diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,14586 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 2426,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.47945205479452e-07,
+      "loss": 0.7251,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.095890410958904e-06,
+      "loss": 0.7015,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6438356164383561e-06,
+      "loss": 0.711,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.191780821917808e-06,
+      "loss": 0.7199,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7397260273972604e-06,
+      "loss": 0.7162,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2876712328767123e-06,
+      "loss": 0.6828,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8356164383561645e-06,
+      "loss": 0.6762,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.383561643835616e-06,
+      "loss": 0.636,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.931506849315069e-06,
+      "loss": 0.6087,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.479452054794521e-06,
+      "loss": 0.6188,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.027397260273973e-06,
+      "loss": 0.6265,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.5753424657534245e-06,
+      "loss": 0.6262,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.123287671232877e-06,
+      "loss": 0.6285,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.671232876712329e-06,
+      "loss": 0.584,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.219178082191782e-06,
+      "loss": 0.5821,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.767123287671233e-06,
+      "loss": 0.5811,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.315068493150685e-06,
+      "loss": 0.5499,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.863013698630138e-06,
+      "loss": 0.5701,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0410958904109589e-05,
+      "loss": 0.572,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0958904109589042e-05,
+      "loss": 0.541,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1506849315068493e-05,
+      "loss": 0.5314,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2054794520547945e-05,
+      "loss": 0.5522,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2602739726027398e-05,
+      "loss": 0.5133,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3150684931506849e-05,
+      "loss": 0.507,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3698630136986302e-05,
+      "loss": 0.5164,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4246575342465754e-05,
+      "loss": 0.526,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4794520547945205e-05,
+      "loss": 0.5296,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5342465753424658e-05,
+      "loss": 0.5002,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.589041095890411e-05,
+      "loss": 0.4886,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6438356164383563e-05,
+      "loss": 0.5641,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6986301369863014e-05,
+      "loss": 0.5053,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7534246575342465e-05,
+      "loss": 0.509,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.808219178082192e-05,
+      "loss": 0.4915,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.863013698630137e-05,
+      "loss": 0.4765,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9178082191780822e-05,
+      "loss": 0.4906,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9726027397260276e-05,
+      "loss": 0.4889,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0273972602739727e-05,
+      "loss": 0.4874,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0821917808219178e-05,
+      "loss": 0.4639,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1369863013698632e-05,
+      "loss": 0.4976,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1917808219178083e-05,
+      "loss": 0.4837,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2465753424657538e-05,
+      "loss": 0.5033,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3013698630136985e-05,
+      "loss": 0.4741,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.356164383561644e-05,
+      "loss": 0.4888,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.410958904109589e-05,
+      "loss": 0.455,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4657534246575345e-05,
+      "loss": 0.4867,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5205479452054796e-05,
+      "loss": 0.4778,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.575342465753425e-05,
+      "loss": 0.4937,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6301369863013698e-05,
+      "loss": 0.4524,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6849315068493153e-05,
+      "loss": 0.4843,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7397260273972603e-05,
+      "loss": 0.4661,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7945205479452058e-05,
+      "loss": 0.4423,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.849315068493151e-05,
+      "loss": 0.5043,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9041095890410963e-05,
+      "loss": 0.4708,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.958904109589041e-05,
+      "loss": 0.4749,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0136986301369865e-05,
+      "loss": 0.4997,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0684931506849316e-05,
+      "loss": 0.5062,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.123287671232877e-05,
+      "loss": 0.4853,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.178082191780822e-05,
+      "loss": 0.4886,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2328767123287676e-05,
+      "loss": 0.4577,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.287671232876713e-05,
+      "loss": 0.4639,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.342465753424658e-05,
+      "loss": 0.4474,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.397260273972603e-05,
+      "loss": 0.4879,
+      "step": 62
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.452054794520548e-05,
+      "loss": 0.4559,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.506849315068493e-05,
+      "loss": 0.4667,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.561643835616439e-05,
