731542e-06,
+      "loss": 1.2772,
+      "step": 2802
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.082893680762619e-06,
+      "loss": 1.2479,
+      "step": 2803
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.076376833150457e-06,
+      "loss": 1.281,
+      "step": 2804
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.069860833768692e-06,
+      "loss": 1.1656,
+      "step": 2805
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.063345685490589e-06,
+      "loss": 1.166,
+      "step": 2806
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.056831391189024e-06,
+      "loss": 1.1941,
+      "step": 2807
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.050317953736511e-06,
+      "loss": 1.3127,
+      "step": 2808
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.043805376005177e-06,
+      "loss": 1.2381,
+      "step": 2809
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.03729366086677e-06,
+      "loss": 1.2754,
+      "step": 2810
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.030782811192668e-06,
+      "loss": 1.2491,
+      "step": 2811
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.024272829853852e-06,
+      "loss": 1.2586,
+      "step": 2812
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.017763719720938e-06,
+      "loss": 1.1979,
+      "step": 2813
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.011255483664135e-06,
+      "loss": 1.1241,
+      "step": 2814
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.004748124553284e-06,
+      "loss": 1.3839,
+      "step": 2815
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.99824164525784e-06,
+      "loss": 1.2906,
+      "step": 2816
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.991736048646859e-06,
+      "loss": 1.1297,
+      "step": 2817
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.985231337589019e-06,
+      "loss": 1.1454,
+      "step": 2818
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.978727514952595e-06,
+      "loss": 1.3406,
+      "step": 2819
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.972224583605484e-06,
+      "loss": 1.0924,
+      "step": 2820
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.965722546415173e-06,
+      "loss": 1.1714,
+      "step": 2821
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.959221406248776e-06,
+      "loss": 1.2317,
+      "step": 2822
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.952721165972996e-06,
+      "loss": 1.2715,
+      "step": 2823
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.946221828454143e-06,
+      "loss": 1.2298,
+      "step": 2824
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.939723396558133e-06,
+      "loss": 1.1361,
+      "step": 2825
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.93322587315047e-06,
+      "loss": 1.3244,
+      "step": 2826
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.926729261096277e-06,
+      "loss": 1.1272,
+      "step": 2827
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.92023356326026e-06,
+      "loss": 1.3007,
+      "step": 2828
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.913738782506727e-06,
+      "loss": 1.2885,
+      "step": 2829
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.907244921699582e-06,
+      "loss": 1.1812,
+      "step": 2830
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.900751983702318e-06,
+      "loss": 1.1976,
+      "step": 2831
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.89425997137803e-06,
+      "loss": 1.1273,
+      "step": 2832
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.887768887589403e-06,
+      "loss": 1.1267,
+      "step": 2833
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.881278735198704e-06,
+      "loss": 1.3099,
+      "step": 2834
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.874789517067801e-06,
+      "loss": 1.2669,
+      "step": 2835
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.868301236058138e-06,
+      "loss": 1.2302,
+      "step": 2836
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.861813895030754e-06,
+      "loss": 1.2481,
+      "step": 2837
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.855327496846277e-06,
+      "loss": 1.3035,
+      "step": 2838
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.848842044364906e-06,
+      "loss": 1.3086,
+      "step": 2839
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.842357540446437e-06,
+      "loss": 1.173,
+      "step": 2840
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.835873987950238e-06,
+      "loss": 1.2108,
+      "step": 2841
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.829391389735258e-06,
+      "loss": 1.223,
+      "step": 2842
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.822909748660039e-06,
+      "loss": 1.232,
+      "step": 2843
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.81642906758268e-06,
+      "loss": 1.2034,
+      "step": 2844
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.809949349360872e-06,
+      "loss": 1.1002,
+      "step": 2845
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.803470596851872e-06,
+      "loss": 1.2189,
+      "step": 2846
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.796992812912517e-06,
+      "loss": 1.1398,
+      "step": 2847
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.790516000399218e-06,
+      "loss": 1.2596,
+      "step": 2848
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.784040162167954e-06,
+      "loss": 1.181,
+      "step": 2849
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.777565301074274e-06,
+      "loss": 1.1228,
+      "step": 2850
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.7710914199733e-06,
+      "loss": 1.2665,
+      "step": 2851
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.764618521719716e-06,
+      "loss": 1.2316,
+      "step": 2852
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.758146609167773e-06,
+      "loss": 1.2488,
+      "step": 2853
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.751675685171301e-06,
+      "loss": 1.0805,
+      "step": 2854
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.745205752583673e-06,
+      "loss": 1.2245,
+      "step": 2855
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.738736814257844e-06,
+      "loss": 1.2706,
+      "step": 2856
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.732268873046313e-06,
+      "loss": 1.2375,
+      "step": 2857
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.72580193180115e-06,
+      "loss": 1.1873,
+      "step": 2858
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.719335993373989e-06,
+      "loss": 1.3396,
+      "step": 2859
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.71287106061601e-06,
+      "loss": 1.0663,
+      "step": 2860
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.706407136377956e-06,
+      "loss": 1.1971,
+      "step": 2861
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.69994422351012e-06,
+      "loss": 1.1144,
+      "step": 2862
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.69348232486236e-06,
+      "loss": 1.3066,
+      "step": 2863
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.687021443284071e-06,
+      "loss": 1.1565,
+      "step": 2864
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.680561581624213e-06,
+      "loss": 1.2525,
+      "step": 2865
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.674102742731294e-06,
+      "loss": 1.499,
+      "step": 2866
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.667644929453362e-06,
+      "loss": 1.1221,
+      "step": 2867
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.661188144638027e-06,
+      "loss": 1.1232,
+      "step": 2868
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.65473239113243e-06,
+      "loss": 1.4858,
+      "step": 2869
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.648277671783267e-06,
+      "loss": 1.144,
+      "step": 2870
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.64182398943678e-06,
+      "loss": 1.2586,
+      "step": 2871
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.635371346938746e-06,
+      "loss": 1.1665,
+      "step": 2872
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.628919747134489e-06,
+      "loss": 1.348,
+      "step": 2873
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.622469192868866e-06,
+      "loss": 1.3372,
+      "step": 2874
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.616019686986284e-06,
+      "loss": 1.1008,
+      "step": 2875
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.609571232330685e-06,
+      "loss": 1.2349,
+      "step": 2876
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.603123831745536e-06,
+      "loss": 1.2182,
+      "step": 2877
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.596677488073854e-06,
+      "loss": 1.0648,
+      "step": 2878
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5902322041581784e-06,
+      "loss": 1.3388,
+      "step": 2879
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.583787982840588e-06,
+      "loss": 1.1433,
+      "step": 2880
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.577344826962697e-06,
+      "loss": 1.3395,
+      "step": 2881
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5709027393656375e-06,
+      "loss": 1.265,
+      "step": 2882
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.564461722890082e-06,
+      "loss": 1.2952,
+      "step": 2883
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.558021780376223e-06,
+      "loss": 1.2287,
+      "step": 2884
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.551582914663781e-06,
+      "loss": 1.3146,
+      "step": 2885
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.545145128592009e-06,
+      "loss": 1.3374,
+      "step": 2886
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.538708424999675e-06,
+      "loss": 1.0336,
+      "step": 2887
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.532272806725072e-06,
+      "loss": 1.2254,
+      "step": 2888
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.525838276606016e-06,
+      "loss": 1.184,
+      "step": 2889
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.519404837479838e-06,
+      "loss": 1.0759,
+      "step": 2890
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.512972492183401e-06,
+      "loss": 1.1701,
+      "step": 2891
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.506541243553072e-06,
+      "loss": 1.2394,
+      "step": 2892
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.50011109442474e-06,
+      "loss": 1.2644,
+      "step": 2893
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.493682047633808e-06,
+      "loss": 1.3306,
+      "step": 2894
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4872541060151945e-06,
+      "loss": 1.3451,
+      "step": 2895
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.480827272403326e-06,
+      "loss": 1.2607,
+      "step": 2896
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.474401549632148e-06,
+      "loss": 1.3476,
+      "step": 2897
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.467976940535113e-06,
+      "loss": 1.2987,
+      "step": 2898
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.461553447945179e-06,
+      "loss": 1.1503,
+      "step": 2899
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.455131074694817e-06,
+      "loss": 1.3003,
+      "step": 2900
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.448709823615995e-06,
+      "loss": 1.1312,
+      "step": 2901
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4422896975402004e-06,
+      "loss": 1.1725,
+      "step": 2902
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.4358706992984164e-06,
+      "loss": 1.1013,
+      "step": 2903
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.429452831721128e-06,
+      "loss": 1.246,
+      "step": 2904
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.42303609763832e-06,
+      "loss": 1.2295,
+      "step": 2905
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.416620499879481e-06,
+      "loss": 1.0999,
+      "step": 2906
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.410206041273605e-06,
+      "loss": 1.1233,
+      "step": 2907
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.403792724649171e-06,
+      "loss": 1.1565,
+      "step": 2908
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.397380552834162e-06,
+      "loss": 1.1725,
+      "step": 2909
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.390969528656051e-06,
+      "loss": 1.2502,
+      "step": 2910
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.384559654941814e-06,
+      "loss": 1.3919,
+      "step": 2911
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.378150934517907e-06,
+      "loss": 1.1161,
+      "step": 2912
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.371743370210289e-06,
+      "loss": 1.0972,
+      "step": 2913
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.365336964844408e-06,
+      "loss": 1.2334,
+      "step": 2914
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.358931721245191e-06,
+      "loss": 1.3691,
+      "step": 2915
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.352527642237064e-06,
+      "loss": 1.2475,
+      "step": 2916
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.346124730643928e-06,
+      "loss": 1.2073,
+      "step": 2917
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.339722989289184e-06,
+      "loss": 1.2032,
+      "step": 2918
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.333322420995708e-06,
+      "loss": 1.1572,
+      "step": 2919
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.326923028585855e-06,
+      "loss": 1.0601,
+      "step": 2920
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.320524814881471e-06,
+      "loss": 1.2434,
+      "step": 2921
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.31412778270387e-06,
+      "loss": 1.3472,
+      "step": 2922
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.307731934873861e-06,
+      "loss": 1.1913,
+      "step": 2923
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.301337274211722e-06,
+      "loss": 1.1355,
+      "step": 2924
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.294943803537202e-06,
+      "loss": 1.2653,
+      "step": 2925
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.288551525669536e-06,
+      "loss": 1.276,
+      "step": 2926
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.282160443427424e-06,
+      "loss": 1.2339,
+      "step": 2927
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.275770559629042e-06,
+      "loss": 1.2532,
+      "step": 2928
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.269381877092046e-06,
+      "loss": 1.2595,
+      "step": 2929
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.262994398633548e-06,
+      "loss": 1.2507,
+      "step": 2930
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.256608127070137e-06,
+      "loss": 1.3127,
+      "step": 2931
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.250223065217869e-06,
+      "loss": 1.2122,
+      "step": 2932
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.243839215892263e-06,
+      "loss": 1.3168,
+      "step": 2933
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2374565819083155e-06,
+      "loss": 1.1456,
+      "step": 2934
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.231075166080468e-06,
+      "loss": 1.2514,
+      "step": 2935
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.224694971222641e-06,
+      "loss": 1.2326,
+      "step": 2936
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.2183160001482075e-06,
+      "loss": 1.3189,
+      "step": 2937
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.211938255670004e-06,
+      "loss": 1.2305,
+      "step": 2938
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.2055617406003285e-06,
+      "loss": 1.1677,
+      "step": 2939
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.199186457750931e-06,
+      "loss": 1.0443,
+      "step": 2940
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.192812409933026e-06,
+      "loss": 1.2447,
+      "step": 2941
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.186439599957273e-06,
+      "loss": 1.1176,
+      "step": 2942
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.180068030633798e-06,
+      "loss": 1.1699,
+      "step": 2943
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.173697704772164e-06,
+      "loss": 1.1879,
+      "step": 2944
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.167328625181405e-06,
+      "loss": 1.3318,
+      "step": 2945
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.160960794669993e-06,
+      "loss": 1.2646,
+      "step": 2946
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.15459421604585e-06,
+      "loss": 1.1781,
+      "step": 2947
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.148228892116351e-06,
+      "loss": 1.2955,
+      "step": 2948
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.141864825688307e-06,
+      "loss": 1.2141,
+      "step": 2949
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.13550201956799e-06,
+      "loss": 1.1736,
+      "step": 2950
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1291404765611084e-06,
+      "loss": 1.0989,
+      "step": 2951
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.122780199472808e-06,
+      "loss": 1.1207,
+      "step": 2952
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.116421191107687e-06,
+      "loss": 1.3529,
+      "step": 2953
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.110063454269778e-06,
+      "loss": 1.2208,
+      "step": 2954
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.1037069917625456e-06,
+      "loss": 1.2068,
+      "step": 2955
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.097351806388914e-06,
+      "loss": 1.3344,
+      "step": 2956
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.090997900951227e-06,
+      "loss": 1.2765,
+      "step": 2957
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.084645278251264e-06,
+      "loss": 1.2252,
+      "step": 2958
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.078293941090248e-06,
+      "loss": 1.2464,
+      "step": 2959
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.071943892268822e-06,
+      "loss": 1.3503,
+      "step": 2960
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.065595134587078e-06,
+      "loss": 1.2262,
+      "step": 2961
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.059247670844528e-06,
+      "loss": 1.2678,
+      "step": 2962
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.052901503840111e-06,
+      "loss": 1.2646,
+      "step": 2963
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.046556636372202e-06,
+      "loss": 1.2231,
+      "step": 2964
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.040213071238593e-06,
+      "loss": 1.0681,
+      "step": 2965
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.033870811236516e-06,
+      "loss": 1.1796,
+      "step": 2966
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.027529859162616e-06,
+      "loss": 1.1204,
+      "step": 2967
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.0211902178129655e-06,
+      "loss": 1.2553,
+      "step": 2968
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.014851889983058e-06,
+      "loss": 1.2685,
+      "step": 2969
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.008514878467805e-06,
+      "loss": 1.0615,
+      "step": 2970
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.002179186061542e-06,
+      "loss": 1.1945,
+      "step": 2971
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.995844815558026e-06,
+      "loss": 1.2625,
+      "step": 2972
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.989511769750421e-06,
+      "loss": 1.2704,
+      "step": 2973
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.983180051431315e-06,
+      "loss": 1.1343,
+      "step": 2974
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.976849663392708e-06,
+      "loss": 1.2909,
+      "step": 2975
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.97052060842601e-06,
+      "loss": 1.2774,
+      "step": 2976
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.964192889322052e-06,
+      "loss": 1.3532,
+      "step": 2977
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.957866508871068e-06,
+      "loss": 1.1862,
+      "step": 2978
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.951541469862706e-06,
+      "loss": 1.1634,
+      "step": 2979
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.945217775086017e-06,
+      "loss": 1.3039,
+      "step": 2980
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.938895427329463e-06,
+      "loss": 1.3144,
+      "step": 2981
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.932574429380918e-06,
+      "loss": 1.2895,
+      "step": 2982
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.926254784027648e-06,
+      "loss": 1.192,
+      "step": 2983
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.919936494056336e-06,
+      "loss": 1.096,
+      "step": 2984
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.913619562253055e-06,
+      "loss": 1.2414,
+      "step": 2985
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.907303991403289e-06,
+      "loss": 1.1724,
+      "step": 2986
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.900989784291912e-06,
+      "loss": 1.399,
+      "step": 2987
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.894676943703207e-06,
+      "loss": 1.3381,
+      "step": 2988
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.888365472420852e-06,
+      "loss": 1.2138,
+      "step": 2989
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.882055373227914e-06,
+      "loss": 1.1171,
+      "step": 2990
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.875746648906862e-06,
+      "loss": 1.1342,
+      "step": 2991
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.869439302239556e-06,
+      "loss": 1.2785,
+      "step": 2992
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.863133336007248e-06,
+      "loss": 1.3908,
+      "step": 2993
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.85682875299059e-06,
+      "loss": 1.204,
+      "step": 2994
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8505255559696075e-06,
+      "loss": 1.2581,
+      "step": 2995
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.844223747723728e-06,
+      "loss": 1.1434,
+      "step": 2996
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.837923331031761e-06,
+      "loss": 1.2085,
+      "step": 2997
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.831624308671905e-06,
+      "loss": 1.2299,
+      "step": 2998
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8253266834217445e-06,
+      "loss": 1.2631,
+      "step": 2999
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.819030458058243e-06,
+      "loss": 1.0902,
+      "step": 3000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.8127356353577535e-06,
+      "loss": 1.249,
+      "step": 3001
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.806442218096001e-06,
+      "loss": 1.3046,
+      "step": 3002
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.800150209048098e-06,
+      "loss": 1.171,
+      "step": 3003
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.79385961098854e-06,
+      "loss": 1.2519,
+      "step": 3004
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.787570426691189e-06,
+      "loss": 1.3299,
+      "step": 3005
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7812826589292936e-06,
+      "loss": 1.3209,
+      "step": 3006
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7749963104754725e-06,
+      "loss": 1.2653,
+      "step": 3007
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.768711384101712e-06,
+      "loss": 1.2535,
+      "step": 3008
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.76242788257939e-06,
+      "loss": 1.2719,
+      "step": 3009
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.756145808679243e-06,
+      "loss": 1.3123,
+      "step": 3010
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.749865165171375e-06,
+      "loss": 1.1228,
+      "step": 3011
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.74358595482527e-06,
+      "loss": 1.2798,
+      "step": 3012
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.737308180409767e-06,
+      "loss": 1.1235,
+      "step": 3013
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.731031844693088e-06,
+      "loss": 1.266,
+      "step": 3014
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.724756950442808e-06,
+      "loss": 1.3759,
+      "step": 3015
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.718483500425868e-06,
+      "loss": 1.2322,
+      "step": 3016
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.712211497408579e-06,
+      "loss": 1.2885,
+      "step": 3017
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.705940944156604e-06,
+      "loss": 1.2338,
+      "step": 3018
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.699671843434973e-06,
+      "loss": 1.278,
+      "step": 3019
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.69340419800808e-06,
+      "loss": 1.2889,
+      "step": 3020
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.687138010639667e-06,
+      "loss": 1.2528,
+      "step": 3021
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.680873284092839e-06,
+      "loss": 1.271,
+      "step": 3022
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6746100211300545e-06,
+      "loss": 1.2565,
+      "step": 3023
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.668348224513127e-06,
+      "loss": 1.1268,
+      "step": 3024
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6620878970032295e-06,
+      "loss": 1.1682,
+      "step": 3025
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.655829041360877e-06,
+      "loss": 1.2272,
+      "step": 3026
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.649571660345945e-06,
+      "loss": 1.1538,
+      "step": 3027
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.643315756717647e-06,
+      "loss": 1.2615,
+      "step": 3028
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.637061333234557e-06,
+      "loss": 1.2014,
+      "step": 3029
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.630808392654593e-06,
+      "loss": 1.3074,
+      "step": 3030
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.624556937735014e-06,
+      "loss": 1.2252,
+      "step": 3031
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.618306971232431e-06,
+      "loss": 1.273,
+      "step": 3032
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.612058495902791e-06,
+      "loss": 1.2439,
+      "step": 3033
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.605811514501392e-06,
+      "loss": 1.2136,
+      "step": 3034
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.599566029782863e-06,
+      "loss": 1.2445,
+      "step": 3035
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.593322044501185e-06,
+      "loss": 1.4015,
+      "step": 3036
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.587079561409672e-06,
+      "loss": 1.2279,
+      "step": 3037
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.580838583260969e-06,
+      "loss": 1.2643,
+      "step": 3038
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.574599112807071e-06,
+      "loss": 1.1962,
+      "step": 3039
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.568361152799293e-06,
+      "loss": 1.3277,
+      "step": 3040
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.562124705988297e-06,
+      "loss": 1.1256,
+      "step": 3041
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.555889775124077e-06,
+      "loss": 1.193,
+      "step": 3042
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.549656362955945e-06,
+      "loss": 1.1913,
+      "step": 3043
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.54342447223256e-06,
+      "loss": 1.1779,
+      "step": 3044
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.537194105701896e-06,
+      "loss": 1.1885,
+      "step": 3045
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5309652661112635e-06,
+      "loss": 1.3098,
+      "step": 3046
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.524737956207303e-06,
+      "loss": 1.0112,
+      "step": 3047
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5185121787359686e-06,
+      "loss": 1.2237,
+      "step": 3048
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.51228793644255e-06,
+      "loss": 1.1617,
+      "step": 3049
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.50606523207165e-06,
+      "loss": 1.2901,
+      "step": 3050
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4998440683672e-06,
+      "loss": 1.2851,
+      "step": 3051
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4936244480724575e-06,
+      "loss": 1.1697,
+      "step": 3052
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.487406373929983e-06,
+      "loss": 1.1988,
+      "step": 3053
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.48118984868167e-06,
+      "loss": 1.1065,
+      "step": 3054
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.474974875068722e-06,
+      "loss": 1.0892,
+      "step": 3055
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.468761455831656e-06,
+      "loss": 1.1785,
+      "step": 3056
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.462549593710317e-06,
+      "loss": 1.118,
+      "step": 3057
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.456339291443845e-06,
+      "loss": 1.1386,
+      "step": 3058
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.450130551770707e-06,
+      "loss": 1.1744,
+      "step": 3059
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.443923377428672e-06,
+      "loss": 1.1019,
+      "step": 3060
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.437717771154821e-06,
+      "loss": 1.2028,
+      "step": 3061
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.431513735685544e-06,