+      "loss": 0.4847,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.616438356164384e-05,
+      "loss": 0.4687,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.671232876712329e-05,
+      "loss": 0.4918,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.726027397260274e-05,
+      "loss": 0.4597,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.780821917808219e-05,
+      "loss": 0.4847,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.8356164383561644e-05,
+      "loss": 0.4634,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.89041095890411e-05,
+      "loss": 0.4623,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.945205479452055e-05,
+      "loss": 0.472,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4e-05,
+      "loss": 0.4767,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999998217392102e-05,
+      "loss": 0.4744,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999992869571584e-05,
+      "loss": 0.4729,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9999839565479795e-05,
+      "loss": 0.464,
+      "step": 76
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999971478337176e-05,
+      "loss": 0.4694,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9999554349614194e-05,
+      "loss": 0.5059,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999935826449307e-05,
+      "loss": 0.4843,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999912652835794e-05,
+      "loss": 0.4873,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999885914162189e-05,
+      "loss": 0.4697,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999855610476158e-05,
+      "loss": 0.4548,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999821741831719e-05,
+      "loss": 0.453,
+      "step": 83
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999784308289247e-05,
+      "loss": 0.4822,
+      "step": 84
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999743309915472e-05,
+      "loss": 0.493,
+      "step": 85
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999698746783477e-05,
+      "loss": 0.439,
+      "step": 86
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999650618972702e-05,
+      "loss": 0.4778,
+      "step": 87
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999598926568938e-05,
+      "loss": 0.5153,
+      "step": 88
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9995436696643344e-05,
+      "loss": 0.4787,
+      "step": 89
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999484848357391e-05,
+      "loss": 0.4377,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999422462752964e-05,
+      "loss": 0.4885,
+      "step": 91
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999356512962262e-05,
+      "loss": 0.4927,
+      "step": 92
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9992869991028484e-05,
+      "loss": 0.4507,
+      "step": 93
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999213921298638e-05,
+      "loss": 0.4489,
+      "step": 94
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999137279679901e-05,
+      "loss": 0.4567,
+      "step": 95
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999057074383259e-05,
+      "loss": 0.4966,
+      "step": 96
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9989733055516855e-05,
+      "loss": 0.4575,
+      "step": 97
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998885973334509e-05,
+      "loss": 0.4908,
+      "step": 98
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998795077887408e-05,
+      "loss": 0.4556,
+      "step": 99
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998700619372413e-05,
+      "loss": 0.469,
+      "step": 100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998602597957908e-05,
+      "loss": 0.4664,
+      "step": 101
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998501013818625e-05,
+      "loss": 0.4642,
+      "step": 102
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9983958671356485e-05,
+      "loss": 0.4598,
+      "step": 103
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998287158096416e-05,
+      "loss": 0.4862,
+      "step": 104
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998174886894711e-05,
+      "loss": 0.4863,
+      "step": 105
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.99805905373067e-05,
+      "loss": 0.4477,
+      "step": 106
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9979396588107784e-05,
+      "loss": 0.4713,
+      "step": 107
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9978167023478695e-05,
+      "loss": 0.4639,
+      "step": 108
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.997690184561127e-05,
+      "loss": 0.4891,
+      "step": 109
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997560105676082e-05,
+      "loss": 0.457,
+      "step": 110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997426465924616e-05,
+      "loss": 0.4499,
+      "step": 111
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997289265544954e-05,
+      "loss": 0.4736,
+      "step": 112
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997148504781672e-05,
+      "loss": 0.4282,
+      "step": 113
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.99700418388569e-05,
+      "loss": 0.4611,
+      "step": 114
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9968563031142766e-05,
+      "loss": 0.4882,
+      "step": 115
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996704862731045e-05,
+      "loss": 0.4803,
+      "step": 116
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996549863005954e-05,
+      "loss": 0.4692,
+      "step": 117
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996391304215306e-05,
+      "loss": 0.4515,
+      "step": 118
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996229186641752e-05,
+      "loss": 0.4577,
+      "step": 119
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9960635105742815e-05,
+      "loss": 0.4481,
+      "step": 120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9958942763082315e-05,
+      "loss": 0.4715,
+      "step": 121
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995721484145279e-05,
+      "loss": 0.4607,
+      "step": 122
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995545134393446e-05,