+      "loss": 1.3428,
+      "step": 3062
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4253112737565426e-06,
+      "loss": 1.2478,
+      "step": 3063
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.419110388102818e-06,
+      "loss": 1.4234,
+      "step": 3064
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.41291108145868e-06,
+      "loss": 1.1231,
+      "step": 3065
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.406713356557739e-06,
+      "loss": 1.0659,
+      "step": 3066
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.40051721613291e-06,
+      "loss": 1.247,
+      "step": 3067
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3943226629164144e-06,
+      "loss": 1.0794,
+      "step": 3068
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3881296996397624e-06,
+      "loss": 1.2002,
+      "step": 3069
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.381938329033774e-06,
+      "loss": 1.1788,
+      "step": 3070
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.37574855382856e-06,
+      "loss": 1.2598,
+      "step": 3071
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3695603767535275e-06,
+      "loss": 1.2185,
+      "step": 3072
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.363373800537388e-06,
+      "loss": 1.1609,
+      "step": 3073
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.357188827908134e-06,
+      "loss": 1.234,
+      "step": 3074
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.351005461593063e-06,
+      "loss": 1.1114,
+      "step": 3075
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.344823704318752e-06,
+      "loss": 1.2228,
+      "step": 3076
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.338643558811082e-06,
+      "loss": 1.2862,
+      "step": 3077
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.332465027795208e-06,
+      "loss": 1.2987,
+      "step": 3078
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.326288113995589e-06,
+      "loss": 1.3619,
+      "step": 3079
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.320112820135962e-06,
+      "loss": 1.2394,
+      "step": 3080
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.313939148939348e-06,
+      "loss": 1.0974,
+      "step": 3081
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.307767103128058e-06,
+      "loss": 1.1839,
+      "step": 3082
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.301596685423679e-06,
+      "loss": 1.4732,
+      "step": 3083
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.295427898547091e-06,
+      "loss": 1.2497,
+      "step": 3084
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2892607452184476e-06,
+      "loss": 1.206,
+      "step": 3085
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.283095228157179e-06,
+      "loss": 1.2441,
+      "step": 3086
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2769313500820025e-06,
+      "loss": 1.2438,
+      "step": 3087
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.270769113710904e-06,
+      "loss": 1.3696,
+      "step": 3088
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.264608521761153e-06,
+      "loss": 1.2112,
+      "step": 3089
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.258449576949292e-06,
+      "loss": 1.2208,
+      "step": 3090
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.252292281991132e-06,
+      "loss": 1.2944,
+      "step": 3091
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.246136639601763e-06,
+      "loss": 1.3976,
+      "step": 3092
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.239982652495539e-06,
+      "loss": 1.312,
+      "step": 3093
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.233830323386092e-06,
+      "loss": 1.4503,
+      "step": 3094
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.227679654986323e-06,
+      "loss": 1.1689,
+      "step": 3095
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.221530650008391e-06,
+      "loss": 1.2649,
+      "step": 3096
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.215383311163734e-06,
+      "loss": 1.2083,
+      "step": 3097
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.209237641163041e-06,
+      "loss": 1.3635,
+      "step": 3098
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.203093642716278e-06,
+      "loss": 1.1644,
+      "step": 3099
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.196951318532671e-06,
+      "loss": 1.1891,
+      "step": 3100
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.190810671320704e-06,
+      "loss": 1.1783,
+      "step": 3101
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.184671703788123e-06,
+      "loss": 1.2172,
+      "step": 3102
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.178534418641933e-06,
+      "loss": 1.1414,
+      "step": 3103
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.172398818588394e-06,
+      "loss": 1.2855,
+      "step": 3104
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.166264906333038e-06,
+      "loss": 1.0882,
+      "step": 3105
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.160132684580632e-06,
+      "loss": 1.0945,
+      "step": 3106
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1540021560352125e-06,
+      "loss": 1.2378,
+      "step": 3107
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.147873323400058e-06,
+      "loss": 1.193,
+      "step": 3108
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.141746189377711e-06,
+      "loss": 1.2871,
+      "step": 3109
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.135620756669954e-06,
+      "loss": 1.248,
+      "step": 3110
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.129497027977829e-06,
+      "loss": 1.1641,
+      "step": 3111
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.123375006001622e-06,
+      "loss": 1.3239,
+      "step": 3112
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.117254693440864e-06,
+      "loss": 1.4828,
+      "step": 3113
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.111136092994334e-06,
+      "loss": 1.2628,
+      "step": 3114
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.105019207360057e-06,
+      "loss": 1.1152,
+      "step": 3115
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.098904039235305e-06,
+      "loss": 1.2896,
+      "step": 3116
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.092790591316586e-06,
+      "loss": 1.1845,
+      "step": 3117
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.086678866299656e-06,
+      "loss": 1.2317,
+      "step": 3118
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.080568866879505e-06,
+      "loss": 1.1433,
+      "step": 3119
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.074460595750362e-06,
+      "loss": 1.1847,
+      "step": 3120
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.068354055605705e-06,
+      "loss": 1.2266,
+      "step": 3121
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.062249249138235e-06,
+      "loss": 1.1049,
+      "step": 3122
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.056146179039899e-06,
+      "loss": 1.2853,
+      "step": 3123
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.050044848001866e-06,
+      "loss": 1.315,
+      "step": 3124
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.043945258714553e-06,
+      "loss": 1.2694,
+      "step": 3125
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.037847413867595e-06,
+      "loss": 1.2105,
+      "step": 3126
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.031751316149869e-06,
+      "loss": 1.1953,
+      "step": 3127
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0256569682494795e-06,
+      "loss": 1.3132,
+      "step": 3128
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.01956437285375e-06,
+      "loss": 1.0918,
+      "step": 3129
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.013473532649246e-06,
+      "loss": 1.1408,
+      "step": 3130
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.007384450321742e-06,
+      "loss": 1.1369,
+      "step": 3131
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.0012971285562535e-06,
+      "loss": 1.2912,
+      "step": 3132
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.995211570037013e-06,
+      "loss": 1.2189,
+      "step": 3133
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.989127777447471e-06,
+      "loss": 1.3475,
+      "step": 3134
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.983045753470308e-06,
+      "loss": 1.3321,
+      "step": 3135
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.976965500787414e-06,
+      "loss": 1.3071,
+      "step": 3136
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9708870220799095e-06,
+      "loss": 1.1346,
+      "step": 3137
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.964810320028129e-06,
+      "loss": 1.2361,
+      "step": 3138
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.958735397311617e-06,
+      "loss": 1.2837,
+      "step": 3139
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9526622566091404e-06,
+      "loss": 1.4368,
+      "step": 3140
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9465909005986765e-06,
+      "loss": 1.1253,
+      "step": 3141
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.940521331957419e-06,
+      "loss": 1.3117,
+      "step": 3142
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.934453553361775e-06,
+      "loss": 1.1264,
+      "step": 3143
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9283875674873524e-06,
+      "loss": 1.2597,
+      "step": 3144
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.922323377008981e-06,
+      "loss": 1.2948,
+      "step": 3145
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.91626098460069e-06,
+      "loss": 1.2191,
+      "step": 3146
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.910200392935718e-06,
+      "loss": 1.1933,
+      "step": 3147
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.904141604686516e-06,
+      "loss": 1.204,
+      "step": 3148
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.898084622524729e-06,
+      "loss": 1.2718,
+      "step": 3149
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.892029449121214e-06,
+      "loss": 1.2953,
+      "step": 3150
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.885976087146024e-06,
+      "loss": 1.2345,
+      "step": 3151
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.879924539268421e-06,
+      "loss": 1.3125,
+      "step": 3152
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.873874808156856e-06,
+      "loss": 1.334,
+      "step": 3153
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.867826896478992e-06,
+      "loss": 1.3496,
+      "step": 3154
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.861780806901682e-06,
+      "loss": 1.2617,
+      "step": 3155
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8557365420909725e-06,
+      "loss": 1.1874,
+      "step": 3156
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8496941047121166e-06,
+      "loss": 1.2976,
+      "step": 3157
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.843653497429546e-06,
+      "loss": 1.3516,
+      "step": 3158
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8376147229069e-06,
+      "loss": 1.2656,
+      "step": 3159
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.831577783807005e-06,
+      "loss": 1.2233,
+      "step": 3160
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.825542682791874e-06,
+      "loss": 1.2027,
+      "step": 3161
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.819509422522711e-06,
+      "loss": 1.2808,
+      "step": 3162
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.813478005659905e-06,
+      "loss": 1.193,
+      "step": 3163
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.80744843486305e-06,
+      "loss": 1.231,
+      "step": 3164
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.801420712790905e-06,
+      "loss": 1.3009,
+      "step": 3165
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.795394842101423e-06,
+      "loss": 1.1306,
+      "step": 3166
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.789370825451737e-06,
+      "loss": 1.2605,
+      "step": 3167
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.78334866549816e-06,
+      "loss": 1.3126,
+      "step": 3168
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7773283648961995e-06,
+      "loss": 1.276,
+      "step": 3169
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.771309926300535e-06,
+      "loss": 1.3401,
+      "step": 3170
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7652933523650205e-06,
+      "loss": 1.3101,
+      "step": 3171
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.759278645742692e-06,
+      "loss": 1.2947,
+      "step": 3172
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7532658090857574e-06,
+      "loss": 1.2317,
+      "step": 3173
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.747254845045609e-06,
+      "loss": 1.0691,
+      "step": 3174
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.741245756272813e-06,
+      "loss": 1.3264,
+      "step": 3175
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.735238545417101e-06,
+      "loss": 1.1405,
+      "step": 3176
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.729233215127378e-06,
+      "loss": 1.21,
+      "step": 3177
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.723229768051719e-06,
+      "loss": 1.3071,
+      "step": 3178
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.717228206837375e-06,
+      "loss": 1.2616,
+      "step": 3179
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.711228534130766e-06,
+      "loss": 1.2487,
+      "step": 3180
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.70523075257747e-06,
+      "loss": 1.1329,
+      "step": 3181
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.699234864822239e-06,
+      "loss": 1.199,
+      "step": 3182
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.693240873508981e-06,
+      "loss": 1.2266,
+      "step": 3183
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.687248781280781e-06,
+      "loss": 1.1915,
+      "step": 3184
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.681258590779873e-06,
+      "loss": 1.2146,
+      "step": 3185
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.675270304647665e-06,
+      "loss": 1.1971,
+      "step": 3186
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.669283925524716e-06,
+      "loss": 1.2539,
+      "step": 3187
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.663299456050743e-06,
+      "loss": 1.1801,
+      "step": 3188
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.657316898864634e-06,
+      "loss": 1.2433,
+      "step": 3189
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.651336256604416e-06,
+      "loss": 1.1763,
+      "step": 3190
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.645357531907289e-06,
+      "loss": 1.18,
+      "step": 3191
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.639380727409594e-06,
+      "loss": 1.2719,
+      "step": 3192
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6334058457468265e-06,
+      "loss": 1.2819,
+      "step": 3193
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.627432889553646e-06,
+      "loss": 1.2319,
+      "step": 3194
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.621461861463846e-06,
+      "loss": 1.3225,
+      "step": 3195
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.615492764110388e-06,
+      "loss": 1.2688,
+      "step": 3196
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.609525600125368e-06,
+      "loss": 1.3052,
+      "step": 3197
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.603560372140029e-06,
+      "loss": 1.0922,
+      "step": 3198
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.597597082784776e-06,
+      "loss": 1.3629,
+      "step": 3199
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.59163573468914e-06,
+      "loss": 1.1848,
+      "step": 3200
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.585676330481806e-06,
+      "loss": 1.2642,
+      "step": 3201
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.579718872790607e-06,
+      "loss": 1.2592,
+      "step": 3202
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5737633642425e-06,
+      "loss": 1.3126,
+      "step": 3203
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.567809807463606e-06,
+      "loss": 1.3768,
+      "step": 3204
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5618582050791645e-06,
+      "loss": 1.1632,
+      "step": 3205
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.555908559713561e-06,
+      "loss": 1.3178,
+      "step": 3206
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5499608739903256e-06,
+      "loss": 1.3062,
+      "step": 3207
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.544015150532109e-06,
+      "loss": 1.0675,
+      "step": 3208
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.538071391960715e-06,
+      "loss": 1.3199,
+      "step": 3209
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.532129600897065e-06,
+      "loss": 1.3061,
+      "step": 3210
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5261897799612155e-06,
+      "loss": 1.2167,
+      "step": 3211
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.520251931772365e-06,
+      "loss": 1.1282,
+      "step": 3212
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.514316058948828e-06,
+      "loss": 1.2489,
+      "step": 3213
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.508382164108059e-06,
+      "loss": 1.0672,
+      "step": 3214
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.502450249866638e-06,
+      "loss": 1.1245,
+      "step": 3215
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.496520318840266e-06,
+      "loss": 1.1194,
+      "step": 3216
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4905923736437685e-06,
+      "loss": 1.2461,
+      "step": 3217
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.484666416891109e-06,
+      "loss": 1.1712,
+      "step": 3218
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.478742451195358e-06,
+      "loss": 1.2176,
+      "step": 3219
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.472820479168721e-06,
+      "loss": 1.2621,
+      "step": 3220
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.466900503422516e-06,
+      "loss": 1.1351,
+      "step": 3221
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.46098252656718e-06,
+      "loss": 1.2081,
+      "step": 3222
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.455066551212278e-06,
+      "loss": 1.2408,
+      "step": 3223
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.44915257996648e-06,
+      "loss": 1.245,
+      "step": 3224
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.443240615437586e-06,
+      "loss": 1.2256,
+      "step": 3225
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.437330660232498e-06,
+      "loss": 1.1859,
+      "step": 3226
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.431422716957236e-06,
+      "loss": 1.2498,
+      "step": 3227
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.425516788216942e-06,
+      "loss": 1.2832,
+      "step": 3228
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.419612876615854e-06,
+      "loss": 1.1837,
+      "step": 3229
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.413710984757335e-06,
+      "loss": 1.1538,
+      "step": 3230
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.40781111524385e-06,
+      "loss": 1.2111,
+      "step": 3231
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.401913270676971e-06,
+      "loss": 1.1896,
+      "step": 3232
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3960174536573765e-06,
+      "loss": 1.1434,
+      "step": 3233
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3901236667848586e-06,
+      "loss": 1.2968,
+      "step": 3234
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.384231912658311e-06,
+      "loss": 1.2088,
+      "step": 3235
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.378342193875729e-06,
+      "loss": 1.2186,
+      "step": 3236
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.372454513034208e-06,
+      "loss": 1.3677,
+      "step": 3237
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.366568872729944e-06,
+      "loss": 1.228,
+      "step": 3238
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.360685275558244e-06,
+      "loss": 1.1476,
+      "step": 3239
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.354803724113506e-06,
+      "loss": 1.1301,
+      "step": 3240
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.348924220989227e-06,
+      "loss": 1.1432,
+      "step": 3241
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.343046768777999e-06,
+      "loss": 1.1056,
+      "step": 3242
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.337171370071508e-06,
+      "loss": 1.1849,
+      "step": 3243
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.33129802746054e-06,
+      "loss": 1.177,
+      "step": 3244
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.325426743534978e-06,
+      "loss": 1.2559,
+      "step": 3245
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3195575208837865e-06,
+      "loss": 1.2029,
+      "step": 3246
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.313690362095028e-06,
+      "loss": 1.1687,
+      "step": 3247
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.307825269755846e-06,
+      "loss": 1.1648,
+      "step": 3248
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.301962246452486e-06,
+      "loss": 1.0714,
+      "step": 3249
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.296101294770276e-06,
+      "loss": 1.2359,
+      "step": 3250
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.290242417293628e-06,
+      "loss": 1.273,
+      "step": 3251
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.28438561660604e-06,
+      "loss": 1.1143,
+      "step": 3252
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.278530895290091e-06,
+      "loss": 1.1741,
+      "step": 3253
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.27267825592745e-06,
+      "loss": 1.1118,
+      "step": 3254
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.266827701098871e-06,
+      "loss": 1.2696,
+      "step": 3255
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.260979233384178e-06,
+      "loss": 1.1997,
+      "step": 3256
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.255132855362277e-06,
+      "loss": 1.3202,
+      "step": 3257
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.249288569611155e-06,
+      "loss": 1.2355,
+      "step": 3258
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2434463787078815e-06,
+      "loss": 1.1377,
+      "step": 3259
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.237606285228592e-06,
+      "loss": 1.4336,
+      "step": 3260
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.231768291748506e-06,
+      "loss": 1.2329,
+      "step": 3261
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.225932400841912e-06,
+      "loss": 1.1774,
+      "step": 3262
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.22009861508217e-06,
+      "loss": 1.2557,
+      "step": 3263
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.214266937041721e-06,
+      "loss": 1.5202,
+      "step": 3264
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.208437369292061e-06,
+      "loss": 1.1467,
+      "step": 3265
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.202609914403773e-06,
+      "loss": 1.1098,
+      "step": 3266
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.196784574946496e-06,
+      "loss": 1.3064,
+      "step": 3267
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.190961353488941e-06,
+      "loss": 1.3324,
+      "step": 3268
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.185140252598881e-06,
+      "loss": 1.148,
+      "step": 3269
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1793212748431565e-06,
+      "loss": 1.2666,
+      "step": 3270
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1735044227876795e-06,
+      "loss": 1.0463,
+      "step": 3271
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.167689698997414e-06,
+      "loss": 1.3582,
+      "step": 3272
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.161877106036386e-06,
+      "loss": 1.1772,
+      "step": 3273
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.156066646467683e-06,
+      "loss": 1.2048,
+      "step": 3274
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.150258322853461e-06,
+      "loss": 1.3268,
+      "step": 3275
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.144452137754921e-06,
+      "loss": 1.2367,
+      "step": 3276
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.13864809373233e-06,
+      "loss": 1.1122,
+      "step": 3277
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.132846193345007e-06,
+      "loss": 1.2769,
+      "step": 3278
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.127046439151322e-06,
+      "loss": 1.1536,
+      "step": 3279
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.121248833708712e-06,
+      "loss": 1.2282,
+      "step": 3280
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.115453379573647e-06,
+      "loss": 1.2053,
+      "step": 3281
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.109660079301668e-06,
+      "loss": 1.1834,
+      "step": 3282
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.103868935447354e-06,
+      "loss": 1.1816,
+      "step": 3283
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.098079950564332e-06,
+      "loss": 1.2367,
+      "step": 3284
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.092293127205287e-06,
+      "loss": 1.2959,
+      "step": 3285
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.086508467921943e-06,
+      "loss": 1.1586,
+      "step": 3286
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.080725975265073e-06,
+      "loss": 1.1397,
+      "step": 3287
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.074945651784491e-06,
+      "loss": 1.2753,
+      "step": 3288
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0691675000290566e-06,
+      "loss": 1.2994,
+      "step": 3289
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.063391522546677e-06,
+      "loss": 1.2316,
+      "step": 3290
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.057617721884293e-06,
+      "loss": 1.3515,
+      "step": 3291
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.051846100587883e-06,
+      "loss": 1.1893,
+      "step": 3292
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.046076661202479e-06,
+      "loss": 1.1378,
+      "step": 3293
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.040309406272135e-06,
+      "loss": 1.2265,
+      "step": 3294
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.034544338339953e-06,
+      "loss": 1.2637,
+      "step": 3295
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.028781459948063e-06,
+      "loss": 1.178,
+      "step": 3296
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.02302077363763e-06,
+      "loss": 1.2474,
+      "step": 3297
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.01726228194886e-06,
+      "loss": 1.2741,
+      "step": 3298
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.011505987420983e-06,
+      "loss": 1.3037,
+      "step": 3299
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.005751892592266e-06,
+      "loss": 1.2827,
+      "step": 3300
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 1.2567,
+      "step": 3301
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.994250312180511e-06,
+      "loss": 1.1708,
+      "step": 3302
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.98850283166915e-06,
+      "loss": 1.1695,
+      "step": 3303
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.98275756100029e-06,
+      "loss": 1.2116,
+      "step": 3304
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.977014502707341e-06,
+      "loss": 1.2607,
+      "step": 3305
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9712736593227285e-06,
+      "loss": 1.2417,
+      "step": 3306
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.965535033377902e-06,
+      "loss": 1.0922,
+      "step": 3307
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.959798627403331e-06,
+      "loss": 1.1845,
+      "step": 3308
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.954064443928515e-06,
+      "loss": 1.2731,
+      "step": 3309
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.948332485481972e-06,
+      "loss": 1.4129,
+      "step": 3310
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.942602754591229e-06,
+      "loss": 1.1736,
+      "step": 3311
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.936875253782841e-06,
+      "loss": 1.2427,
+      "step": 3312
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.931149985582368e-06,
+      "loss": 1.2025,
+      "step": 3313
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.925426952514399e-06,
+      "loss": 1.2647,
+      "step": 3314
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.919706157102533e-06,
+      "loss": 1.2088,
+      "step": 3315
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.91398760186938e-06,
+      "loss": 1.2307,
+      "step": 3316
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.908271289336561e-06,
+      "loss": 1.1425,
+      "step": 3317
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9025572220247074e-06,
+      "loss": 1.0238,
+      "step": 3318
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.896845402453466e-06,
+      "loss": 1.2632,
+      "step": 3319
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.891135833141495e-06,
+      "loss": 1.3421,
+      "step": 3320
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.885428516606449e-06,
+      "loss": 1.1414,
+      "step": 3321
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.879723455364997e-06,
+      "loss": 1.2919,
+      "step": 3322
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.87402065193281e-06,
+      "loss": 1.3276,
+      "step": 3323
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.868320108824564e-06,
+      "loss": 1.2611,
+      "step": 3324
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.862621828553941e-06,
+      "loss": 1.2209,
+      "step": 3325
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.856925813633628e-06,
+      "loss": 1.1751,
+      "step": 3326
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.851232066575304e-06,
+      "loss": 1.1502,
+      "step": 3327
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.845540589889656e-06,
+      "loss": 1.2423,
+      "step": 3328
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.839851386086358e-06,
+      "loss": 1.2409,
+      "step": 3329
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.834164457674099e-06,
+      "loss": 1.2216,
+      "step": 3330
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8284798071605565e-06,
+      "loss": 1.3147,
+      "step": 3331
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8227974370524e-06,
+      "loss": 1.2244,
+      "step": 3332
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.817117349855297e-06,
+      "loss": 1.1438,
+      "step": 3333
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.811439548073903e-06,
+      "loss": 1.2928,
+      "step": 3334
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.805764034211876e-06,
+      "loss": 1.1591,
+      "step": 3335
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8000908107718625e-06,
+      "loss": 1.2627,
+      "step": 3336
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.794419880255492e-06,
+      "loss": 1.2534,
+      "step": 3337
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.7887512451633874e-06,
+      "loss": 1.2297,
+      "step": 3338
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.783084907995156e-06,
+      "loss": 1.2474,
+      "step": 3339
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.777420871249399e-06,
+      "loss": 1.1493,
+      "step": 3340
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.7717591374237e-06,
+      "loss": 1.2552,
+      "step": 3341
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.766099709014628e-06,
+      "loss": 1.2785,
+      "step": 3342
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.760442588517729e-06,
+      "loss": 1.2563,
+      "step": 3343
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.754787778427533e-06,
+      "loss": 1.1794,
+      "step": 3344
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.74913528123756e-06,
+      "loss": 1.3383,
+      "step": 3345
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.743485099440307e-06,
+      "loss": 1.2304,
+      "step": 3346
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7378372355272435e-06,
+      "loss": 1.2727,
+      "step": 3347
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.732191691988822e-06,
+      "loss": 1.2368,
+      "step": 3348
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.726548471314465e-06,
+      "loss": 1.1599,
+      "step": 3349
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.720907575992585e-06,
+      "loss": 1.2386,
+      "step": 3350
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.715269008510552e-06,
+      "loss": 1.228,
+      "step": 3351
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.709632771354727e-06,
+      "loss": 1.244,
+      "step": 3352
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.703998867010431e-06,
+      "loss": 1.2314,
+      "step": 3353
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.698367297961954e-06,
+      "loss": 1.2816,
+      "step": 3354
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.69273806669257e-06,
+      "loss": 1.0855,
+      "step": 3355
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.687111175684509e-06,
+      "loss": 1.2332,
+      "step": 3356
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.681486627418979e-06,
+      "loss": 1.1775,
+      "step": 3357
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.675864424376146e-06,
+      "loss": 1.0634,
+      "step": 3358
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.670244569035145e-06,
+      "loss": 1.1775,
+      "step": 3359
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.664627063874083e-06,
+      "loss": 1.1731,
+      "step": 3360
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.659011911370017e-06,
+      "loss": 1.231,
+      "step": 3361
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.653399113998981e-06,
+      "loss": 1.3224,
+      "step": 3362
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.647788674235961e-06,
+      "loss": 1.3008,
+      "step": 3363
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.642180594554901e-06,
+      "loss": 1.2022,
+      "step": 3364
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.636574877428718e-06,
+      "loss": 1.1906,
+      "step": 3365
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.630971525329274e-06,
+      "loss": 1.357,
+      "step": 3366
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.625370540727389e-06,
+      "loss": 1.1882,
+      "step": 3367
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6197719260928494e-06,
+      "loss": 1.2653,
+      "step": 3368
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.614175683894384e-06,
+      "loss": 1.2329,
+      "step": 3369
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.608581816599688e-06,
+      "loss": 1.2312,
+      "step": 3370
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6029903266753975e-06,
+      "loss": 1.1434,
+      "step": 3371
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.597401216587104e-06,
+      "loss": 1.2826,
+      "step": 3372
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.5918144887993575e-06,
+      "loss": 1.1862,
+      "step": 3373
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.586230145775647e-06,
+      "loss": 1.1845,
+      "step": 3374
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.58064818997841e-06,
+      "loss": 1.1815,
+      "step": 3375
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5750686238690455e-06,
+      "loss": 1.2498,
+      "step": 3376
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.569491449907878e-06,
+      "loss": 1.3113,
+      "step": 3377
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.563916670554196e-06,
+      "loss": 1.2865,
+      "step": 3378
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.55834428826622e-06,
+      "loss": 1.1134,
+      "step": 3379
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.552774305501115e-06,
+      "loss": 1.2662,
+      "step": 3380
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.547206724714996e-06,
+      "loss": 1.4008,
+      "step": 3381
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.541641548362911e-06,
+      "loss": 1.1612,
+      "step": 3382
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.536078778898845e-06,
+      "loss": 1.2733,
+      "step": 3383
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.530518418775734e-06,
+      "loss": 1.1122,
+      "step": 3384
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.524960470445437e-06,
+      "loss": 1.2143,
+      "step": 3385
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.519404936358763e-06,
+      "loss": 1.2724,
+      "step": 3386
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.513851818965449e-06,
+      "loss": 1.0872,
+      "step": 3387
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.508301120714162e-06,
+      "loss": 1.235,
+      "step": 3388
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5027528440525134e-06,
+      "loss": 1.2296,
+      "step": 3389
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.497206991427036e-06,
+      "loss": 1.3127,
+      "step": 3390
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4916635652832076e-06,
+      "loss": 1.2498,
+      "step": 3391
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4861225680654196e-06,
+      "loss": 1.2037,
+      "step": 3392
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.480584002216999e-06,
+      "loss": 1.3434,
+      "step": 3393
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.475047870180207e-06,
+      "loss": 1.2851,
+      "step": 3394
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.469514174396221e-06,
+      "loss": 1.203,
+      "step": 3395
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.463982917305155e-06,
+      "loss": 1.1462,
+      "step": 3396
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.45845410134604e-06,
+      "loss": 1.1997,
+      "step": 3397
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.452927728956832e-06,
+      "loss": 1.0916,
+      "step": 3398
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.447403802574406e-06,
+      "loss": 1.1735,
+      "step": 3399
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.441882324634565e-06,
+      "loss": 1.171,
+      "step": 3400
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.436363297572036e-06,
+      "loss": 1.2119,
+      "step": 3401
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.430846723820453e-06,
+      "loss": 1.1988,
+      "step": 3402
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.425332605812377e-06,
+      "loss": 1.0945,
+      "step": 3403
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.419820945979278e-06,
+      "loss": 1.2398,
+      "step": 3404
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4143117467515514e-06,
+      "loss": 1.2515,
+      "step": 3405
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4088050105585075e-06,
+      "loss": 1.2387,
+      "step": 3406
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.403300739828364e-06,
+      "loss": 1.0942,
+      "step": 3407
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.397798936988251e-06,
+      "loss": 1.104,
+      "step": 3408
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.392299604464213e-06,
+      "loss": 1.1661,
+      "step": 3409
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.386802744681209e-06,
+      "loss": 1.1952,
+      "step": 3410
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.381308360063107e-06,
+      "loss": 1.2586,
+      "step": 3411
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3758164530326785e-06,
+      "loss": 1.3828,
+      "step": 3412
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3703270260116025e-06,
+      "loss": 1.2805,
+      "step": 3413
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3648400814204665e-06,
+      "loss": 1.1712,
+      "step": 3414
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.359355621678765e-06,
+      "loss": 1.3427,
+      "step": 3415
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.353873649204899e-06,
+      "loss": 1.1013,
+      "step": 3416
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.348394166416169e-06,
+      "loss": 1.0994,
+      "step": 3417
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.342917175728773e-06,
+      "loss": 1.1985,
+      "step": 3418
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.337442679557815e-06,
+      "loss": 1.1748,
+      "step": 3419
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3319706803173e-06,
+      "loss": 1.1071,
+      "step": 3420
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.326501180420137e-06,
+      "loss": 1.2526,
+      "step": 3421
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.321034182278122e-06,
+      "loss": 1.2809,
+      "step": 3422
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.315569688301953e-06,
+      "loss": 1.4285,
+      "step": 3423
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.310107700901224e-06,
+      "loss": 1.2628,
+      "step": 3424
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.304648222484414e-06,
+      "loss": 1.2805,
+      "step": 3425
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.299191255458922e-06,
+      "loss": 1.2328,
+      "step": 3426
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.293736802231014e-06,
+      "loss": 1.1596,
+      "step": 3427
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.288284865205856e-06,
+      "loss": 1.3575,
+      "step": 3428
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.282835446787504e-06,
+      "loss": 1.2775,
+      "step": 3429
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.277388549378902e-06,
+      "loss": 1.1963,
+      "step": 3430
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.271944175381886e-06,
+      "loss": 1.2221,
+      "step": 3431
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.266502327197182e-06,
+      "loss": 1.2081,
+      "step": 3432
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.261063007224393e-06,
+      "loss": 1.1365,
+      "step": 3433
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.255626217862014e-06,
+      "loss": 1.3953,
+      "step": 3434
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.250191961507416e-06,
+      "loss": 1.2461,
+      "step": 3435
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.244760240556864e-06,
+      "loss": 1.2521,
+      "step": 3436
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.239331057405505e-06,
+      "loss": 1.1401,
+      "step": 3437
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.233904414447355e-06,
+      "loss": 1.3424,
+      "step": 3438
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.2284803140753205e-06,
+      "loss": 1.3518,
+      "step": 3439
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.223058758681178e-06,
+      "loss": 1.2338,
+      "step": 3440
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.217639750655595e-06,
+      "loss": 1.2445,
+      "step": 3441
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.212223292388098e-06,
+      "loss": 1.1791,
+      "step": 3442
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.20680938626711e-06,
+      "loss": 1.3444,
+      "step": 3443
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.201398034679911e-06,
+      "loss": 1.2356,
+      "step": 3444
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.195989240012659e-06,
+      "loss": 1.3411,
+      "step": 3445
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.190583004650393e-06,
+      "loss": 1.2442,
+      "step": 3446
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.185179330977011e-06,
+      "loss": 1.158,
+      "step": 3447
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.179778221375295e-06,
+      "loss": 1.164,
+      "step": 3448
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.174379678226883e-06,
+      "loss": 1.2001,
+      "step": 3449
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.168983703912285e-06,
+      "loss": 1.1506,
+      "step": 3450
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.163590300810888e-06,
+      "loss": 1.2353,
+      "step": 3451
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.15819947130093e-06,
+      "loss": 1.2,
+      "step": 3452
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.152811217759529e-06,
+      "loss": 1.5004,
+      "step": 3453
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.147425542562658e-06,
+      "loss": 1.2721,
+      "step": 3454
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.142042448085148e-06,
+      "loss": 1.246,
+      "step": 3455
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1366619367007086e-06,
+      "loss": 1.2222,
+      "step": 3456
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.131284010781896e-06,
+      "loss": 1.2523,
+      "step": 3457
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.125908672700129e-06,
+      "loss": 1.2296,
+      "step": 3458
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1205359248256946e-06,
+      "loss": 1.1398,
+      "step": 3459
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1151657695277225e-06,
+      "loss": 1.1226,
+      "step": 3460
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1097982091742124e-06,
+      "loss": 1.2321,
+      "step": 3461
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.104433246132015e-06,
+      "loss": 1.1852,
+      "step": 3462
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.099070882766829e-06,
+      "loss": 1.2087,
+      "step": 3463
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0937111214432204e-06,
+      "loss": 1.2413,
+      "step": 3464
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.088353964524593e-06,
+      "loss": 1.2788,
+      "step": 3465
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.082999414373217e-06,
+      "loss": 1.2299,
+      "step": 3466
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.077647473350201e-06,
+      "loss": 1.2241,
+      "step": 3467
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.072298143815508e-06,
+      "loss": 1.1394,
+      "step": 3468
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.066951428127952e-06,
+      "loss": 1.142,
+      "step": 3469
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.061607328645186e-06,
+      "loss": 1.2439,
+      "step": 3470
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0562658477237235e-06,
+      "loss": 1.2062,
+      "step": 3471
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.050926987718911e-06,
+      "loss": 1.3418,
+      "step": 3472
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0455907509849405e-06,
+      "loss": 1.3302,
+      "step": 3473
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.040257139874848e-06,
+      "loss": 1.2466,
+      "step": 3474
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.034926156740518e-06,
+      "loss": 1.0073,
+      "step": 3475
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.029597803932676e-06,
+      "loss": 1.2661,
+      "step": 3476
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.024272083800877e-06,
+      "loss": 1.2419,
+      "step": 3477
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0189489986935225e-06,
+      "loss": 1.2697,
+      "step": 3478
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.01362855095785e-06,
+      "loss": 1.1534,
+      "step": 3479
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.008310742939939e-06,
+      "loss": 1.3174,
+      "step": 3480
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.002995576984696e-06,
+      "loss": 1.1303,
+      "step": 3481
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.997683055435875e-06,
+      "loss": 1.1713,
+      "step": 3482
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.992373180636051e-06,
+      "loss": 1.3027,
+      "step": 3483
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.987065954926636e-06,
+      "loss": 1.1618,
+      "step": 3484
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.981761380647881e-06,
+      "loss": 1.0613,
+      "step": 3485
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9764594601388565e-06,
+      "loss": 1.1989,
+      "step": 3486
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.971160195737474e-06,
+      "loss": 1.3413,
+      "step": 3487
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.965863589780467e-06,
+      "loss": 1.3163,
+      "step": 3488
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.960569644603395e-06,
+      "loss": 1.2634,
+      "step": 3489
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.955278362540647e-06,
+      "loss": 1.3457,
+      "step": 3490
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9499897459254375e-06,
+      "loss": 1.1065,
+      "step": 3491
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.944703797089814e-06,
+      "loss": 1.2765,
+      "step": 3492
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9394205183646325e-06,
+      "loss": 1.1493,
+      "step": 3493
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.934139912079582e-06,
+      "loss": 1.2662,
+      "step": 3494
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.928861980563164e-06,
+      "loss": 1.1034,
+      "step": 3495
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.923586726142711e-06,
+      "loss": 1.243,
+      "step": 3496
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.918314151144372e-06,
+      "loss": 1.26,
+      "step": 3497
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.913044257893114e-06,
+      "loss": 1.1983,
+      "step": 3498
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.907777048712715e-06,
+      "loss": 1.2064,
+      "step": 3499
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.902512525925776e-06,
+      "loss": 1.1064,
+      "step": 3500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.8972506918537125e-06,
+      "loss": 1.1612,
+      "step": 3501
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.8919915488167596e-06,
+      "loss": 1.1069,
+      "step": 3502
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.886735099133956e-06,
+      "loss": 1.2712,
+      "step": 3503
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.881481345123158e-06,
+      "loss": 1.1576,
+      "step": 3504
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.876230289101028e-06,
+      "loss": 1.2892,
+      "step": 3505
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8709819333830455e-06,
+      "loss": 1.1906,
+      "step": 3506
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.865736280283503e-06,
+      "loss": 1.1763,
+      "step": 3507
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.86049333211549e-06,
+      "loss": 1.2539,
+      "step": 3508
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.855253091190909e-06,
+      "loss": 1.1922,
+      "step": 3509
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.850015559820465e-06,
+      "loss": 1.0804,
+      "step": 3510
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8447807403136724e-06,
+      "loss": 1.2242,
+      "step": 3511
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.839548634978856e-06,
+      "loss": 1.1964,
+      "step": 3512
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.834319246123129e-06,
+      "loss": 1.3758,
+      "step": 3513
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.829092576052417e-06,
+      "loss": 1.0985,
+      "step": 3514
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.82386862707144e-06,
+      "loss": 1.1649,
+      "step": 3515
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.818647401483724e-06,
+      "loss": 1.2311,
+      "step": 3516
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.813428901591598e-06,
+      "loss": 1.1609,
+      "step": 3517
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8082131296961775e-06,
+      "loss": 1.2197,
+      "step": 3518
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.803000088097384e-06,
+      "loss": 1.2337,
+      "step": 3519
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.7977897790939255e-06,
+      "loss": 1.269,
+      "step": 3520
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7925822049833194e-06,
+      "loss": 1.3895,
+      "step": 3521
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7873773680618618e-06,
+      "loss": 1.3412,
+      "step": 3522
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7821752706246583e-06,
+      "loss": 1.2858,
+      "step": 3523
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7769759149655914e-06,
+      "loss": 1.163,
+      "step": 3524
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.771779303377342e-06,
+      "loss": 1.2004,
+      "step": 3525
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7665854381513754e-06,
+      "loss": 1.4415,
+      "step": 3526
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7613943215779557e-06,
+      "loss": 1.2107,
+      "step": 3527
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7562059559461293e-06,
+      "loss": 1.2594,
+      "step": 3528
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.75102034354373e-06,
+      "loss": 1.3401,
+      "step": 3529
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7458374866573742e-06,
+      "loss": 1.2787,
+      "step": 3530
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.740657387572464e-06,
+      "loss": 1.1844,
+      "step": 3531
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7354800485731945e-06,
+      "loss": 1.1854,
+      "step": 3532
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7303054719425314e-06,
+      "loss": 1.1882,
+      "step": 3533
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7251336599622323e-06,
+      "loss": 1.172,
+      "step": 3534
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.719964614912829e-06,
+      "loss": 1.2269,
+      "step": 3535
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.714798339073632e-06,
+      "loss": 1.1421,
+      "step": 3536
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7096348347227406e-06,
+      "loss": 1.2576,
+      "step": 3537
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.704474104137019e-06,
+      "loss": 1.2041,
+      "step": 3538
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6993161495921225e-06,
+      "loss": 1.1928,
+      "step": 3539
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6941609733624707e-06,
+      "loss": 1.2278,
+      "step": 3540
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.689008577721258e-06,
+      "loss": 1.1417,
+      "step": 3541
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6838589649404643e-06,
+      "loss": 1.1138,
+      "step": 3542
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6787121372908274e-06,
+      "loss": 1.2804,
+      "step": 3543
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6735680970418706e-06,
+      "loss": 1.1411,
+      "step": 3544
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6684268464618812e-06,
+      "loss": 0.9647,
+      "step": 3545
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6632883878179103e-06,
+      "loss": 1.1565,
+      "step": 3546
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.658152723375794e-06,
+      "loss": 1.2492,
+      "step": 3547
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.653019855400123e-06,
+      "loss": 1.1577,
+      "step": 3548
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.647889786154254e-06,
+      "loss": 1.1762,
+      "step": 3549
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6427625179003223e-06,
+      "loss": 1.0836,
+      "step": 3550
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.637638052899213e-06,
+      "loss": 1.4689,
+      "step": 3551
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.632516393410589e-06,
+      "loss": 1.1998,
+      "step": 3552
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6273975416928675e-06,
+      "loss": 1.2215,
+      "step": 3553
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.622281500003224e-06,
+      "loss": 1.3213,
+      "step": 3554
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6171682705976086e-06,
+      "loss": 1.1684,
+      "step": 3555
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6120578557307152e-06,
+      "loss": 1.1203,
+      "step": 3556
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6069502576560124e-06,
+      "loss": 1.1544,
+      "step": 3557
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.601845478625714e-06,
+      "loss": 1.1092,
+      "step": 3558
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.596743520890794e-06,
+      "loss": 1.1622,
+      "step": 3559
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.59164438670099e-06,
+      "loss": 1.1967,
+      "step": 3560
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5865480783047802e-06,
+      "loss": 1.2456,
+      "step": 3561
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.581454597949414e-06,
+      "loss": 1.2798,
+      "step": 3562
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5763639478808808e-06,
+      "loss": 1.3202,
+      "step": 3563
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5712761303439246e-06,
+      "loss": 1.1393,
+      "step": 3564
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.56619114758204e-06,
+      "loss": 1.1278,
+      "step": 3565
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.561109001837475e-06,
+      "loss": 1.1519,
+      "step": 3566
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5560296953512296e-06,
+      "loss": 1.2924,
+      "step": 3567
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5509532303630446e-06,
+      "loss": 1.2223,
+      "step": 3568
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.545879609111408e-06,
+      "loss": 1.1946,
+      "step": 3569
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5408088338335543e-06,
+      "loss": 1.0989,
+      "step": 3570
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.535740906765468e-06,
+      "loss": 1.2753,
+      "step": 3571
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5306758301418777e-06,
+      "loss": 1.2395,
+      "step": 3572
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.525613606196249e-06,
+      "loss": 1.2116,
+      "step": 3573
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5205542371607924e-06,
+      "loss": 1.14,
+      "step": 3574
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.515497725266458e-06,
+      "loss": 1.183,
+      "step": 3575
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5104440727429378e-06,
+      "loss": 1.5394,
+      "step": 3576
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5053932818186696e-06,
+      "loss": 1.2689,
+      "step": 3577
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5003453547208176e-06,
+      "loss": 1.2941,
+      "step": 3578
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4953002936752876e-06,
+      "loss": 1.297,
+      "step": 3579
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4902581009067237e-06,
+      "loss": 1.1454,
+      "step": 3580
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4852187786384996e-06,
+      "loss": 1.3612,
+      "step": 3581
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.480182329092732e-06,
+      "loss": 1.2643,