+      "loss": 0.4912,
+      "step": 123
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995365227367094e-05,
+      "loss": 0.4692,
+      "step": 124
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995181763386928e-05,
+      "loss": 0.4481,
+      "step": 125
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.99499474277999e-05,
+      "loss": 0.4534,
+      "step": 126
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994804165879667e-05,
+      "loss": 0.4811,
+      "step": 127
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994610033025681e-05,
+      "loss": 0.4497,
+      "step": 128
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9944123445640966e-05,
+      "loss": 0.4561,
+      "step": 129
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994211100847312e-05,
+      "loss": 0.4494,
+      "step": 130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994006302234068e-05,
+      "loss": 0.4687,
+      "step": 131
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9937979490894406e-05,
+      "loss": 0.4691,
+      "step": 132
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.99358604178484e-05,
+      "loss": 0.465,
+      "step": 133
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9933705806980155e-05,
+      "loss": 0.4502,
+      "step": 134
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.993151566213048e-05,
+      "loss": 0.4582,
+      "step": 135
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9929289987203565e-05,
+      "loss": 0.4514,
+      "step": 136
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9927028786166904e-05,
+      "loss": 0.4744,
+      "step": 137
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9924732063051325e-05,
+      "loss": 0.4605,
+      "step": 138
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9922399821950994e-05,
+      "loss": 0.4647,
+      "step": 139
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9920032067023384e-05,
+      "loss": 0.4466,
+      "step": 140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.991762880248927e-05,
+      "loss": 0.4304,
+      "step": 141
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.991519003263274e-05,
+      "loss": 0.4507,
+      "step": 142
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.991271576180114e-05,
+      "loss": 0.4373,
+      "step": 143
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.991020599440515e-05,
+      "loss": 0.4588,
+      "step": 144
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9907660734918685e-05,
+      "loss": 0.4488,
+      "step": 145
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9905079987878956e-05,
+      "loss": 0.4698,
+      "step": 146
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9902463757886414e-05,
+      "loss": 0.465,
+      "step": 147
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989981204960478e-05,
+      "loss": 0.4521,
+      "step": 148
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9897124867760996e-05,
+      "loss": 0.4551,
+      "step": 149
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989440221714527e-05,
+      "loss": 0.4283,
+      "step": 150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989164410261102e-05,
+      "loss": 0.4701,
+      "step": 151
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9888850529074865e-05,
+      "loss": 0.473,
+      "step": 152
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.988602150151667e-05,
+      "loss": 0.441,
+      "step": 153
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.988315702497946e-05,
+      "loss": 0.443,
+      "step": 154
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.988025710456951e-05,
+      "loss": 0.457,
+      "step": 155
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.98773217454562e-05,
+      "loss": 0.4393,
+      "step": 156
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9874350952872156e-05,
+      "loss": 0.4631,
+      "step": 157
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9871344732113113e-05,
+      "loss": 0.428,
+      "step": 158
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9868303088538e-05,
+      "loss": 0.4361,
+      "step": 159
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.986522602756887e-05,
+      "loss": 0.4312,
+      "step": 160
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.986211355469092e-05,
+      "loss": 0.4261,
+      "step": 161
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9858965675452454e-05,
+      "loss": 0.4439,
+      "step": 162
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.985578239546492e-05,
+      "loss": 0.4284,
+      "step": 163
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9852563720402854e-05,
+      "loss": 0.4236,
+      "step": 164
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.984930965600389e-05,
+      "loss": 0.4426,
+      "step": 165
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9846020208068764e-05,
+      "loss": 0.4703,
+      "step": 166
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9842695382461244e-05,
+      "loss": 0.432,
+      "step": 167
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9839335185108214e-05,
+      "loss": 0.456,
+      "step": 168
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.983593962199958e-05,
+      "loss": 0.4106,
+      "step": 169
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.98325086991883e-05,
+      "loss": 0.4263,
+      "step": 170
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.982904242279036e-05,
+      "loss": 0.4221,
+      "step": 171
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.982554079898477e-05,
+      "loss": 0.453,
+      "step": 172
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9822003834013564e-05,
+      "loss": 0.4355,
+      "step": 173
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.981843153418176e-05,
+      "loss": 0.426,
+      "step": 174
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.981482390585735e-05,
+      "loss": 0.4444,
+      "step": 175
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.981118095547135e-05,
+      "loss": 0.3989,
+      "step": 176
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.98075026895177e-05,
+      "loss": 0.4394,
+      "step": 177
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9803789114553295e-05,