+      "step": 3582
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4751487544902673e-06,
+      "loss": 1.2641,
+      "step": 3583
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.470118057050681e-06,
+      "loss": 1.2447,
+      "step": 3584
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4650902389922826e-06,
+      "loss": 1.2951,
+      "step": 3585
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4600653025321085e-06,
+      "loss": 1.177,
+      "step": 3586
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4550432498859287e-06,
+      "loss": 1.1394,
+      "step": 3587
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4500240832682454e-06,
+      "loss": 1.2223,
+      "step": 3588
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4450078048922776e-06,
+      "loss": 1.1811,
+      "step": 3589
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4399944169699783e-06,
+      "loss": 1.1874,
+      "step": 3590
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4349839217120194e-06,
+      "loss": 1.3675,
+      "step": 3591
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.429976321327805e-06,
+      "loss": 1.1687,
+      "step": 3592
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4249716180254622e-06,
+      "loss": 1.1977,
+      "step": 3593
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4199698140118354e-06,
+      "loss": 1.2334,
+      "step": 3594
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.414970911492491e-06,
+      "loss": 1.1296,
+      "step": 3595
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4099749126717175e-06,
+      "loss": 1.176,
+      "step": 3596
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.404981819752524e-06,
+      "loss": 1.2527,
+      "step": 3597
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3999916349366414e-06,
+      "loss": 1.3249,
+      "step": 3598
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.395004360424512e-06,
+      "loss": 1.219,
+      "step": 3599
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.390019998415297e-06,
+      "loss": 1.1724,
+      "step": 3600
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3850385511068697e-06,
+      "loss": 1.2975,
+      "step": 3601
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3800600206958256e-06,
+      "loss": 1.2542,
+      "step": 3602
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3750844093774736e-06,
+      "loss": 1.1724,
+      "step": 3603
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.37011171934583e-06,
+      "loss": 1.2396,
+      "step": 3604
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3651419527936223e-06,
+      "loss": 1.3523,
+      "step": 3605
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.360175111912292e-06,
+      "loss": 1.1118,
+      "step": 3606
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.35521119889199e-06,
+      "loss": 1.5058,
+      "step": 3607
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.350250215921581e-06,
+      "loss": 1.303,
+      "step": 3608
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3452921651886315e-06,
+      "loss": 1.3241,
+      "step": 3609
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.340337048879414e-06,
+      "loss": 1.2616,
+      "step": 3610
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.335384869178908e-06,
+      "loss": 1.2115,
+      "step": 3611
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.330435628270806e-06,
+      "loss": 1.281,
+      "step": 3612
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.325489328337491e-06,
+      "loss": 1.1141,
+      "step": 3613
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.320545971560063e-06,
+      "loss": 1.1712,
+      "step": 3614
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3156055601183156e-06,
+      "loss": 1.1923,
+      "step": 3615
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.310668096190741e-06,
+      "loss": 1.2661,
+      "step": 3616
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3057335819545444e-06,
+      "loss": 1.2637,
+      "step": 3617
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.300802019585615e-06,
+      "loss": 1.2005,
+      "step": 3618
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2958734112585546e-06,
+      "loss": 1.3246,
+      "step": 3619
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.290947759146651e-06,
+      "loss": 1.0924,
+      "step": 3620
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2860250654218916e-06,
+      "loss": 1.2772,
+      "step": 3621
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.281105332254966e-06,
+      "loss": 1.1687,
+      "step": 3622
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.276188561815249e-06,
+      "loss": 1.24,
+      "step": 3623
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2712747562708115e-06,
+      "loss": 1.2751,
+      "step": 3624
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2663639177884234e-06,
+      "loss": 1.1891,
+      "step": 3625
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2614560485335357e-06,
+      "loss": 1.3565,
+      "step": 3626
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2565511506703007e-06,
+      "loss": 1.1288,
+      "step": 3627
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.251649226361555e-06,
+      "loss": 1.2373,
+      "step": 3628
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2467502777688187e-06,
+      "loss": 1.1209,
+      "step": 3629
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2418543070523134e-06,
+      "loss": 1.1509,
+      "step": 3630
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2369613163709356e-06,
+      "loss": 1.3192,
+      "step": 3631
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2320713078822684e-06,
+      "loss": 1.1068,
+      "step": 3632
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2271842837425917e-06,
+      "loss": 1.1852,
+      "step": 3633
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.222300246106852e-06,
+      "loss": 1.3136,
+      "step": 3634
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.217419197128695e-06,
+      "loss": 1.0852,
+      "step": 3635
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2125411389604397e-06,
+      "loss": 1.1246,
+      "step": 3636
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.207666073753084e-06,
+      "loss": 1.2367,
+      "step": 3637
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2027940036563167e-06,
+      "loss": 1.2773,
+      "step": 3638
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1979249308184955e-06,
+      "loss": 1.185,
+      "step": 3639
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1930588573866584e-06,
+      "loss": 1.1388,
+      "step": 3640
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.188195785506527e-06,
+      "loss": 1.0554,
+      "step": 3641
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.18333571732249e-06,
+      "loss": 1.2488,
+      "step": 3642
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.178478654977624e-06,
+      "loss": 1.2584,
+      "step": 3643
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1736246006136663e-06,
+      "loss": 1.2391,
+      "step": 3644
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.168773556371034e-06,
+      "loss": 1.3248,
+      "step": 3645
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.163925524388822e-06,
+      "loss": 1.3782,
+      "step": 3646
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.159080506804787e-06,
+      "loss": 1.2603,
+      "step": 3647
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.154238505755367e-06,
+      "loss": 1.2474,
+      "step": 3648
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.149399523375661e-06,
+      "loss": 1.162,
+      "step": 3649
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1445635617994384e-06,
+      "loss": 1.2591,
+      "step": 3650
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.139730623159144e-06,
+      "loss": 1.2443,
+      "step": 3651
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1349007095858785e-06,
+      "loss": 1.2417,
+      "step": 3652
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1300738232094186e-06,
+      "loss": 1.1635,
+      "step": 3653
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1252499661582016e-06,
+      "loss": 1.1872,
+      "step": 3654
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.120429140559327e-06,
+      "loss": 1.2366,
+      "step": 3655
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.115611348538556e-06,
+      "loss": 1.127,
+      "step": 3656
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.110796592220322e-06,
+      "loss": 1.2247,
+      "step": 3657
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1059848737277145e-06,
+      "loss": 1.3768,
+      "step": 3658
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1011761951824803e-06,
+      "loss": 1.1466,
+      "step": 3659
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0963705587050286e-06,
+      "loss": 1.3207,
+      "step": 3660
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.091567966414424e-06,
+      "loss": 1.2329,
+      "step": 3661
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.086768420428392e-06,
+      "loss": 1.2113,
+      "step": 3662
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.081971922863319e-06,
+      "loss": 1.273,
+      "step": 3663
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0771784758342405e-06,
+      "loss": 1.3853,
+      "step": 3664
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.072388081454848e-06,
+      "loss": 1.2121,
+      "step": 3665
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.067600741837483e-06,
+      "loss": 1.1715,
+      "step": 3666
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.062816459093151e-06,
+      "loss": 1.2095,
+      "step": 3667
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0580352353315022e-06,
+      "loss": 1.2999,
+      "step": 3668
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.05325707266084e-06,
+      "loss": 1.1221,
+      "step": 3669
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0484819731881144e-06,
+      "loss": 1.389,
+      "step": 3670
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.043709939018925e-06,
+      "loss": 1.2171,
+      "step": 3671
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0389409722575258e-06,
+      "loss": 1.2371,
+      "step": 3672
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0341750750068165e-06,
+      "loss": 1.2032,
+      "step": 3673
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0294122493683374e-06,
+      "loss": 1.281,
+      "step": 3674
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0246524974422808e-06,
+      "loss": 1.1743,
+      "step": 3675
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0198958213274753e-06,
+      "loss": 1.2476,
+      "step": 3676
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0151422231214024e-06,
+      "loss": 1.2087,
+      "step": 3677
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.010391704920187e-06,
+      "loss": 1.2551,
+      "step": 3678
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0056442688185885e-06,
+      "loss": 1.2171,
+      "step": 3679
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0008999169100107e-06,
+      "loss": 1.0905,
+      "step": 3680
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9961586512864947e-06,
+      "loss": 1.1597,
+      "step": 3681
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.991420474038721e-06,
+      "loss": 1.1243,
+      "step": 3682
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9866853872560197e-06,
+      "loss": 1.1646,
+      "step": 3683
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9819533930263434e-06,
+      "loss": 1.2303,
+      "step": 3684
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9772244934362883e-06,
+      "loss": 1.1158,
+      "step": 3685
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.972498690571082e-06,
+      "loss": 1.2993,
+      "step": 3686
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.967775986514585e-06,
+      "loss": 1.3,
+      "step": 3687
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.963056383349301e-06,
+      "loss": 1.3714,
+      "step": 3688
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9583398831563628e-06,
+      "loss": 1.277,
+      "step": 3689
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9536264880155286e-06,
+      "loss": 1.2664,
+      "step": 3690
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.948916200005192e-06,
+      "loss": 1.3293,
+      "step": 3691
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.944209021202372e-06,
+      "loss": 1.1972,
+      "step": 3692
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.939504953682726e-06,
+      "loss": 1.1254,
+      "step": 3693
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.934803999520535e-06,
+      "loss": 1.1271,
+      "step": 3694
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.930106160788703e-06,
+      "loss": 1.1394,
+      "step": 3695
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.925411439558764e-06,
+      "loss": 1.2576,
+      "step": 3696
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.920719837900874e-06,
+      "loss": 1.1227,
+      "step": 3697
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9160313578838186e-06,
+      "loss": 1.2085,
+      "step": 3698
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9113460015750073e-06,
+      "loss": 1.2381,
+      "step": 3699
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.906663771040468e-06,
+      "loss": 1.1823,
+      "step": 3700
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.901984668344848e-06,
+      "loss": 1.377,
+      "step": 3701
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8973086955514194e-06,
+      "loss": 1.2109,
+      "step": 3702
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.892635854722079e-06,
+      "loss": 1.052,
+      "step": 3703
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8879661479173306e-06,
+      "loss": 1.2624,
+      "step": 3704
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.883299577196308e-06,
+      "loss": 1.0849,
+      "step": 3705
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8786361446167553e-06,
+      "loss": 1.2515,
+      "step": 3706
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8739758522350305e-06,
+      "loss": 1.2022,
+      "step": 3707
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8693187021061164e-06,
+      "loss": 1.1052,
+      "step": 3708
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8646646962836e-06,
+      "loss": 1.2363,
+      "step": 3709
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8600138368196907e-06,
+      "loss": 1.0747,
+      "step": 3710
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8553661257652045e-06,
+      "loss": 1.2788,
+      "step": 3711
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8507215651695665e-06,
+      "loss": 1.2028,
+      "step": 3712
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.846080157080824e-06,
+      "loss": 1.1405,
+      "step": 3713
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8414419035456197e-06,
+      "loss": 1.1698,
+      "step": 3714
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8368068066092204e-06,
+      "loss": 1.1561,
+      "step": 3715
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8321748683154893e-06,
+      "loss": 1.3022,
+      "step": 3716
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8275460907068974e-06,
+      "loss": 1.307,
+      "step": 3717
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.822920475824531e-06,
+      "loss": 1.0839,
+      "step": 3718
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8182980257080748e-06,
+      "loss": 1.1405,
+      "step": 3719
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8136787423958147e-06,
+      "loss": 1.3006,
+      "step": 3720
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8090626279246504e-06,
+      "loss": 1.1575,
+      "step": 3721
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.804449684330074e-06,
+      "loss": 1.2997,
+      "step": 3722
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7998399136461884e-06,
+      "loss": 1.3315,
+      "step": 3723
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7952333179056912e-06,
+      "loss": 1.1229,
+      "step": 3724
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7906298991398783e-06,
+      "loss": 1.017,
+      "step": 3725
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7860296593786553e-06,
+      "loss": 1.2871,
+      "step": 3726
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.781432600650511e-06,
+      "loss": 1.1813,
+      "step": 3727
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7768387249825456e-06,
+      "loss": 1.2759,
+      "step": 3728
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7722480344004487e-06,
+      "loss": 1.3579,
+      "step": 3729
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.767660530928503e-06,
+      "loss": 1.2796,
+      "step": 3730
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7630762165895887e-06,
+      "loss": 1.1777,
+      "step": 3731
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7584950934051823e-06,
+      "loss": 1.2686,
+      "step": 3732
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.753917163395353e-06,
+      "loss": 1.2304,
+      "step": 3733
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7493424285787584e-06,
+      "loss": 1.3414,
+      "step": 3734
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.744770890972648e-06,
+      "loss": 1.2151,
+      "step": 3735
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7402025525928587e-06,
+      "loss": 1.2852,
+      "step": 3736
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.735637415453826e-06,
+      "loss": 1.2575,
+      "step": 3737
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7310754815685627e-06,
+      "loss": 1.2283,
+      "step": 3738
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.72651675294868e-06,
+      "loss": 1.0637,
+      "step": 3739
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7219612316043673e-06,
+      "loss": 1.2088,
+      "step": 3740
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7174089195443985e-06,
+      "loss": 1.1851,
+      "step": 3741
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7128598187761444e-06,
+      "loss": 1.3074,
+      "step": 3742
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7083139313055428e-06,
+      "loss": 1.3842,
+      "step": 3743
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7037712591371324e-06,
+      "loss": 1.2165,
+      "step": 3744
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6992318042740208e-06,
+      "loss": 1.2105,
+      "step": 3745
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6946955687179e-06,
+      "loss": 1.2731,
+      "step": 3746
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6901625544690436e-06,
+      "loss": 1.2454,
+      "step": 3747
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6856327635263045e-06,
+      "loss": 1.2856,
+      "step": 3748
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6811061978871213e-06,
+      "loss": 1.1341,
+      "step": 3749
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6765828595474985e-06,
+      "loss": 1.3615,
+      "step": 3750
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6720627505020224e-06,
+      "loss": 1.2243,
+      "step": 3751
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.667545872743854e-06,
+      "loss": 1.0995,
+      "step": 3752
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6630322282647336e-06,
+      "loss": 1.212,
+      "step": 3753
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6585218190549768e-06,
+      "loss": 1.3104,
+      "step": 3754
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6540146471034657e-06,
+      "loss": 1.2105,
+      "step": 3755
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6495107143976573e-06,
+      "loss": 1.2489,
+      "step": 3756
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.64501002292358e-06,
+      "loss": 1.2964,
+      "step": 3757
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6405125746658357e-06,
+      "loss": 1.3576,
+      "step": 3758
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6360183716075984e-06,
+      "loss": 1.2139,
+      "step": 3759
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6315274157306038e-06,
+      "loss": 1.3208,
+      "step": 3760
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.62703970901516e-06,
+      "loss": 1.1007,
+      "step": 3761
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.622555253440139e-06,
+      "loss": 1.2319,
+      "step": 3762
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6180740509829848e-06,
+      "loss": 1.2557,
+      "step": 3763
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6135961036197065e-06,
+      "loss": 1.3594,
+      "step": 3764
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.609121413324872e-06,
+      "loss": 1.1137,
+      "step": 3765
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6046499820716178e-06,
+      "loss": 1.0717,
+      "step": 3766
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.600181811831638e-06,
+      "loss": 1.3631,
+      "step": 3767
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5957169045751962e-06,
+      "loss": 1.1137,
+      "step": 3768
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.591255262271115e-06,
+      "loss": 1.2479,
+      "step": 3769
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5867968868867743e-06,
+      "loss": 1.2926,
+      "step": 3770
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5823417803881145e-06,
+      "loss": 1.1995,
+      "step": 3771
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5778899447396333e-06,
+      "loss": 1.2605,
+      "step": 3772
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5734413819043892e-06,
+      "loss": 1.3707,
+      "step": 3773
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5689960938440006e-06,
+      "loss": 1.0726,
+      "step": 3774
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5645540825186333e-06,
+      "loss": 1.2892,
+      "step": 3775
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5601153498870137e-06,
+      "loss": 1.3027,
+      "step": 3776
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.555679897906417e-06,
+      "loss": 1.1605,
+      "step": 3777
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.551247728532682e-06,
+      "loss": 1.1585,
+      "step": 3778
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.546818843720189e-06,
+      "loss": 1.2076,
+      "step": 3779
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5423932454218804e-06,
+      "loss": 1.3447,
+      "step": 3780
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.53797093558924e-06,
+      "loss": 1.3374,
+      "step": 3781
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5335519161723045e-06,
+      "loss": 1.3768,
+      "step": 3782
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5291361891196654e-06,
+      "loss": 1.2307,
+      "step": 3783
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.524723756378452e-06,
+      "loss": 1.3409,
+      "step": 3784
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5203146198943516e-06,
+      "loss": 1.3271,
+      "step": 3785
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.515908781611591e-06,
+      "loss": 1.2722,
+      "step": 3786
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.511506243472944e-06,
+      "loss": 1.2869,
+      "step": 3787
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.507107007419729e-06,
+      "loss": 1.1438,
+      "step": 3788
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5027110753918093e-06,
+      "loss": 1.1925,
+      "step": 3789
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.498318449327595e-06,
+      "loss": 1.318,
+      "step": 3790
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4939291311640324e-06,
+      "loss": 1.1163,
+      "step": 3791
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4895431228366106e-06,
+      "loss": 1.2891,
+      "step": 3792
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.485160426279357e-06,
+      "loss": 1.1239,
+      "step": 3793
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4807810434248493e-06,
+      "loss": 1.2183,
+      "step": 3794
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4764049762041874e-06,
+      "loss": 1.3469,
+      "step": 3795
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4720322265470253e-06,
+      "loss": 1.3629,
+      "step": 3796
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4676627963815445e-06,
+      "loss": 1.1697,
+      "step": 3797
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.463296687634462e-06,
+      "loss": 1.2299,
+      "step": 3798
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4589339022310386e-06,
+      "loss": 1.1699,
+      "step": 3799
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.454574442095059e-06,
+      "loss": 1.2473,
+      "step": 3800
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.450218309148853e-06,
+      "loss": 1.2201,
+      "step": 3801
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.445865505313274e-06,
+      "loss": 1.2451,
+      "step": 3802
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.441516032507708e-06,
+      "loss": 1.1168,
+      "step": 3803
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.437169892650081e-06,
+      "loss": 1.1354,
+      "step": 3804
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.432827087656836e-06,
+      "loss": 1.1609,
+      "step": 3805
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4284876194429595e-06,
+      "loss": 1.2386,
+      "step": 3806
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.424151489921958e-06,
+      "loss": 1.2778,
+      "step": 3807
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4198187010058615e-06,
+      "loss": 1.3069,
+      "step": 3808
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.415489254605242e-06,
+      "loss": 1.2312,
+      "step": 3809
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4111631526291846e-06,
+      "loss": 1.2182,
+      "step": 3810
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.406840396985299e-06,
+      "loss": 1.1166,
+      "step": 3811
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.402520989579733e-06,
+      "loss": 1.2323,
+      "step": 3812
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.398204932317141e-06,
+      "loss": 1.2708,
+      "step": 3813
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3938922271007147e-06,
+      "loss": 1.3251,
+      "step": 3814
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.389582875832157e-06,
+      "loss": 1.1455,
+      "step": 3815
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3852768804116955e-06,
+      "loss": 1.1225,
+      "step": 3816
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.380974242738082e-06,
+      "loss": 1.2234,
+      "step": 3817
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3766749647085785e-06,
+      "loss": 1.3352,
+      "step": 3818
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.372379048218979e-06,
+      "loss": 1.1764,
+      "step": 3819
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.368086495163583e-06,
+      "loss": 1.2814,
+      "step": 3820
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3637973074352117e-06,
+      "loss": 1.1836,
+      "step": 3821
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.359511486925199e-06,
+      "loss": 1.347,
+      "step": 3822
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3552290355233998e-06,
+      "loss": 1.1812,
+      "step": 3823
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3509499551181826e-06,
+      "loss": 1.1885,
+      "step": 3824
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3466742475964266e-06,
+      "loss": 1.2147,
+      "step": 3825
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.342401914843523e-06,
+      "loss": 1.321,
+      "step": 3826
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.338132958743373e-06,
+      "loss": 1.2628,
+      "step": 3827
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3338673811783974e-06,
+      "loss": 1.4171,
+      "step": 3828
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.329605184029523e-06,
+      "loss": 1.1771,
+      "step": 3829
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3253463691761824e-06,
+      "loss": 1.3103,
+      "step": 3830
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.32109093849632e-06,
+      "loss": 1.0133,
+      "step": 3831
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3168388938663834e-06,
+      "loss": 1.377,
+      "step": 3832
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.312590237161335e-06,
+      "loss": 1.1947,
+      "step": 3833
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3083449702546425e-06,
+      "loss": 1.2052,
+      "step": 3834
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.304103095018271e-06,
+      "loss": 1.258,
+      "step": 3835
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2998646133226963e-06,
+      "loss": 1.0759,
+      "step": 3836
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2956295270368965e-06,
+      "loss": 1.161,
+      "step": 3837
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2913978380283453e-06,
+      "loss": 1.2088,
+      "step": 3838
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2871695481630373e-06,
+      "loss": 1.0782,
+      "step": 3839
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2829446593054495e-06,
+      "loss": 1.1049,
+      "step": 3840
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.278723173318568e-06,
+      "loss": 1.2552,
+      "step": 3841
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2745050920638733e-06,
+      "loss": 1.2839,
+      "step": 3842
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2702904174013475e-06,
+      "loss": 1.2531,
+      "step": 3843
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.266079151189472e-06,
+      "loss": 1.1376,
+      "step": 3844
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2618712952852274e-06,
+      "loss": 1.0697,
+      "step": 3845
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2576668515440824e-06,
+      "loss": 1.3459,
+      "step": 3846
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.253465821820008e-06,
+      "loss": 1.2152,
+      "step": 3847
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2492682079654636e-06,
+      "loss": 1.1038,
+      "step": 3848
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2450740118314083e-06,
+      "loss": 1.2054,
+      "step": 3849
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.240883235267295e-06,
+      "loss": 1.3403,
+      "step": 3850
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2366958801210634e-06,
+      "loss": 1.2719,
+      "step": 3851
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2325119482391466e-06,
+      "loss": 1.262,
+      "step": 3852
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.228331441466466e-06,
+      "loss": 1.2473,
+      "step": 3853
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.224154361646438e-06,
+      "loss": 1.1779,
+      "step": 3854
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.219980710620967e-06,
+      "loss": 1.3041,
+      "step": 3855
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.21581049023044e-06,
+      "loss": 1.2164,
+      "step": 3856
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.211643702313736e-06,
+      "loss": 1.187,
+      "step": 3857
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2074803487082164e-06,
+      "loss": 1.206,
+      "step": 3858
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2033204312497335e-06,
+      "loss": 1.2576,
+      "step": 3859
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.199163951772624e-06,
+      "loss": 1.323,
+      "step": 3860
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1950109121097043e-06,
+      "loss": 1.342,
+      "step": 3861
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1908613140922763e-06,
+      "loss": 1.2161,
+      "step": 3862
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.18671515955012e-06,
+      "loss": 1.2761,
+      "step": 3863
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1825724503115064e-06,
+      "loss": 1.2253,
+      "step": 3864
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.178433188203182e-06,
+      "loss": 1.233,
+      "step": 3865
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1742973750503726e-06,
+      "loss": 1.1603,
+      "step": 3866
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1701650126767824e-06,
+      "loss": 1.3724,
+      "step": 3867
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.166036102904594e-06,
+      "loss": 1.158,
+      "step": 3868
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.161910647554474e-06,
+      "loss": 1.1512,
+      "step": 3869
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1577886484455535e-06,
+      "loss": 1.394,
+      "step": 3870
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.153670107395456e-06,
+      "loss": 1.2222,
+      "step": 3871
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.149555026220265e-06,
+      "loss": 1.2518,
+      "step": 3872
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.145443406734542e-06,
+      "loss": 1.2362,
+      "step": 3873
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.141335250751331e-06,
+      "loss": 1.2232,
+      "step": 3874
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1372305600821355e-06,
+      "loss": 1.1569,
+      "step": 3875
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1331293365369444e-06,
+      "loss": 1.221,
+      "step": 3876
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.129031581924207e-06,
+      "loss": 1.1838,
+      "step": 3877
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1249372980508433e-06,
+      "loss": 1.238,
+      "step": 3878
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1208464867222544e-06,
+      "loss": 1.2138,
+      "step": 3879
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1167591497422945e-06,
+      "loss": 1.1423,
+      "step": 3880
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.112675288913301e-06,
+      "loss": 1.319,
+      "step": 3881
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1085949060360654e-06,
+      "loss": 1.1726,
+      "step": 3882
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1045180029098512e-06,
+      "loss": 1.2478,
+      "step": 3883
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.10044458133239e-06,
+      "loss": 1.0603,
+      "step": 3884
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0963746430998753e-06,
+      "loss": 1.2587,
+      "step": 3885
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.092308190006962e-06,
+      "loss": 1.3159,
+      "step": 3886
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0882452238467755e-06,
+      "loss": 1.11,
+      "step": 3887
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.084185746410894e-06,
+      "loss": 1.2735,
+      "step": 3888
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.080129759489369e-06,
+      "loss": 1.2846,
+      "step": 3889
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0760772648707016e-06,
+      "loss": 1.2542,
+      "step": 3890
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0720282643418577e-06,
+      "loss": 1.0768,
+      "step": 3891
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.067982759688266e-06,
+      "loss": 1.1675,
+      "step": 3892
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0639407526938084e-06,
+      "loss": 1.1401,
+      "step": 3893
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0599022451408226e-06,
+      "loss": 1.3439,
+      "step": 3894
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.055867238810113e-06,
+      "loss": 1.2074,
+      "step": 3895
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0518357354809293e-06,
+      "loss": 1.1411,
+      "step": 3896
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0478077369309856e-06,
+      "loss": 1.2041,
+      "step": 3897
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.043783244936445e-06,
+      "loss": 1.1994,
+      "step": 3898
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.03976226127192e-06,
+      "loss": 1.1594,
+      "step": 3899
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0357447877104897e-06,
+      "loss": 1.1866,
+      "step": 3900
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0317308260236735e-06,
+      "loss": 1.1675,
+      "step": 3901
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0277203779814446e-06,
+      "loss": 1.232,
+      "step": 3902
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.023713445352232e-06,
+      "loss": 1.1535,
+      "step": 3903
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.019710029902906e-06,
+      "loss": 1.3042,
+      "step": 3904
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0157101333987973e-06,
+      "loss": 1.2416,
+      "step": 3905
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.011713757603675e-06,
+      "loss": 1.216,
+      "step": 3906
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0077209042797562e-06,
+      "loss": 1.3168,
+      "step": 3907
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.003731575187714e-06,
+      "loss": 1.2159,
+      "step": 3908
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9997457720866554e-06,
+      "loss": 1.4784,
+      "step": 3909
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.995763496734143e-06,
+      "loss": 1.1362,
+      "step": 3910
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.991784750886178e-06,
+      "loss": 1.3416,
+      "step": 3911
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9878095362972037e-06,
+      "loss": 1.1466,
+      "step": 3912
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.983837854720113e-06,
+      "loss": 1.2118,
+      "step": 3913
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9798697079062336e-06,
+      "loss": 1.3506,
+      "step": 3914
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9759050976053407e-06,
+      "loss": 1.2316,
+      "step": 3915
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9719440255656474e-06,
+      "loss": 1.2355,
+      "step": 3916
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9679864935338045e-06,
+      "loss": 1.3258,
+      "step": 3917
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9640325032549023e-06,
+      "loss": 1.1559,
+      "step": 3918
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.960082056472473e-06,
+      "loss": 1.1225,
+      "step": 3919
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9561351549284866e-06,
+      "loss": 1.2102,
+      "step": 3920
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9521918003633446e-06,
+      "loss": 1.276,
+      "step": 3921
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9482519945158873e-06,
+      "loss": 1.2378,
+      "step": 3922
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.944315739123388e-06,
+      "loss": 1.1416,
+      "step": 3923
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.940383035921558e-06,
+      "loss": 1.1954,
+      "step": 3924
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9364538866445436e-06,
+      "loss": 1.1212,
+      "step": 3925
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.93252829302492e-06,
+      "loss": 1.1802,
+      "step": 3926
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9286062567936935e-06,
+      "loss": 1.1042,
+      "step": 3927
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.924687779680302e-06,
+      "loss": 1.3032,
+      "step": 3928
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9207728634126188e-06,
+      "loss": 1.2361,
+      "step": 3929
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.916861509716945e-06,
+      "loss": 1.3148,
+      "step": 3930
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9129537203180105e-06,
+      "loss": 1.1956,
+      "step": 3931
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9090494969389693e-06,
+      "loss": 1.0993,
+      "step": 3932
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9051488413014063e-06,
+      "loss": 1.2864,
+      "step": 3933
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9012517551253352e-06,
+      "loss": 1.1131,
+      "step": 3934
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8973582401291968e-06,
+      "loss": 1.2017,
+      "step": 3935
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8934682980298502e-06,
+      "loss": 1.0996,
+      "step": 3936
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8895819305425856e-06,
+      "loss": 1.111,
+      "step": 3937
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8856991393811097e-06,
+      "loss": 1.25,
+      "step": 3938
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8818199262575598e-06,
+      "loss": 1.2142,
+      "step": 3939
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8779442928824965e-06,
+      "loss": 1.0934,
+      "step": 3940
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.874072240964895e-06,
+      "loss": 1.2867,
+      "step": 3941
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8702037722121523e-06,
+      "loss": 1.1584,
+      "step": 3942
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8663388883300903e-06,
+      "loss": 1.238,
+      "step": 3943
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.862477591022942e-06,
+      "loss": 1.2641,
+      "step": 3944
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8586198819933688e-06,
+      "loss": 1.1755,
+      "step": 3945
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8547657629424454e-06,
+      "loss": 1.1908,
+      "step": 3946
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8509152355696625e-06,
+      "loss": 1.2367,
+      "step": 3947
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8470683015729273e-06,
+      "loss": 1.2604,
+      "step": 3948
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.84322496264856e-06,
+      "loss": 1.1744,
+      "step": 3949
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8393852204913e-06,
+      "loss": 1.2932,
+      "step": 3950
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.835549076794303e-06,
+      "loss": 1.1765,
+      "step": 3951
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8317165332491305e-06,
+      "loss": 1.1792,
+      "step": 3952
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8278875915457616e-06,
+      "loss": 1.2067,
+      "step": 3953
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8240622533725816e-06,
+      "loss": 1.1146,
+      "step": 3954
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8202405204163942e-06,
+      "loss": 1.2964,
+      "step": 3955
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8164223943624115e-06,
+      "loss": 1.2314,
+      "step": 3956
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8126078768942512e-06,
+      "loss": 1.1817,
+      "step": 3957
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8087969696939435e-06,
+      "loss": 1.2862,
+      "step": 3958
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8049896744419215e-06,
+      "loss": 1.3034,
+      "step": 3959
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8011859928170338e-06,
+      "loss": 1.407,
+      "step": 3960
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7973859264965287e-06,
+      "loss": 1.2059,
+      "step": 3961
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7935894771560647e-06,
+      "loss": 1.3943,
+      "step": 3962
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7897966464697036e-06,
+      "loss": 1.1664,
+      "step": 3963
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7860074361099068e-06,
+      "loss": 1.1985,
+      "step": 3964
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7822218477475496e-06,
+      "loss": 1.1689,
+      "step": 3965
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7784398830519002e-06,
+      "loss": 1.1241,
+      "step": 3966
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7746615436906366e-06,
+      "loss": 0.9783,
+      "step": 3967
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7708868313298334e-06,
+      "loss": 1.1788,
+      "step": 3968
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.767115747633965e-06,
+      "loss": 1.1885,
+      "step": 3969
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.763348294265912e-06,
+      "loss": 1.3159,
+      "step": 3970
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.759584472886945e-06,
+      "loss": 1.0892,
+      "step": 3971
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7558242851567442e-06,
+      "loss": 1.3464,
+      "step": 3972
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.752067732733378e-06,
+      "loss": 1.2881,
+      "step": 3973
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7483148172733144e-06,
+      "loss": 1.267,
+      "step": 3974
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.744565540431421e-06,
+      "loss": 1.1785,
+      "step": 3975
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7408199038609586e-06,
+      "loss": 1.1149,
+      "step": 3976
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7370779092135792e-06,
+      "loss": 1.3408,
+      "step": 3977
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7333395581393364e-06,
+      "loss": 1.1535,
+      "step": 3978
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7296048522866694e-06,
+      "loss": 1.2721,
+      "step": 3979
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7258737933024183e-06,
+      "loss": 1.2407,
+      "step": 3980
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7221463828318074e-06,
+      "loss": 1.2831,
+      "step": 3981
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7184226225184554e-06,
+      "loss": 1.1569,
+      "step": 3982
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7147025140043727e-06,
+      "loss": 1.2004,
+      "step": 3983
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7109860589299554e-06,
+      "loss": 1.2318,
+      "step": 3984
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7072732589339958e-06,
+      "loss": 1.2525,
+      "step": 3985
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7035641156536674e-06,
+      "loss": 1.2754,
+      "step": 3986
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6998586307245313e-06,
+      "loss": 1.2093,
+      "step": 3987
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6961568057805433e-06,
+      "loss": 1.3082,
+      "step": 3988
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6924586424540346e-06,
+      "loss": 1.3604,
+      "step": 3989
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6887641423757328e-06,
+      "loss": 1.2215,
+      "step": 3990
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6850733071747406e-06,
+      "loss": 1.1597,
+      "step": 3991
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6813861384785501e-06,
+      "loss": 1.2608,
+      "step": 3992
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6777026379130323e-06,
+      "loss": 1.3145,
+      "step": 3993
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6740228071024455e-06,
+      "loss": 1.3047,
+      "step": 3994
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6703466476694307e-06,
+      "loss": 1.2314,
+      "step": 3995
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6666741612350034e-06,
+      "loss": 1.2718,
+      "step": 3996
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.663005349418566e-06,
+      "loss": 1.2513,
+      "step": 3997
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6593402138378933e-06,
+      "loss": 1.0002,
+      "step": 3998
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6556787561091492e-06,
+      "loss": 1.2209,
+      "step": 3999
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6520209778468644e-06,
+      "loss": 1.2421,
+      "step": 4000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6483668806639585e-06,
+      "loss": 1.2191,
+      "step": 4001
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6447164661717197e-06,
+      "loss": 1.193,
+      "step": 4002
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6410697359798122e-06,
+      "loss": 1.4049,
+      "step": 4003
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6374266916962832e-06,
+      "loss": 1.1188,
+      "step": 4004
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6337873349275457e-06,
+      "loss": 1.2376,
+      "step": 4005
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6301516672783945e-06,
+      "loss": 1.1039,
+      "step": 4006
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.626519690351992e-06,
+      "loss": 1.341,
+      "step": 4007
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6228914057498747e-06,
+      "loss": 1.2729,
+      "step": 4008
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6192668150719482e-06,
+      "loss": 1.0266,
+      "step": 4009
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6156459199164954e-06,
+      "loss": 1.2703,
+      "step": 4010
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.612028721880169e-06,
+      "loss": 1.2438,
+      "step": 4011
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6084152225579864e-06,
+      "loss": 1.1764,
+      "step": 4012
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6048054235433364e-06,
+      "loss": 1.3392,
+      "step": 4013
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6011993264279735e-06,
+      "loss": 1.1308,
+      "step": 4014
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5975969328020257e-06,
+      "loss": 1.2734,
+      "step": 4015
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.593998244253988e-06,
+      "loss": 1.1418,
+      "step": 4016
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5904032623707144e-06,
+      "loss": 1.2096,
+      "step": 4017
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5868119887374301e-06,
+      "loss": 1.348,
+      "step": 4018
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5832244249377203e-06,
+      "loss": 1.3291,
+      "step": 4019
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5796405725535401e-06,
+      "loss": 1.2908,
+      "step": 4020
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5760604331652074e-06,
+      "loss": 1.1897,
+      "step": 4021
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5724840083514003e-06,
+      "loss": 1.1753,
+      "step": 4022
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5689112996891576e-06,
+      "loss": 1.1948,
+      "step": 4023
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5653423087538799e-06,
+      "loss": 1.2106,
+      "step": 4024
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.561777037119333e-06,
+      "loss": 1.3735,
+      "step": 4025
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5582154863576415e-06,
+      "loss": 1.2625,
+      "step": 4026
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5546576580392848e-06,
+      "loss": 1.3194,
+      "step": 4027
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5511035537331042e-06,
+      "loss": 1.1796,
+      "step": 4028
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5475531750062955e-06,
+      "loss": 1.2238,
+      "step": 4029
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5440065234244162e-06,
+      "loss": 1.2462,
+      "step": 4030
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5404636005513808e-06,
+      "loss": 1.053,
+      "step": 4031
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5369244079494561e-06,
+      "loss": 1.2059,
+      "step": 4032
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.533388947179264e-06,
+      "loss": 1.1669,
+      "step": 4033
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5298572197997796e-06,
+      "loss": 1.2443,
+      "step": 4034
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5263292273683405e-06,
+      "loss": 1.1908,
+      "step": 4035
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5228049714406246e-06,
+      "loss": 1.3673,
+      "step": 4036
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.519284453570674e-06,
+      "loss": 1.3105,
+      "step": 4037
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5157676753108752e-06,
+      "loss": 1.1421,
+      "step": 4038
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.512254638211964e-06,
+      "loss": 1.2746,
+      "step": 4039
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.508745343823037e-06,
+      "loss": 1.1955,
+      "step": 4040
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5052397936915264e-06,
+      "loss": 1.1533,
+      "step": 4041
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5017379893632255e-06,
+      "loss": 1.2364,
+      "step": 4042
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4982399323822705e-06,
+      "loss": 1.2229,
+      "step": 4043
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4947456242911407e-06,
+      "loss": 1.2243,
+      "step": 4044
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4912550666306746e-06,
+      "loss": 1.2135,
+      "step": 4045
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4877682609400423e-06,
+      "loss": 1.1704,
+      "step": 4046
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4842852087567727e-06,
+      "loss": 1.338,
+      "step": 4047
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4808059116167306e-06,
+      "loss": 1.2297,
+      "step": 4048
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4773303710541275e-06,
+      "loss": 1.1436,
+      "step": 4049
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4738585886015178e-06,
+      "loss": 1.2984,
+      "step": 4050
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4703905657898043e-06,
+      "loss": 1.3043,
+      "step": 4051
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4669263041482219e-06,
+      "loss": 1.2508,
+      "step": 4052
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4634658052043581e-06,
+      "loss": 1.1396,
+      "step": 4053
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.460009070484132e-06,
+      "loss": 1.284,
+      "step": 4054
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4565561015118057e-06,
+      "loss": 1.1728,
+      "step": 4055
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.453106899809985e-06,
+      "loss": 1.203,
+      "step": 4056
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.449661466899608e-06,
+      "loss": 1.2632,
+      "step": 4057
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4462198042999565e-06,
+      "loss": 1.192,
+      "step": 4058
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.442781913528647e-06,
+      "loss": 1.2472,
+      "step": 4059
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.43934779610163e-06,
+      "loss": 1.1861,
+      "step": 4060
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4359174535331998e-06,
+      "loss": 1.2404,
+      "step": 4061
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4324908873359766e-06,
+      "loss": 1.2171,
+      "step": 4062
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.429068099020926e-06,
+      "loss": 1.209,
+      "step": 4063
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4256490900973386e-06,
+      "loss": 1.2464,
+      "step": 4064
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4222338620728405e-06,
+      "loss": 1.2334,
+      "step": 4065
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4188224164533971e-06,
+      "loss": 1.273,
+      "step": 4066
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4154147547432972e-06,
+      "loss": 1.1845,
+      "step": 4067
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4120108784451625e-06,
+      "loss": 1.2495,
+      "step": 4068
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4086107890599543e-06,
+      "loss": 1.2217,
+      "step": 4069
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.405214488086951e-06,
+      "loss": 1.2695,
+      "step": 4070
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4018219770237718e-06,
+      "loss": 1.3278,
+      "step": 4071
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3984332573663583e-06,
+      "loss": 1.1544,
+      "step": 4072
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3950483306089802e-06,
+      "loss": 1.2568,
+      "step": 4073
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3916671982442387e-06,
+      "loss": 1.1553,
+      "step": 4074
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.388289861763057e-06,
+      "loss": 1.3899,
+      "step": 4075
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3849163226546902e-06,
+      "loss": 1.2162,
+      "step": 4076
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3815465824067154e-06,
+      "loss": 1.2048,
+      "step": 4077
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3781806425050303e-06,
+      "loss": 1.1687,
+      "step": 4078
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.374818504433867e-06,
+      "loss": 1.0686,
+      "step": 4079
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3714601696757713e-06,
+      "loss": 1.3296,
+      "step": 4080
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.36810563971162e-06,
+      "loss": 1.1662,
+      "step": 4081
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3647549160206074e-06,
+      "loss": 0.9286,
+      "step": 4082
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3614080000802498e-06,
+      "loss": 1.1799,
+      "step": 4083
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.358064893366382e-06,
+      "loss": 1.2605,
+      "step": 4084
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3547255973531648e-06,
+      "loss": 1.1539,
+      "step": 4085
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.351390113513078e-06,
+      "loss": 1.171,
+      "step": 4086
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3480584433169176e-06,
+      "loss": 1.2427,
+      "step": 4087
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3447305882337968e-06,
+      "loss": 1.1881,
+      "step": 4088
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.341406549731148e-06,
+      "loss": 1.2282,
+      "step": 4089
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3380863292747214e-06,
+      "loss": 1.2963,
+      "step": 4090
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3347699283285875e-06,
+      "loss": 1.204,
+      "step": 4091
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3314573483551253e-06,
+      "loss": 1.1386,
+      "step": 4092
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3281485908150315e-06,
+      "loss": 1.1519,
+      "step": 4093
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3248436571673162e-06,
+      "loss": 1.1634,
+      "step": 4094
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.321542548869308e-06,
+      "loss": 1.254,
+      "step": 4095
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3182452673766454e-06,
+      "loss": 1.1148,
+      "step": 4096
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3149518141432805e-06,
+      "loss": 1.2214,
+      "step": 4097
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3116621906214744e-06,
+      "loss": 1.2253,
+      "step": 4098
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3083763982618026e-06,
+      "loss": 1.0011,
+      "step": 4099
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3050944385131448e-06,
+      "loss": 1.2438,
+      "step": 4100
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3018163128227058e-06,
+      "loss": 1.2636,
+      "step": 4101
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2985420226359846e-06,
+      "loss": 1.2989,