+      "loss": 0.4453,
+      "step": 178
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.980004023719799e-05,
+      "loss": 0.4449,
+      "step": 179
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9796256064134574e-05,
+      "loss": 0.4261,
+      "step": 180
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.979243660210873e-05,
+      "loss": 0.4531,
+      "step": 181
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.978858185792907e-05,
+      "loss": 0.4353,
+      "step": 182
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.978469183846709e-05,
+      "loss": 0.4177,
+      "step": 183
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.978076655065715e-05,
+      "loss": 0.4571,
+      "step": 184
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9776806001496536e-05,
+      "loss": 0.4358,
+      "step": 185
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.977281019804532e-05,
+      "loss": 0.4461,
+      "step": 186
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.976877914742648e-05,
+      "loss": 0.4772,
+      "step": 187
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9764712856825775e-05,
+      "loss": 0.4784,
+      "step": 188
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.976061133349182e-05,
+      "loss": 0.4358,
+      "step": 189
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.975647458473603e-05,
+      "loss": 0.4412,
+      "step": 190
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9752302617932586e-05,
+      "loss": 0.4148,
+      "step": 191
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.974809544051848e-05,
+      "loss": 0.454,
+      "step": 192
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.974385305999346e-05,
+      "loss": 0.4124,
+      "step": 193
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9739575483920026e-05,
+      "loss": 0.4509,
+      "step": 194
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9735262719923424e-05,
+      "loss": 0.4305,
+      "step": 195
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.973091477569161e-05,
+      "loss": 0.4436,
+      "step": 196
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.972653165897527e-05,
+      "loss": 0.4594,
+      "step": 197
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.972211337758779e-05,
+      "loss": 0.4539,
+      "step": 198
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.971765993940522e-05,
+      "loss": 0.4258,
+      "step": 199
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.97131713523663e-05,
+      "loss": 0.4403,
+      "step": 200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.970864762447242e-05,
+      "loss": 0.3987,
+      "step": 201
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.970408876378761e-05,
+      "loss": 0.4218,
+      "step": 202
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9699494778438544e-05,
+      "loss": 0.4403,
+      "step": 203
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.969486567661448e-05,
+      "loss": 0.4623,
+      "step": 204
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.96902014665673e-05,
+      "loss": 0.4419,
+      "step": 205
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.968550215661147e-05,
+      "loss": 0.46,
+      "step": 206
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9680767755123995e-05,
+      "loss": 0.4291,
+      "step": 207
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.967599827054447e-05,
+      "loss": 0.4007,
+      "step": 208
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.967119371137502e-05,
+      "loss": 0.4722,
+      "step": 209
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.966635408618028e-05,
+      "loss": 0.4333,
+      "step": 210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.966147940358741e-05,
+      "loss": 0.4362,
+      "step": 211
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.965656967228606e-05,
+      "loss": 0.437,
+      "step": 212
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.965162490102835e-05,
+      "loss": 0.3975,
+      "step": 213
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9646645098628866e-05,
+      "loss": 0.3996,
+      "step": 214
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9641630273964654e-05,
+      "loss": 0.4239,
+      "step": 215
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9636580435975174e-05,
+      "loss": 0.4142,
+      "step": 216
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.963149559366231e-05,
+      "loss": 0.4139,
+      "step": 217
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.962637575609033e-05,
+      "loss": 0.5019,
+      "step": 218
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.962122093238591e-05,
+      "loss": 0.3886,
+      "step": 219
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.961603113173808e-05,
+      "loss": 0.465,
+      "step": 220
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.961080636339821e-05,
+      "loss": 0.4312,
+      "step": 221
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.960554663668002e-05,
+      "loss": 0.4211,
+      "step": 222
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9600251960959536e-05,
+      "loss": 0.448,
+      "step": 223
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.959492234567509e-05,
+      "loss": 0.4422,
+      "step": 224
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.958955780032731e-05,
+      "loss": 0.4292,
+      "step": 225
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.958415833447905e-05,
+      "loss": 0.618,
+      "step": 226
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.957872395775546e-05,
+      "loss": 0.4206,
+      "step": 227
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.95732546798439e-05,
+      "loss": 0.4482,
+      "step": 228
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9567750510493944e-05,
+      "loss": 0.4374,
+      "step": 229
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9562211459517375e-05,
+      "loss": 0.4129,
+      "step": 230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9556637536788144e-05,
+      "loss": 0.4563,
+      "step": 231
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.955102875224237e-05,
+      "loss": 0.443,
+      "step": 232