+      "step": 4102
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2952715693967966e-06,
+      "loss": 1.3581,
+      "step": 4103
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2920049545472602e-06,
+      "loss": 1.1186,
+      "step": 4104
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2887421795278043e-06,
+      "loss": 1.2546,
+      "step": 4105
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.285483245777165e-06,
+      "loss": 1.1042,
+      "step": 4106
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2822281547323867e-06,
+      "loss": 1.2266,
+      "step": 4107
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2789769078288151e-06,
+      "loss": 1.1848,
+      "step": 4108
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2757295065001008e-06,
+      "loss": 1.4246,
+      "step": 4109
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2724859521781996e-06,
+      "loss": 1.2078,
+      "step": 4110
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2692462462933742e-06,
+      "loss": 1.1866,
+      "step": 4111
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2660103902741871e-06,
+      "loss": 1.2565,
+      "step": 4112
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.262778385547504e-06,
+      "loss": 1.0842,
+      "step": 4113
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2595502335384912e-06,
+      "loss": 1.1811,
+      "step": 4114
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2563259356706147e-06,
+      "loss": 1.2173,
+      "step": 4115
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2531054933656462e-06,
+      "loss": 1.2582,
+      "step": 4116
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.249888908043655e-06,
+      "loss": 1.2953,
+      "step": 4117
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2466761811230099e-06,
+      "loss": 1.3379,
+      "step": 4118
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2434673140203746e-06,
+      "loss": 1.3141,
+      "step": 4119
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2402623081507126e-06,
+      "loss": 1.2557,
+      "step": 4120
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.237061164927288e-06,
+      "loss": 1.2217,
+      "step": 4121
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2338638857616615e-06,
+      "loss": 1.2423,
+      "step": 4122
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2306704720636852e-06,
+      "loss": 1.2729,
+      "step": 4123
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.227480925241511e-06,
+      "loss": 1.2012,
+      "step": 4124
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.22429524670158e-06,
+      "loss": 1.262,
+      "step": 4125
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2211134378486378e-06,
+      "loss": 1.1684,
+      "step": 4126
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2179355000857117e-06,
+      "loss": 1.1041,
+      "step": 4127
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2147614348141334e-06,
+      "loss": 1.1938,
+      "step": 4128
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2115912434335187e-06,
+      "loss": 1.2139,
+      "step": 4129
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.208424927341776e-06,
+      "loss": 1.3566,
+      "step": 4130
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2052624879351105e-06,
+      "loss": 1.1915,
+      "step": 4131
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2021039266080104e-06,
+      "loss": 1.1671,
+      "step": 4132
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1989492447532613e-06,
+      "loss": 1.343,
+      "step": 4133
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.195798443761933e-06,
+      "loss": 1.3895,
+      "step": 4134
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1926515250233839e-06,
+      "loss": 1.2022,
+      "step": 4135
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1895084899252662e-06,
+      "loss": 1.1501,
+      "step": 4136
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1863693398535115e-06,
+      "loss": 1.2591,
+      "step": 4137
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1832340761923444e-06,
+      "loss": 1.2932,
+      "step": 4138
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.180102700324275e-06,
+      "loss": 1.1123,
+      "step": 4139
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1769752136300927e-06,
+      "loss": 1.1369,
+      "step": 4140
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1738516174888837e-06,
+      "loss": 1.3077,
+      "step": 4141
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1707319132780071e-06,
+      "loss": 1.1459,
+      "step": 4142
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1676161023731115e-06,
+      "loss": 1.1665,
+      "step": 4143
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1645041861481287e-06,
+      "loss": 1.2221,
+      "step": 4144
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1613961659752716e-06,
+      "loss": 1.1469,
+      "step": 4145
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.158292043225039e-06,
+      "loss": 1.1385,
+      "step": 4146
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.155191819266205e-06,
+      "loss": 1.234,
+      "step": 4147
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1520954954658247e-06,
+      "loss": 1.0665,
+      "step": 4148
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1490030731892422e-06,
+      "loss": 1.222,
+      "step": 4149
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1459145538000703e-06,
+      "loss": 1.2669,
+      "step": 4150
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1428299386602104e-06,
+      "loss": 1.3227,
+      "step": 4151
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.139749229129834e-06,
+      "loss": 1.268,
+      "step": 4152
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1366724265673935e-06,
+      "loss": 1.1455,
+      "step": 4153
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1335995323296223e-06,
+      "loss": 1.112,
+      "step": 4154
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1305305477715256e-06,
+      "loss": 1.2653,
+      "step": 4155
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1274654742463842e-06,
+      "loss": 1.186,
+      "step": 4156
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1244043131057591e-06,
+      "loss": 1.1549,
+      "step": 4157
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1213470656994818e-06,
+      "loss": 1.2957,
+      "step": 4158
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1182937333756582e-06,
+      "loss": 1.2423,
+      "step": 4159
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1152443174806726e-06,
+      "loss": 1.1424,
+      "step": 4160
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1121988193591737e-06,
+      "loss": 1.2541,
+      "step": 4161
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.109157240354094e-06,
+      "loss": 1.1576,
+      "step": 4162
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1061195818066283e-06,
+      "loss": 1.2649,
+      "step": 4163
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1030858450562443e-06,
+      "loss": 1.3035,
+      "step": 4164
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.100056031440685e-06,
+      "loss": 1.18,
+      "step": 4165
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0970301422959583e-06,
+      "loss": 1.2145,
+      "step": 4166
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0940081789563462e-06,
+      "loss": 1.212,
+      "step": 4167
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0909901427543968e-06,
+      "loss": 1.2478,
+      "step": 4168
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0879760350209234e-06,
+      "loss": 1.2087,
+      "step": 4169
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0849658570850153e-06,
+      "loss": 1.1506,
+      "step": 4170
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0819596102740194e-06,
+      "loss": 1.1616,
+      "step": 4171
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0789572959135597e-06,
+      "loss": 1.2793,
+      "step": 4172
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0759589153275162e-06,
+      "loss": 1.3768,
+      "step": 4173
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0729644698380403e-06,
+      "loss": 1.1945,
+      "step": 4174
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0699739607655434e-06,
+      "loss": 1.3475,
+      "step": 4175
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0669873894287052e-06,
+      "loss": 1.3182,
+      "step": 4176
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0640047571444724e-06,
+      "loss": 1.2877,
+      "step": 4177
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0610260652280468e-06,
+      "loss": 1.2328,
+      "step": 4178
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0580513149928961e-06,
+      "loss": 1.2223,
+      "step": 4179
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0550805077507475e-06,
+      "loss": 1.1105,
+      "step": 4180
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0521136448115954e-06,
+      "loss": 1.0581,
+      "step": 4181
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0491507274836922e-06,
+      "loss": 1.277,
+      "step": 4182
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0461917570735492e-06,
+      "loss": 1.3633,
+      "step": 4183
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0432367348859362e-06,
+      "loss": 1.3473,
+      "step": 4184
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0402856622238832e-06,
+      "loss": 1.1121,
+      "step": 4185
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0373385403886793e-06,
+      "loss": 1.1363,
+      "step": 4186
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.034395370679876e-06,
+      "loss": 1.2657,
+      "step": 4187
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0314561543952728e-06,
+      "loss": 1.255,
+      "step": 4188
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.028520892830932e-06,
+      "loss": 1.2611,
+      "step": 4189
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0255895872811683e-06,
+      "loss": 1.2381,
+      "step": 4190
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0226622390385553e-06,
+      "loss": 1.2773,
+      "step": 4191
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0197388493939243e-06,
+      "loss": 1.2914,
+      "step": 4192
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0168194196363534e-06,
+      "loss": 1.2656,
+      "step": 4193
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.01390395105318e-06,
+      "loss": 1.2651,
+      "step": 4194
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0109924449299901e-06,
+      "loss": 1.3258,
+      "step": 4195
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0080849025506277e-06,
+      "loss": 1.2586,
+      "step": 4196
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0051813251971897e-06,
+      "loss": 1.241,
+      "step": 4197
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0022817141500196e-06,
+      "loss": 1.2132,
+      "step": 4198
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.993860706877135e-07,
+      "loss": 1.1685,
+      "step": 4199
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.964943960871187e-07,
+      "loss": 1.1644,
+      "step": 4200
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.93606691623329e-07,
+      "loss": 1.2493,
+      "step": 4201
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.907229585696986e-07,
+      "loss": 1.1968,
+      "step": 4202
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.878431981978177e-07,
+      "loss": 1.3696,
+      "step": 4203
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.849674117775298e-07,
+      "loss": 1.2651,
+      "step": 4204
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.820956005769277e-07,
+      "loss": 1.3745,
+      "step": 4205
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.792277658623461e-07,
+      "loss": 1.193,
+      "step": 4206
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.763639088983723e-07,
+      "loss": 1.255,
+      "step": 4207
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.735040309478395e-07,
+      "loss": 1.2581,
+      "step": 4208
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.70648133271821e-07,
+      "loss": 1.2238,
+      "step": 4209
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.677962171296395e-07,
+      "loss": 1.3497,
+      "step": 4210
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.649482837788559e-07,
+      "loss": 1.2769,
+      "step": 4211
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.621043344752834e-07,
+      "loss": 1.2864,
+      "step": 4212
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.592643704729754e-07,
+      "loss": 1.2653,
+      "step": 4213
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.564283930242258e-07,
+      "loss": 1.1906,
+      "step": 4214
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.535964033795709e-07,
+      "loss": 1.2881,
+      "step": 4215
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.507684027877884e-07,
+      "loss": 1.198,
+      "step": 4216
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.479443924959009e-07,
+      "loss": 1.1833,
+      "step": 4217
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.451243737491656e-07,
+      "loss": 1.1798,
+      "step": 4218
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.423083477910855e-07,
+      "loss": 1.2697,
+      "step": 4219
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.394963158633996e-07,
+      "loss": 1.289,
+      "step": 4220
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.366882792060827e-07,
+      "loss": 1.137,
+      "step": 4221
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.338842390573566e-07,
+      "loss": 1.2653,
+      "step": 4222
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.31084196653671e-07,
+      "loss": 1.1495,
+      "step": 4223
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.282881532297205e-07,
+      "loss": 1.1702,
+      "step": 4224
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.254961100184334e-07,
+      "loss": 1.2926,
+      "step": 4225
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.227080682509692e-07,
+      "loss": 1.1945,
+      "step": 4226
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.199240291567335e-07,
+      "loss": 1.2516,
+      "step": 4227
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.171439939633564e-07,
+      "loss": 1.0932,
+      "step": 4228
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.143679638967107e-07,
+      "loss": 1.1226,
+      "step": 4229
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.115959401808983e-07,
+      "loss": 1.1747,
+      "step": 4230
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.088279240382535e-07,
+      "loss": 1.2714,
+      "step": 4231
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.060639166893493e-07,
+      "loss": 1.3042,
+      "step": 4232
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.033039193529858e-07,
+      "loss": 1.3283,
+      "step": 4233
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.00547933246193e-07,
+      "loss": 1.2573,
+      "step": 4234
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.977959595842411e-07,
+      "loss": 1.2337,
+      "step": 4235
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.950479995806216e-07,
+      "loss": 1.2833,
+      "step": 4236
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.923040544470629e-07,
+      "loss": 1.1996,
+      "step": 4237
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.895641253935183e-07,
+      "loss": 1.2929,
+      "step": 4238
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.868282136281703e-07,
+      "loss": 1.1681,
+      "step": 4239
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.840963203574349e-07,
+      "loss": 1.2985,
+      "step": 4240
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.813684467859507e-07,
+      "loss": 1.1863,
+      "step": 4241
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.786445941165878e-07,
+      "loss": 1.108,
+      "step": 4242
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.759247635504409e-07,
+      "loss": 1.1513,
+      "step": 4243
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.732089562868295e-07,
+      "loss": 1.1304,
+      "step": 4244
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.704971735233048e-07,
+      "loss": 1.2881,
+      "step": 4245
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.677894164556356e-07,
+      "loss": 1.2483,
+      "step": 4246
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.650856862778245e-07,
+      "loss": 1.2746,
+      "step": 4247
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.623859841820902e-07,
+      "loss": 1.1269,
+      "step": 4248
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.596903113588806e-07,
+      "loss": 1.0687,
+      "step": 4249
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.569986689968613e-07,
+      "loss": 1.1851,
+      "step": 4250
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.543110582829272e-07,
+      "loss": 1.1631,
+      "step": 4251
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.516274804021929e-07,
+      "loss": 1.2185,
+      "step": 4252
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.489479365379948e-07,
+      "loss": 1.2179,
+      "step": 4253
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.462724278718882e-07,
+      "loss": 1.2437,
+      "step": 4254
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.4360095558365e-07,
+      "loss": 1.2125,
+      "step": 4255
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.409335208512803e-07,
+      "loss": 1.1679,
+      "step": 4256
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.38270124850995e-07,
+      "loss": 1.3194,
+      "step": 4257
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.356107687572323e-07,
+      "loss": 1.351,
+      "step": 4258
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.329554537426465e-07,
+      "loss": 1.133,
+      "step": 4259
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.303041809781087e-07,
+      "loss": 1.308,
+      "step": 4260
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.276569516327149e-07,
+      "loss": 1.0994,
+      "step": 4261
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.250137668737668e-07,
+      "loss": 1.1952,
+      "step": 4262
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.223746278667944e-07,
+      "loss": 1.2225,
+      "step": 4263
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.197395357755355e-07,
+      "loss": 1.17,
+      "step": 4264
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.171084917619454e-07,
+      "loss": 1.3327,
+      "step": 4265
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.144814969861936e-07,
+      "loss": 1.0587,
+      "step": 4266
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.11858552606668e-07,
+      "loss": 1.1654,
+      "step": 4267
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.09239659779969e-07,
+      "loss": 1.2607,
+      "step": 4268
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.066248196609073e-07,
+      "loss": 1.1979,
+      "step": 4269
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.040140334025082e-07,
+      "loss": 1.2475,
+      "step": 4270
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.014073021560087e-07,
+      "loss": 1.254,
+      "step": 4271
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.988046270708615e-07,
+      "loss": 1.2093,
+      "step": 4272
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.962060092947276e-07,
+      "loss": 1.1889,
+      "step": 4273
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.936114499734792e-07,
+      "loss": 1.148,
+      "step": 4274
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.91020950251199e-07,
+      "loss": 1.2301,
+      "step": 4275
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.884345112701764e-07,
+      "loss": 1.1858,
+      "step": 4276
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.858521341709169e-07,
+      "loss": 1.3006,
+      "step": 4277
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.83273820092133e-07,
+      "loss": 1.1801,
+      "step": 4278
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.806995701707421e-07,
+      "loss": 1.1596,
+      "step": 4279
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.781293855418704e-07,
+      "loss": 1.2484,
+      "step": 4280
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.755632673388524e-07,
+      "loss": 1.1977,
+      "step": 4281
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.730012166932299e-07,
+      "loss": 1.2996,
+      "step": 4282
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.704432347347534e-07,
+      "loss": 1.1348,
+      "step": 4283
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.678893225913742e-07,
+      "loss": 1.3058,
+      "step": 4284
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.653394813892523e-07,
+      "loss": 1.2409,
+      "step": 4285
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.627937122527507e-07,
+      "loss": 1.2416,
+      "step": 4286
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.60252016304438e-07,
+      "loss": 1.1436,
+      "step": 4287
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.577143946650889e-07,
+      "loss": 1.2546,
+      "step": 4288
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.551808484536782e-07,
+      "loss": 1.1297,
+      "step": 4289
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.526513787873834e-07,
+      "loss": 1.2093,
+      "step": 4290
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.501259867815847e-07,
+      "loss": 1.3013,
+      "step": 4291
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.476046735498677e-07,
+      "loss": 1.1097,
+      "step": 4292
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.450874402040176e-07,
+      "loss": 1.4562,
+      "step": 4293
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.425742878540177e-07,
+      "loss": 1.26,
+      "step": 4294
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.400652176080558e-07,
+      "loss": 1.2376,
+      "step": 4295
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.375602305725138e-07,
+      "loss": 1.1925,
+      "step": 4296
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.350593278519824e-07,
+      "loss": 1.3567,
+      "step": 4297
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.325625105492429e-07,
+      "loss": 1.2075,
+      "step": 4298
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.3006977976528e-07,
+      "loss": 1.143,
+      "step": 4299
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.275811365992736e-07,
+      "loss": 1.1159,
+      "step": 4300
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.250965821486012e-07,
+      "loss": 1.2067,
+      "step": 4301
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.22616117508842e-07,
+      "loss": 1.3593,
+      "step": 4302
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.201397437737634e-07,
+      "loss": 1.2739,
+      "step": 4303
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.176674620353374e-07,
+      "loss": 1.2932,
+      "step": 4304
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.151992733837276e-07,
+      "loss": 1.2086,
+      "step": 4305
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.127351789072912e-07,
+      "loss": 1.2157,
+      "step": 4306
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.102751796925811e-07,
+      "loss": 1.1026,
+      "step": 4307
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.078192768243486e-07,
+      "loss": 1.248,
+      "step": 4308
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.053674713855318e-07,
+      "loss": 1.3275,
+      "step": 4309
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.029197644572693e-07,
+      "loss": 1.2173,
+      "step": 4310
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.004761571188856e-07,
+      "loss": 1.2338,
+      "step": 4311
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.980366504479009e-07,
+      "loss": 1.3105,
+      "step": 4312
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.956012455200279e-07,
+      "loss": 1.1929,
+      "step": 4313
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.931699434091676e-07,
+      "loss": 1.3267,
+      "step": 4314
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.907427451874161e-07,
+      "loss": 1.1624,
+      "step": 4315
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.883196519250568e-07,
+      "loss": 1.2497,
+      "step": 4316
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.85900664690562e-07,
+      "loss": 1.2322,
+      "step": 4317
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.834857845505971e-07,
+      "loss": 1.1819,
+      "step": 4318
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.810750125700127e-07,
+      "loss": 1.1953,
+      "step": 4319
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.786683498118518e-07,
+      "loss": 1.2915,
+      "step": 4320
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.762657973373432e-07,
+      "loss": 1.2602,
+      "step": 4321
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.738673562059006e-07,
+      "loss": 1.2459,
+      "step": 4322
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.714730274751303e-07,
+      "loss": 1.2242,
+      "step": 4323
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.69082812200823e-07,
+      "loss": 1.2394,
+      "step": 4324
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.666967114369505e-07,
+      "loss": 1.2337,
+      "step": 4325
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.643147262356809e-07,
+      "loss": 1.2436,
+      "step": 4326
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.61936857647355e-07,
+      "loss": 1.0703,
+      "step": 4327
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.59563106720511e-07,
+      "loss": 1.2517,
+      "step": 4328
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.571934745018627e-07,
+      "loss": 1.2823,
+      "step": 4329
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.548279620363074e-07,
+      "loss": 1.3301,
+      "step": 4330
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.52466570366933e-07,
+      "loss": 1.1426,
+      "step": 4331
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.501093005350023e-07,
+      "loss": 1.1952,
+      "step": 4332
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.477561535799682e-07,
+      "loss": 1.2567,
+      "step": 4333
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.454071305394583e-07,
+      "loss": 1.2877,
+      "step": 4334
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.430622324492853e-07,
+      "loss": 1.1422,
+      "step": 4335
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.407214603434442e-07,
+      "loss": 1.2405,
+      "step": 4336
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.383848152541072e-07,
+      "loss": 1.2424,
+      "step": 4337
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.360522982116301e-07,
+      "loss": 1.0968,
+      "step": 4338
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.33723910244548e-07,
+      "loss": 1.1403,
+      "step": 4339
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.313996523795718e-07,
+      "loss": 1.0035,
+      "step": 4340
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.290795256415927e-07,
+      "loss": 1.256,
+      "step": 4341
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.267635310536835e-07,
+      "loss": 1.1906,
+      "step": 4342
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.244516696370928e-07,
+      "loss": 1.1817,
+      "step": 4343
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.221439424112463e-07,
+      "loss": 1.3974,
+      "step": 4344
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.198403503937467e-07,
+      "loss": 1.2303,
+      "step": 4345
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.175408946003703e-07,
+      "loss": 1.3083,
+      "step": 4346
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.152455760450749e-07,
+      "loss": 1.136,
+      "step": 4347
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.129543957399942e-07,
+      "loss": 1.1519,
+      "step": 4348
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.106673546954322e-07,
+      "loss": 1.2564,
+      "step": 4349
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.083844539198691e-07,
+      "loss": 1.2162,
+      "step": 4350
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.061056944199605e-07,
+      "loss": 1.2838,
+      "step": 4351
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.038310772005374e-07,
+      "loss": 1.2366,
+      "step": 4352
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.015606032646026e-07,
+      "loss": 1.2273,
+      "step": 4353
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.992942736133322e-07,
+      "loss": 1.2489,
+      "step": 4354
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.970320892460734e-07,
+      "loss": 1.2786,
+      "step": 4355
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.947740511603461e-07,
+      "loss": 1.0506,
+      "step": 4356
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.925201603518416e-07,
+      "loss": 1.1917,
+      "step": 4357
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.90270417814427e-07,
+      "loss": 1.1334,
+      "step": 4358
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.880248245401354e-07,
+      "loss": 1.1193,
+      "step": 4359
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.857833815191705e-07,
+      "loss": 1.1553,
+      "step": 4360
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.835460897399059e-07,
+      "loss": 1.2709,
+      "step": 4361
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.813129501888859e-07,
+      "loss": 1.213,
+      "step": 4362
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.79083963850825e-07,
+      "loss": 1.2079,
+      "step": 4363
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.768591317086047e-07,
+      "loss": 1.1698,
+      "step": 4364
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.746384547432738e-07,
+      "loss": 1.0772,
+      "step": 4365
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.724219339340509e-07,
+      "loss": 1.222,
+      "step": 4366
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.702095702583188e-07,
+      "loss": 1.1703,
+      "step": 4367
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.680013646916316e-07,
+      "loss": 1.2956,
+      "step": 4368
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.657973182077081e-07,
+      "loss": 1.092,
+      "step": 4369
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.63597431778431e-07,
+      "loss": 1.1933,
+      "step": 4370
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.614017063738519e-07,
+      "loss": 1.2219,
+      "step": 4371
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.592101429621821e-07,
+      "loss": 1.2366,
+      "step": 4372
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.570227425098051e-07,
+      "loss": 1.3167,
+      "step": 4373
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.54839505981265e-07,
+      "loss": 1.177,
+      "step": 4374
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.526604343392694e-07,
+      "loss": 1.2506,
+      "step": 4375
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.504855285446897e-07,
+      "loss": 1.2491,
+      "step": 4376
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.483147895565589e-07,
+      "loss": 1.3917,
+      "step": 4377
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.461482183320754e-07,
+      "loss": 1.2705,
+      "step": 4378
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.439858158266009e-07,
+      "loss": 1.2889,
+      "step": 4379
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.418275829936537e-07,
+      "loss": 1.1526,
+      "step": 4380
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.39673520784918e-07,
+      "loss": 1.2827,
+      "step": 4381
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.375236301502352e-07,
+      "loss": 1.2043,
+      "step": 4382
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.353779120376102e-07,
+      "loss": 1.2481,
+      "step": 4383
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.332363673932106e-07,
+      "loss": 1.1821,
+      "step": 4384
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.310989971613568e-07,
+      "loss": 1.2309,
+      "step": 4385
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.289658022845323e-07,
+      "loss": 1.2131,
+      "step": 4386
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.268367837033783e-07,
+      "loss": 1.3656,
+      "step": 4387
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.247119423566982e-07,
+      "loss": 1.3011,
+      "step": 4388
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.22591279181447e-07,
+      "loss": 1.1583,
+      "step": 4389
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.204747951127442e-07,
+      "loss": 1.0689,
+      "step": 4390
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.183624910838603e-07,
+      "loss": 1.2085,
+      "step": 4391
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.162543680262267e-07,
+      "loss": 1.1637,
+      "step": 4392
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.141504268694297e-07,
+      "loss": 1.3039,
+      "step": 4393
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.120506685412108e-07,
+      "loss": 1.2746,
+      "step": 4394
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.099550939674691e-07,
+      "loss": 1.0417,
+      "step": 4395
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.078637040722589e-07,
+      "loss": 1.0843,
+      "step": 4396
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.057764997777848e-07,
+      "loss": 1.2441,
+      "step": 4397
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.036934820044126e-07,
+      "loss": 1.1727,
+      "step": 4398
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.016146516706566e-07,
+      "loss": 1.3554,
+      "step": 4399
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.995400096931846e-07,
+      "loss": 1.1705,
+      "step": 4400
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.974695569868238e-07,
+      "loss": 1.1172,
+      "step": 4401
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.954032944645459e-07,
+      "loss": 1.1661,
+      "step": 4402
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.933412230374812e-07,
+      "loss": 1.1834,
+      "step": 4403
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.9128334361491e-07,
+      "loss": 1.3453,
+      "step": 4404
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.892296571042598e-07,
+      "loss": 1.24,
+      "step": 4405
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.871801644111174e-07,
+      "loss": 1.2478,
+      "step": 4406
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.85134866439212e-07,
+      "loss": 1.3873,
+      "step": 4407
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.830937640904309e-07,
+      "loss": 1.2922,
+      "step": 4408
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.810568582648056e-07,
+      "loss": 1.1187,
+      "step": 4409
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.790241498605175e-07,
+      "loss": 1.1538,
+      "step": 4410
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.769956397739017e-07,
+      "loss": 1.259,
+      "step": 4411
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.749713288994373e-07,
+      "loss": 1.2602,
+      "step": 4412
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.729512181297524e-07,
+      "loss": 1.1328,
+      "step": 4413
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.709353083556267e-07,
+      "loss": 1.1753,
+      "step": 4414
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.689236004659825e-07,
+      "loss": 1.2814,
+      "step": 4415
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.669160953478913e-07,
+      "loss": 1.1648,
+      "step": 4416
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6491279388657495e-07,
+      "loss": 1.279,
+      "step": 4417
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6291369696539354e-07,
+      "loss": 1.2471,
+      "step": 4418
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6091880546586307e-07,
+      "loss": 1.1247,
+      "step": 4419
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.589281202676366e-07,
+      "loss": 1.2846,
+      "step": 4420
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.569416422485151e-07,
+      "loss": 1.2971,
+      "step": 4421
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.549593722844492e-07,
+      "loss": 1.2065,
+      "step": 4422
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.529813112495252e-07,
+      "loss": 1.2095,
+      "step": 4423
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5100746001598193e-07,
+      "loss": 1.1443,
+      "step": 4424
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4903781945419556e-07,
+      "loss": 1.2278,
+      "step": 4425
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.47072390432689e-07,
+      "loss": 1.2245,
+      "step": 4426
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4511117381812796e-07,
+      "loss": 1.1743,
+      "step": 4427
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.431541704753173e-07,
+      "loss": 1.1865,
+      "step": 4428
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.412013812672089e-07,
+      "loss": 1.1788,
+      "step": 4429
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.3925280705489513e-07,
+      "loss": 1.2546,
+      "step": 4430
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3730844869760535e-07,
+      "loss": 1.174,
+      "step": 4431
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3536830705271484e-07,
+      "loss": 1.1066,
+      "step": 4432
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3343238297573695e-07,
+      "loss": 1.2179,
+      "step": 4433
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3150067732032895e-07,
+      "loss": 1.0744,
+      "step": 4434
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.295731909382828e-07,
+      "loss": 1.1226,
+      "step": 4435
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.276499246795329e-07,
+      "loss": 1.1215,
+      "step": 4436
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2573087939215217e-07,
+      "loss": 1.1834,
+      "step": 4437
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.238160559223514e-07,
+      "loss": 1.3064,
+      "step": 4438
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.219054551144841e-07,
+      "loss": 1.287,
+      "step": 4439
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.199990778110363e-07,
+      "loss": 1.3227,
+      "step": 4440
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.180969248526334e-07,
+      "loss": 1.0862,
+      "step": 4441
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1619899707803666e-07,
+      "loss": 1.1359,
+      "step": 4442
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1430529532414886e-07,
+      "loss": 1.2316,
+      "step": 4443
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.124158204260065e-07,
+      "loss": 1.2354,
+      "step": 4444
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.105305732167819e-07,
+      "loss": 1.2807,
+      "step": 4445
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0864955452778244e-07,
+      "loss": 1.0716,
+      "step": 4446
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.067727651884501e-07,
+      "loss": 1.1735,
+      "step": 4447
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0490020602636627e-07,
+      "loss": 1.2372,
+      "step": 4448
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0303187786724486e-07,
+      "loss": 1.1286,
+      "step": 4449
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0116778153493353e-07,
+      "loss": 1.1521,
+      "step": 4450
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9930791785141254e-07,
+      "loss": 1.212,
+      "step": 4451
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.97452287636797e-07,
+      "loss": 1.1806,
+      "step": 4452
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.956008917093357e-07,
+      "loss": 1.2657,
+      "step": 4453
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9375373088541335e-07,
+      "loss": 1.2375,
+      "step": 4454
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.919108059795407e-07,
+      "loss": 1.197,
+      "step": 4455
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9007211780436536e-07,
+      "loss": 1.2232,
+      "step": 4456
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8823766717066226e-07,
+      "loss": 1.2268,
+      "step": 4457
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.864074548873431e-07,
+      "loss": 1.1154,
+      "step": 4458
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8458148176145016e-07,
+      "loss": 1.258,
+      "step": 4459
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.827597485981527e-07,
+      "loss": 1.1557,
+      "step": 4460
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.809422562007525e-07,
+      "loss": 1.1793,
+      "step": 4461
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7912900537067975e-07,
+      "loss": 1.2139,
+      "step": 4462
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.773199969074959e-07,
+      "loss": 1.217,
+      "step": 4463
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.755152316088928e-07,
+      "loss": 1.1742,
+      "step": 4464
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.737147102706906e-07,
+      "loss": 1.2065,
+      "step": 4465
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7191843368683646e-07,
+      "loss": 1.2739,
+      "step": 4466
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.701264026494067e-07,
+      "loss": 1.3062,
+      "step": 4467
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.683386179486037e-07,
+      "loss": 1.3552,
+      "step": 4468
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.665550803727613e-07,
+      "loss": 1.2072,
+      "step": 4469
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6477579070833934e-07,
+      "loss": 1.1043,
+      "step": 4470
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.630007497399224e-07,
+      "loss": 1.0174,
+      "step": 4471
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6122995825022324e-07,
+      "loss": 1.2867,
+      "step": 4472
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5946341702007836e-07,
+      "loss": 1.0401,
+      "step": 4473
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.577011268284547e-07,
+      "loss": 1.277,
+      "step": 4474
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5594308845244284e-07,
+      "loss": 1.1906,
+      "step": 4475
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5418930266725606e-07,
+      "loss": 1.2126,
+      "step": 4476
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5243977024623455e-07,
+      "loss": 1.0902,
+      "step": 4477
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5069449196084126e-07,
+      "loss": 1.2572,
+      "step": 4478
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4895346858066723e-07,
+      "loss": 1.256,
+      "step": 4479
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4721670087342284e-07,
+      "loss": 1.1718,
+      "step": 4480
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4548418960494436e-07,
+      "loss": 1.2845,
+      "step": 4481
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.437559355391917e-07,
+      "loss": 1.2083,
+      "step": 4482
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.420319394382432e-07,
+      "loss": 1.2847,
+      "step": 4483
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.403122020623062e-07,
+      "loss": 1.0924,
+      "step": 4484
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.385967241697041e-07,
+      "loss": 1.3398,
+      "step": 4485
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3688550651688634e-07,
+      "loss": 1.3034,
+      "step": 4486
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3517854985842146e-07,
+      "loss": 1.276,
+      "step": 4487
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3347585494699966e-07,
+      "loss": 1.2656,
+      "step": 4488
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.317774225334336e-07,
+      "loss": 1.2782,
+      "step": 4489
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.300832533666509e-07,
+      "loss": 1.1972,
+      "step": 4490
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.283933481937085e-07,
+      "loss": 1.1141,
+      "step": 4491
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2670770775977467e-07,
+      "loss": 1.2301,
+      "step": 4492
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2502633280813823e-07,
+      "loss": 1.2437,
+      "step": 4493
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2334922408021385e-07,
+      "loss": 1.2828,
+      "step": 4494
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.216763823155278e-07,
+      "loss": 1.221,
+      "step": 4495
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.200078082517255e-07,
+      "loss": 1.3041,
+      "step": 4496
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1834350262457624e-07,
+      "loss": 1.1572,
+      "step": 4497
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.166834661679596e-07,
+      "loss": 1.1747,
+      "step": 4498
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1502769961387903e-07,
+      "loss": 1.0483,
+      "step": 4499
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1337620369245034e-07,
+      "loss": 1.1335,
+      "step": 4500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.117289791319089e-07,
+      "loss": 1.3539,
+      "step": 4501
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.100860266586059e-07,
+      "loss": 1.234,
+      "step": 4502
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0844734699700727e-07,
+      "loss": 1.0857,
+      "step": 4503
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.068129408696996e-07,
+      "loss": 1.2268,
+      "step": 4504
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.051828089973796e-07,
+      "loss": 1.3655,
+      "step": 4505
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0355695209886125e-07,
+      "loss": 1.2368,
+      "step": 4506
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0193537089107326e-07,
+      "loss": 1.1387,
+      "step": 4507
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.003180660890592e-07,
+      "loss": 1.3714,
+      "step": 4508
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.987050384059798e-07,
+      "loss": 1.2131,
+      "step": 4509
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.970962885531037e-07,
+      "loss": 1.1962,
+      "step": 4510
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9549181723981804e-07,
+      "loss": 1.1428,
+      "step": 4511
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.93891625173619e-07,
+      "loss": 1.3303,
+      "step": 4512
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.922957130601223e-07,
+      "loss": 1.4118,
+      "step": 4513
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.907040816030515e-07,
+      "loss": 1.3205,
+      "step": 4514
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8911673150424314e-07,
+      "loss": 1.1429,
+      "step": 4515
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.875336634636472e-07,
+      "loss": 1.2069,
+      "step": 4516
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8595487817932424e-07,
+      "loss": 1.2385,
+      "step": 4517
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.843803763474462e-07,
+      "loss": 1.1692,
+      "step": 4518
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.828101586622978e-07,
+      "loss": 1.2602,
+      "step": 4519
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8124422581627286e-07,
+      "loss": 1.2604,
+      "step": 4520
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.796825784998791e-07,
+      "loss": 1.1141,
+      "step": 4521
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.78125217401729e-07,
+      "loss": 1.2542,
+      "step": 4522
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7657214320854776e-07,
+      "loss": 1.3114,
+      "step": 4523
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.750233566051719e-07,
+      "loss": 1.1177,
+      "step": 4524
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7347885827454733e-07,
+      "loss": 1.2778,
+      "step": 4525
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.71938648897726e-07,
+      "loss": 1.2212,
+      "step": 4526
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7040272915387024e-07,
+      "loss": 1.1962,
+      "step": 4527
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.688710997202504e-07,
+      "loss": 1.229,
+      "step": 4528
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.673437612722463e-07,
+      "loss": 1.3621,
+      "step": 4529
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6582071448334466e-07,
+      "loss": 1.2007,
+      "step": 4530
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6430196002514065e-07,
+      "loss": 1.2553,
+      "step": 4531
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.627874985673351e-07,
+      "loss": 1.2203,
+      "step": 4532
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.61277330777735e-07,
+      "loss": 1.2892,
+      "step": 4533
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.597714573222576e-07,
+      "loss": 1.2779,
+      "step": 4534
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.582698788649263e-07,
+      "loss": 1.122,
+      "step": 4535
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5677259606786686e-07,
+      "loss": 1.1353,
+      "step": 4536
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.552796095913124e-07,
+      "loss": 1.1793,
+      "step": 4537
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5379092009360286e-07,
+      "loss": 1.2211,
+      "step": 4538
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5230652823118204e-07,
+      "loss": 1.2153,
+      "step": 4539
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.50826434658602e-07,
+      "loss": 1.1915,
+      "step": 4540
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.493506400285139e-07,
+      "loss": 1.3731,
+      "step": 4541
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.478791449916773e-07,
+      "loss": 1.3252,
+      "step": 4542
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.464119501969531e-07,
+      "loss": 1.0862,
+      "step": 4543
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.449490562913093e-07,
+      "loss": 1.1383,
+      "step": 4544
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.434904639198155e-07,
+      "loss": 1.3083,
+      "step": 4545
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.420361737256438e-07,
+      "loss": 1.1728,
+      "step": 4546
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4058618635007135e-07,
+      "loss": 1.2659,
+      "step": 4547
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3914050243247446e-07,
+      "loss": 1.1466,
+      "step": 4548
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3769912261033534e-07,
+      "loss": 1.2332,
+      "step": 4549
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3626204751923788e-07,
+      "loss": 1.2385,
+      "step": 4550
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3482927779286624e-07,
+      "loss": 1.1386,
+      "step": 4551
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3340081406300618e-07,
+      "loss": 1.202,
+      "step": 4552
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3197665695954606e-07,
+      "loss": 1.3178,
+      "step": 4553
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3055680711047356e-07,
+      "loss": 1.1997,
+      "step": 4554
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2914126514187784e-07,
+      "loss": 1.093,
+      "step": 4555
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2773003167795071e-07,
+      "loss": 1.1391,
+      "step": 4556
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2632310734097994e-07,
+      "loss": 1.2284,
+      "step": 4557
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2492049275135485e-07,
+      "loss": 1.1538,
+      "step": 4558
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2352218852756624e-07,
+      "loss": 1.1424,
+      "step": 4559
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2212819528620199e-07,
+      "loss": 1.177,
+      "step": 4560
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2073851364195042e-07,
+      "loss": 1.2157,
+      "step": 4561
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1935314420759802e-07,
+      "loss": 1.3099,
+      "step": 4562
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.179720875940272e-07,
+      "loss": 1.0286,
+      "step": 4563
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1659534441022422e-07,
+      "loss": 1.1339,
+      "step": 4564
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.15222915263269e-07,
+      "loss": 1.1989,
+      "step": 4565
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1385480075834076e-07,
+      "loss": 1.1906,
+      "step": 4566
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1249100149871693e-07,
+      "loss": 1.1362,
+      "step": 4567
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1113151808576872e-07,
+      "loss": 1.1196,
+      "step": 4568
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0977635111896656e-07,
+      "loss": 1.3203,
+      "step": 4569
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0842550119588024e-07,
+      "loss": 1.2139,
+      "step": 4570
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0707896891216995e-07,
+      "loss": 1.0192,
+      "step": 4571
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0573675486159738e-07,
+      "loss": 1.1672,
+      "step": 4572
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0439885963601803e-07,
+      "loss": 1.3596,
+      "step": 4573
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0306528382538103e-07,
+      "loss": 1.1464,
+      "step": 4574
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0173602801773496e-07,
+      "loss": 1.2858,
+      "step": 4575
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0041109279921868e-07,
+      "loss": 1.213,
+      "step": 4576
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.990904787540704e-07,
+      "loss": 1.1473,
+      "step": 4577
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.97774186464621e-07,
+      "loss": 1.2194,
+      "step": 4578
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9646221651129393e-07,
+      "loss": 1.2089,
+      "step": 4579
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.951545694726098e-07,
+      "loss": 1.0492,
+      "step": 4580
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9385124592518068e-07,
+      "loss": 1.1839,
+      "step": 4581
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.925522464437135e-07,
+      "loss": 1.0446,
+      "step": 4582
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.91257571601009e-07,
+      "loss": 1.2514,
+      "step": 4583
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8996722196795715e-07,
+      "loss": 1.1882,
+      "step": 4584
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8868119811354611e-07,
+      "loss": 1.3234,
+      "step": 4585
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8739950060485234e-07,
+      "loss": 1.1891,
+      "step": 4586
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8612213000704705e-07,
+      "loss": 1.2248,
+      "step": 4587
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8484908688339187e-07,
+      "loss": 1.2724,
+      "step": 4588
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8358037179524223e-07,
+      "loss": 1.1339,
+      "step": 4589
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8231598530204287e-07,
+      "loss": 1.3947,
+      "step": 4590
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.810559279613322e-07,
+      "loss": 1.4148,
+      "step": 4591
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.798002003287347e-07,
+      "loss": 1.2372,
+      "step": 4592
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7854880295797406e-07,
+      "loss": 1.0318,
+      "step": 4593
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7730173640085445e-07,
+      "loss": 1.2405,
+      "step": 4594
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7605900120728048e-07,
+      "loss": 1.1067,
+      "step": 4595
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.748205979252393e-07,
+      "loss": 1.149,
+      "step": 4596
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.735865271008108e-07,
+      "loss": 1.1196,
+      "step": 4597
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7235678927816301e-07,
+      "loss": 1.2778,
+      "step": 4598
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.711313849995555e-07,
+      "loss": 1.2628,
+      "step": 4599
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6991031480533716e-07,
+      "loss": 1.4392,
+      "step": 4600
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.686935792339439e-07,
+      "loss": 1.1817,
+      "step": 4601
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6748117882189886e-07,
+      "loss": 1.0313,
+      "step": 4602
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6627311410381653e-07,
+      "loss": 1.2266,
+      "step": 4603
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.650693856123997e-07,
+      "loss": 1.1784,
+      "step": 4604
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6386999387843717e-07,
+      "loss": 1.2349,
+      "step": 4605
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.626749394308058e-07,
+      "loss": 1.1398,
+      "step": 4606
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.614842227964708e-07,
+      "loss": 1.1919,
+      "step": 4607
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6029784450048325e-07,
+      "loss": 1.2364,
+      "step": 4608
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5911580506598245e-07,
+      "loss": 1.2958,
+      "step": 4609
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5793810501419482e-07,
+      "loss": 1.1896,
+      "step": 4610
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.567647448644327e-07,
+      "loss": 1.1244,
+      "step": 4611
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5559572513409338e-07,
+      "loss": 1.4833,
+      "step": 4612
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5443104633866113e-07,
+      "loss": 1.2151,
+      "step": 4613
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5327070899170736e-07,
+      "loss": 1.2973,
+      "step": 4614
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.521147136048895e-07,
+      "loss": 1.17,
+      "step": 4615
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5096306068794643e-07,
+      "loss": 1.2587,
+      "step": 4616
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4981575074870636e-07,
+      "loss": 1.2526,
+      "step": 4617
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4867278429308018e-07,
+      "loss": 1.2763,
+      "step": 4618
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4753416182506364e-07,
+      "loss": 1.2275,
+      "step": 4619
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4639988384673843e-07,
+      "loss": 1.285,
+      "step": 4620
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.452699508582689e-07,
+      "loss": 1.1666,
+      "step": 4621
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4414436335790537e-07,
+      "loss": 1.293,
+      "step": 4622
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4302312184197975e-07,
+      "loss": 1.36,
+      "step": 4623
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4190622680490873e-07,
+      "loss": 1.3409,
+      "step": 4624
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.407936787391906e-07,
+      "loss": 1.1026,
+      "step": 4625
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.396854781354129e-07,
+      "loss": 1.0783,
+      "step": 4626
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3858162548223807e-07,
+      "loss": 1.1166,
+      "step": 4627
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3748212126641569e-07,
+      "loss": 1.2658,
+      "step": 4628
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3638696597277678e-07,
+      "loss": 1.2559,
+      "step": 4629
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3529616008423506e-07,
+      "loss": 1.2074,
+      "step": 4630
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3420970408178912e-07,
+      "loss": 1.1325,
+      "step": 4631
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3312759844451352e-07,
+      "loss": 1.2198,
+      "step": 4632
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3204984364956874e-07,
+      "loss": 1.216,
+      "step": 4633
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3097644017219468e-07,
+      "loss": 1.41,
+      "step": 4634
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2990738848571493e-07,
+      "loss": 1.1207,
+      "step": 4635
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2884268906153352e-07,
+      "loss": 1.0583,
+      "step": 4636
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2778234236913155e-07,
+      "loss": 1.1976,
+      "step": 4637
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2672634887607616e-07,
+      "loss": 1.3224,
+      "step": 4638
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.256747090480115e-07,
+      "loss": 1.2319,
+      "step": 4639
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2462742334866218e-07,
+      "loss": 1.2373,
+      "step": 4640
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.235844922398355e-07,
+      "loss": 1.1405,
+      "step": 4641
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2254591618141686e-07,
+      "loss": 1.2162,
+      "step": 4642
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.215116956313689e-07,
+      "loss": 1.2755,
+      "step": 4643
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2048183104573563e-07,
+      "loss": 1.1438,
+      "step": 4644
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1945632287864383e-07,
+      "loss": 1.4386,
+      "step": 4645
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1843517158229289e-07,
+      "loss": 1.2129,
+      "step": 4646
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1741837760696595e-07,
+      "loss": 1.139,
+      "step": 4647
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1640594140102213e-07,
+      "loss": 1.189,
+      "step": 4648
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1539786341089876e-07,
+      "loss": 1.1606,
+      "step": 4649
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1439414408111471e-07,
+      "loss": 1.2194,
+      "step": 4650
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1339478385426261e-07,
+      "loss": 1.1501,
+      "step": 4651
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1239978317101663e-07,
+      "loss": 1.3693,
+      "step": 4652
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.114091424701258e-07,
+      "loss": 1.1584,
+      "step": 4653
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1042286218841736e-07,
+      "loss": 1.2628,
+      "step": 4654
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0944094276079676e-07,
+      "loss": 1.2168,
+      "step": 4655
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0846338462024542e-07,
+      "loss": 1.2573,
+      "step": 4656
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0749018819782298e-07,
+      "loss": 1.1115,
+      "step": 4657
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0652135392266394e-07,
+      "loss": 1.3018,
+      "step": 4658
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.055568822219799e-07,
+      "loss": 1.325,
+      "step": 4659
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0459677352106069e-07,
+      "loss": 1.2929,
+      "step": 4660
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0364102824326871e-07,
+      "loss": 1.1934,
+      "step": 4661
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0268964681004356e-07,
+      "loss": 1.2485,
+      "step": 4662
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.017426296409041e-07,
+      "loss": 1.2737,
+      "step": 4663
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.007999771534396e-07,
+      "loss": 1.219,
+      "step": 4664
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.986168976331866e-08,
+      "loss": 1.2712,
+      "step": 4665
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.892776788428149e-08,
+      "loss": 1.3812,
+      "step": 4666
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.79982119281464e-08,
+      "loss": 1.3012,
+      "step": 4667
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.707302230480553e-08,
+      "loss": 1.1331,
+      "step": 4668
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.615219942222476e-08,
+      "loss": 1.1654,
+      "step": 4669
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.523574368644483e-08,
+      "loss": 1.3515,
+      "step": 4670
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.432365550158251e-08,
+      "loss": 1.1582,
+      "step": 4671
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.341593526982606e-08,
+      "loss": 1.2096,
+      "step": 4672
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.251258339143865e-08,
+      "loss": 1.1917,
+      "step": 4673
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.16136002647583e-08,
+      "loss": 1.3432,
+      "step": 4674
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.071898628619569e-08,
+      "loss": 1.031,
+      "step": 4675
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.982874185023415e-08,
+      "loss": 1.2527,
+      "step": 4676
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.89428673494308e-08,
+      "loss": 1.1984,
+      "step": 4677
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.80613631744176e-08,
+      "loss": 1.1994,
+      "step": 4678
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.718422971389584e-08,
+      "loss": 1.2794,
+      "step": 4679
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.631146735464279e-08,
+      "loss": 1.2308,
+      "step": 4680
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.544307648150729e-08,
+      "loss": 1.1889,
+      "step": 4681
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.45790574774108e-08,
+      "loss": 1.2067,
+      "step": 4682
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.3719410723343e-08,
+      "loss": 1.1103,
+      "step": 4683
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.28641365983729e-08,
+      "loss": 1.2453,
+      "step": 4684
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.201323547963546e-08,
+      "loss": 1.3174,
+      "step": 4685
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.116670774234059e-08,
+      "loss": 1.2052,
+      "step": 4686
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.032455375976744e-08,
+      "loss": 1.3911,
+      "step": 4687
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.948677390326786e-08,
+      "loss": 1.2335,
+      "step": 4688
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.865336854226525e-08,
+      "loss": 1.3902,
+      "step": 4689
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.782433804425226e-08,
+      "loss": 1.1569,
+      "step": 4690
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.699968277479652e-08,
+      "loss": 1.1604,
+      "step": 4691
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.617940309753047e-08,
+      "loss": 1.2157,
+      "step": 4692
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.536349937416143e-08,
+      "loss": 1.2399,
+      "step": 4693
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.455197196446495e-08,
+      "loss": 1.2424,
+      "step": 4694
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.374482122628923e-08,
+      "loss": 1.112,
+      "step": 4695
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.294204751555067e-08,
+      "loss": 1.166,
+      "step": 4696
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.214365118623612e-08,
+      "loss": 1.251,
+      "step": 4697
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.134963259040173e-08,
+      "loss": 1.2139,
+      "step": 4698
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.055999207817188e-08,
+      "loss": 1.253,
+      "step": 4699
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.977472999774471e-08,
+      "loss": 1.1988,
+      "step": 4700
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.899384669538434e-08,
+      "loss": 1.3044,
+      "step": 4701
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.821734251542534e-08,
+      "loss": 1.2148,
+      "step": 4702
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.744521780026936e-08,
+      "loss": 1.2393,
+      "step": 4703
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.66774728903885e-08,
+      "loss": 1.2675,
+      "step": 4704
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.59141081243242e-08,
+      "loss": 1.2476,
+      "step": 4705
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.515512383868605e-08,
+      "loss": 1.1944,
+      "step": 4706
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.440052036815081e-08,
+      "loss": 1.1408,
+      "step": 4707
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.365029804546452e-08,
+      "loss": 1.2583,
+      "step": 4708
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.290445720144145e-08,
+      "loss": 1.2463,
+      "step": 4709
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.216299816496186e-08,
+      "loss": 1.1288,
+      "step": 4710
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.142592126297753e-08,
+      "loss": 1.182,
+      "step": 4711
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.069322682050516e-08,
+      "loss": 1.2524,
+      "step": 4712
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.996491516062964e-08,
+      "loss": 1.1481,
+      "step": 4713
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.9240986604502946e-08,
+      "loss": 1.2798,
+      "step": 4714
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.852144147134531e-08,
+      "loss": 1.0135,
+      "step": 4715
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.7806280078444024e-08,
+      "loss": 1.2251,
+      "step": 4716
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.709550274115128e-08,
+      "loss": 1.3014,
+      "step": 4717
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.6389109772887476e-08,
+      "loss": 1.3455,
+      "step": 4718
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.568710148514123e-08,
+      "loss": 1.2819,
+      "step": 4719
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.498947818746603e-08,
+      "loss": 1.2589,
+      "step": 4720
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.429624018748136e-08,
+      "loss": 1.2098,
+      "step": 4721
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.360738779087382e-08,
+      "loss": 1.2957,
+      "step": 4722
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.292292130139598e-08,
+      "loss": 1.2596,
+      "step": 4723
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.224284102086641e-08,
+      "loss": 1.1762,
+      "step": 4724
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.156714724917078e-08,
+      "loss": 1.1404,
+      "step": 4725
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.0895840284257424e-08,
+      "loss": 1.2224,
+      "step": 4726
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.022892042214289e-08,
+      "loss": 1.2188,
+      "step": 4727
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.9566387956909713e-08,
+      "loss": 1.2165,
+      "step": 4728
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.8908243180704194e-08,
+      "loss": 1.2537,
+      "step": 4729
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.825448638373642e-08,
+      "loss": 1.2389,
+      "step": 4730
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.760511785428468e-08,
+      "loss": 1.2015,
+      "step": 4731
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.696013787869213e-08,
+      "loss": 1.3308,
+      "step": 4732
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.631954674136463e-08,
+      "loss": 1.226,
+      "step": 4733
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.568334472477287e-08,
+      "loss": 1.139,
+      "step": 4734
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.505153210945468e-08,
+      "loss": 1.0113,
+      "step": 4735
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.442410917400941e-08,
+      "loss": 1.171,
+      "step": 4736
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.380107619510243e-08,
+      "loss": 1.2145,
+      "step": 4737
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.318243344746287e-08,
+      "loss": 1.2584,
+      "step": 4738
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.256818120388473e-08,
+      "loss": 1.1433,
+      "step": 4739
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.195831973522468e-08,
+      "loss": 1.1286,
+      "step": 4740
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.1352849310404284e-08,
+      "loss": 1.0429,
+      "step": 4741
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.075177019640775e-08,
+      "loss": 1.0546,
+      "step": 4742
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.015508265828527e-08,
+      "loss": 1.2894,
+      "step": 4743
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.9562786959147505e-08,
+      "loss": 1.267,
+      "step": 4744
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.8974883360169966e-08,
+      "loss": 1.126,
+      "step": 4745
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.8391372120591964e-08,
+      "loss": 1.2385,
+      "step": 4746
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.781225349771545e-08,
+      "loss": 1.2288,
+      "step": 4747
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.7237527746905034e-08,
+      "loss": 1.229,
+      "step": 4748
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.666719512158912e-08,
+      "loss": 1.2016,
+      "step": 4749
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.6101255873257635e-08,
+      "loss": 1.1657,
+      "step": 4750
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.553971025146541e-08,
+      "loss": 1.2329,
+      "step": 4751
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.498255850382659e-08,
+      "loss": 1.3761,
+      "step": 4752
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.442980087602133e-08,
+      "loss": 1.3463,
+      "step": 4753
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.38814376117913e-08,
+      "loss": 1.2366,
+      "step": 4754
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.333746895293755e-08,
+      "loss": 1.1543,
+      "step": 4755
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.2797895139327076e-08,
+      "loss": 1.2818,
+      "step": 4756
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.226271640888734e-08,
+      "loss": 1.2335,
+      "step": 4757
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.173193299760735e-08,
+      "loss": 1.1523,
+      "step": 4758
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.120554513953877e-08,
+      "loss": 1.1394,
+      "step": 4759
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.0683553066793715e-08,
+      "loss": 1.2175,
+      "step": 4760
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.016595700954916e-08,
+      "loss": 1.2173,
+      "step": 4761
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9652757196039217e-08,
+      "loss": 1.2389,
+      "step": 4762
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9143953852562856e-08,
+      "loss": 1.2236,
+      "step": 4763
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.86395472034795e-08,
+      "loss": 1.2154,
+      "step": 4764
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.813953747120901e-08,
+      "loss": 1.0721,
+      "step": 4765
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7643924876232798e-08,
+      "loss": 1.3581,
+      "step": 4766
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.715270963709382e-08,
+      "loss": 1.252,
+      "step": 4767
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6665891970395484e-08,
+      "loss": 1.207,
+      "step": 4768
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6183472090801633e-08,
+      "loss": 1.2807,
+      "step": 4769
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5705450211038764e-08,
+      "loss": 1.2961,
+      "step": 4770
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5231826541891602e-08,
+      "loss": 1.1777,
+      "step": 4771
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.4762601292208642e-08,
+      "loss": 1.2797,
+      "step": 4772
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.429777466889438e-08,
+      "loss": 1.1613,
+      "step": 4773
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3837346876918187e-08,
+      "loss": 1.2037,
+      "step": 4774
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.338131811930655e-08,
+      "loss": 1.1983,
+      "step": 4775
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2929688597147505e-08,
+      "loss": 1.232,
+      "step": 4776
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.248245850959063e-08,
+      "loss": 1.1851,
+      "step": 4777
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2039628053842633e-08,
+      "loss": 1.1281,
+      "step": 4778
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.160119742517064e-08,
+      "loss": 1.3132,
+      "step": 4779
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.116716681690556e-08,
+      "loss": 1.0678,
+      "step": 4780
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.073753642043208e-08,
+      "loss": 1.3149,
+      "step": 4781
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.031230642520088e-08,
+      "loss": 1.2045,
+      "step": 4782
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.989147701871641e-08,
+      "loss": 1.2228,
+      "step": 4783
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9475048386546903e-08,
+      "loss": 1.2217,
+      "step": 4784
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9063020712316582e-08,
+      "loss": 1.1387,
+      "step": 4785
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8655394177712337e-08,
+      "loss": 1.1854,
+      "step": 4786
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8252168962479277e-08,
+      "loss": 1.3369,
+      "step": 4787
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.785334524442073e-08,
+      "loss": 1.2,
+      "step": 4788
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7458923199400456e-08,
+      "loss": 1.2647,
+      "step": 4789
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7068903001339342e-08,
+      "loss": 1.2489,
+      "step": 4790
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6683284822219814e-08,
+      "loss": 1.2986,
+      "step": 4791
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6302068832081407e-08,
+      "loss": 1.1419,
+      "step": 4792
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5925255199024103e-08,
+      "loss": 1.322,
+      "step": 4793
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.555284408920388e-08,
+      "loss": 1.2308,
+      "step": 4794
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.518483566683826e-08,
+      "loss": 1.2904,
+      "step": 4795
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4821230094200779e-08,
+      "loss": 1.1627,
+      "step": 4796
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4462027531627621e-08,
+      "loss": 1.2652,
+      "step": 4797
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4107228137508755e-08,
+      "loss": 1.1136,
+      "step": 4798
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.375683206829459e-08,
+      "loss": 1.1956,
+      "step": 4799
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3410839478493753e-08,
+      "loss": 1.2432,
+      "step": 4800
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3069250520675315e-08,
+      "loss": 1.1954,
+      "step": 4801
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2732065345462118e-08,
+      "loss": 1.2832,
+      "step": 4802
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2399284101538566e-08,
+      "loss": 1.3121,
+      "step": 4803
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2070906935646165e-08,
+      "loss": 1.1691,
+      "step": 4804
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1746933992584642e-08,
+      "loss": 1.1947,
+      "step": 4805
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1427365415209723e-08,
+      "loss": 1.1433,
+      "step": 4806
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1112201344438689e-08,
+      "loss": 1.2185,
+      "step": 4807
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0801441919242594e-08,
+      "loss": 1.2454,
+      "step": 4808
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0495087276654049e-08,
+      "loss": 1.2612,
+      "step": 4809
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0193137551759435e-08,
+      "loss": 1.2115,
+      "step": 4810
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.895592877706694e-09,
+      "loss": 1.309,
+      "step": 4811
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.602453385699762e-09,
+      "loss": 1.2677,
+      "step": 4812
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.313719204997463e-09,
+      "loss": 1.2149,
+      "step": 4813
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.029390462921283e-09,
+      "loss": 1.1428,
+      "step": 4814
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.74946728484538e-09,
+      "loss": 1.159,
+      "step": 4815
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.47394979420324e-09,
+      "loss": 1.1958,
+      "step": 4816
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.202838112486566e-09,
+      "loss": 1.3329,
+      "step": 4817
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.936132359243066e-09,
+      "loss": 1.2174,
+      "step": 4818
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.67383265207755e-09,
+      "loss": 1.318,
+      "step": 4819
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.415939106651948e-09,
+      "loss": 1.2704,
+      "step": 4820
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.162451836685291e-09,
+      "loss": 1.2378,
+      "step": 4821
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.9133709539559445e-09,
+      "loss": 1.2622,
+      "step": 4822
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.6686965682949415e-09,
+      "loss": 1.3181,
+      "step": 4823
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.428428787593755e-09,
+      "loss": 1.1412,
+      "step": 4824
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.192567717798748e-09,
+      "loss": 1.1366,
+      "step": 4825
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.961113462915613e-09,
+      "loss": 1.1974,
+      "step": 4826
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.7340661250038235e-09,
+      "loss": 1.1074,
+      "step": 4827
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.511425804179959e-09,
+      "loss": 1.2446,
+      "step": 4828
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.2931925986210444e-09,
+      "loss": 1.1772,
+      "step": 4829
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.079366604555658e-09,
+      "loss": 1.2794,
+      "step": 4830
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.869947916271711e-09,
+      "loss": 1.2812,
+      "step": 4831
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.664936626114225e-09,
+      "loss": 1.267,
+      "step": 4832
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.464332824483108e-09,
+      "loss": 1.1833,
+      "step": 4833
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.268136599836492e-09,
+      "loss": 1.2392,
+      "step": 4834
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.076348038687394e-09,
+      "loss": 1.3277,
+      "step": 4835
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.888967225604834e-09,
+      "loss": 1.2032,
+      "step": 4836
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.705994243217159e-09,
+      "loss": 1.0748,
+      "step": 4837
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.527429172205388e-09,
+      "loss": 1.1985,
+      "step": 4838
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.353272091309867e-09,
+      "loss": 1.3882,
+      "step": 4839
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.183523077324724e-09,
+      "loss": 1.1223,
+      "step": 4840
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.0181822051023047e-09,
+      "loss": 1.3161,
+      "step": 4841
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.8572495475509553e-09,
+      "loss": 1.2514,
+      "step": 4842
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.7007251756339113e-09,
+      "loss": 1.2303,
+      "step": 4843
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.548609158371518e-09,
+      "loss": 1.2187,
+      "step": 4844
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.400901562840119e-09,
+      "loss": 1.1654,
+      "step": 4845
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.257602454172059e-09,
+      "loss": 1.1465,
+      "step": 4846
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.1187118955556808e-09,
+      "loss": 1.1617,
+      "step": 4847
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.984229948235328e-09,
+      "loss": 1.3754,
+      "step": 4848
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8541566715113423e-09,
+      "loss": 1.1904,
+      "step": 4849
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.7284921227400664e-09,
+      "loss": 1.2327,
+      "step": 4850
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6072363573338412e-09,
+      "loss": 1.1634,
+      "step": 4851
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4903894287610077e-09,
+      "loss": 1.2031,
+      "step": 4852
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3779513885470163e-09,
+      "loss": 1.2114,
+      "step": 4853
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2699222862699868e-09,
+      "loss": 1.1661,
+      "step": 4854
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.166302169566258e-09,
+      "loss": 1.2802,
+      "step": 4855
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.067091084128169e-09,
+      "loss": 1.2324,
+      "step": 4856
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.722890737029478e-10,
+      "loss": 1.4101,
+      "step": 4857
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.818961800949322e-10,
+      "loss": 1.2468,
+      "step": 4858
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.959124431622389e-10,
+      "loss": 1.2837,
+      "step": 4859
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.143379008200946e-10,
+      "loss": 1.2896,
+      "step": 4860
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.371725890386149e-10,
+      "loss": 1.2106,
+      "step": 4861
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.644165418450253e-10,
+      "loss": 1.2835,
+      "step": 4862
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.960697913203305e-10,
+      "loss": 1.2853,
+      "step": 4863
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.3213236760375475e-10,
+      "loss": 1.312,
+      "step": 4864
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.7260429888830163e-10,
+      "loss": 1.1209,
+      "step": 4865
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.174856114229741e-10,
+      "loss": 1.1715,
+      "step": 4866
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.6677632951277456e-10,
+      "loss": 1.2605,
+      "step": 4867
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.2047647551759476e-10,
+      "loss": 1.2873,
+      "step": 4868
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.7858606985443616e-10,
+      "loss": 1.2911,
+      "step": 4869
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4110513099407918e-10,
+      "loss": 1.274,
+      "step": 4870
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.0803367546441401e-10,
+      "loss": 1.3809,
+      "step": 4871
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.937171784933028e-11,
+      "loss": 1.1326,
+      "step": 4872
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.51192707864967e-11,
+      "loss": 1.2368,
+      "step": 4873
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.527634496958143e-11,
+      "loss": 1.3165,
+      "step": 4874
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9842949149362358e-11,
+      "loss": 1.2808,
+      "step": 4875
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.819090131506613e-12,
+      "loss": 1.1958,
+      "step": 4876
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.2047727754603843e-12,
+      "loss": 1.1068,
+      "step": 4877
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 1.2134,
+      "step": 4878
+    },
+    {
+      "epoch": 3.0,
+      "step": 4878,
+      "total_flos": 7.916434027612078e+17,
+      "train_loss": 1.2485257899140472,
+      "train_runtime": 4894.132,
+      "train_samples_per_second": 31.876,
+      "train_steps_per_second": 0.997
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 4878,
+  "num_train_epochs": 3,
+  "save_steps": 1000,
+  "total_flos": 7.916434027612078e+17,
+  "trial_name": null,
+  "trial_params": null
+}