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.95453851158783e-05,
+      "loss": 0.418,
+      "step": 233
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9539706637756354e-05,
+      "loss": 0.4545,
+      "step": 234
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.953399332799902e-05,
+      "loss": 0.4483,
+      "step": 235
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.952824519679089e-05,
+      "loss": 0.4271,
+      "step": 236
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.952246225437862e-05,
+      "loss": 0.453,
+      "step": 237
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9516644511070936e-05,
+      "loss": 0.4368,
+      "step": 238
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.951079197723859e-05,
+      "loss": 0.4466,
+      "step": 239
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.950490466331436e-05,
+      "loss": 0.4612,
+      "step": 240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.949898257979302e-05,
+      "loss": 0.4187,
+      "step": 241
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.949302573723131e-05,
+      "loss": 0.4348,
+      "step": 242
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.948703414624796e-05,
+      "loss": 0.4085,
+      "step": 243
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.948100781752362e-05,
+      "loss": 0.4585,
+      "step": 244
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.947494676180086e-05,
+      "loss": 0.4124,
+      "step": 245
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9468850989884194e-05,
+      "loss": 0.4149,
+      "step": 246
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9462720512639965e-05,
+      "loss": 0.4312,
+      "step": 247
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9456555340996424e-05,
+      "loss": 0.4713,
+      "step": 248
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.945035548594366e-05,
+      "loss": 0.4166,
+      "step": 249
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9444120958533565e-05,
+      "loss": 0.443,
+      "step": 250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.943785176987987e-05,
+      "loss": 0.4328,
+      "step": 251
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9431547931158076e-05,
+      "loss": 0.453,
+      "step": 252
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9425209453605464e-05,
+      "loss": 0.4422,
+      "step": 253
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9418836348521045e-05,
+      "loss": 0.4277,
+      "step": 254
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.941242862726557e-05,
+      "loss": 0.4031,
+      "step": 255
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.940598630126149e-05,
+      "loss": 0.4194,
+      "step": 256
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.939950938199295e-05,
+      "loss": 0.4325,
+      "step": 257
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.939299788100576e-05,
+      "loss": 0.4324,
+      "step": 258
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9386451809907363e-05,
+      "loss": 0.4144,
+      "step": 259
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.937987118036684e-05,
+      "loss": 0.4266,
+      "step": 260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.937325600411489e-05,
+      "loss": 0.423,
+      "step": 261
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9366606292943756e-05,
+      "loss": 0.4247,
+      "step": 262
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9359922058707273e-05,
+      "loss": 0.4509,
+      "step": 263
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.935320331332082e-05,
+      "loss": 0.4068,
+      "step": 264
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.934645006876127e-05,
+      "loss": 0.4058,
+      "step": 265
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9339662337067025e-05,
+      "loss": 0.4552,
+      "step": 266
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.933284013033793e-05,
+      "loss": 0.4351,
+      "step": 267
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.932598346073533e-05,
+      "loss": 0.4335,
+      "step": 268
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9319092340481955e-05,
+      "loss": 0.4479,
+      "step": 269
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.931216678186198e-05,
+      "loss": 0.418,
+      "step": 270
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.930520679722096e-05,
+      "loss": 0.4166,
+      "step": 271
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.929821239896582e-05,
+      "loss": 0.4041,
+      "step": 272
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.929118359956483e-05,
+      "loss": 0.3982,
+      "step": 273
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9284120411547587e-05,
+      "loss": 0.4326,
+      "step": 274
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.927702284750497e-05,
+      "loss": 0.4346,
+      "step": 275
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9269890920089174e-05,
+      "loss": 0.4089,
+      "step": 276
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9262724642013615e-05,
+      "loss": 0.4439,
+      "step": 277
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.925552402605295e-05,
+      "loss": 0.4043,
+      "step": 278
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9248289085043075e-05,
+      "loss": 0.4496,
+      "step": 279
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.924101983188104e-05,
+      "loss": 0.3963,
+      "step": 280
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.923371627952507e-05,
+      "loss": 0.4088,
+      "step": 281
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9226378440994545e-05,
+      "loss": 0.3928,
+      "step": 282
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.921900632936995e-05,
+      "loss": 0.4288,
+      "step": 283
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9211599957792866e-05,
+      "loss": 0.4231,
+      "step": 284
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.920415933946595e-05,
+      "loss": 0.4032,
+      "step": 285
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9196684487652914e-05,
+      "loss": 0.4236,
+      "step": 286
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9189175415678484e-05,
+      "loss": 0.3949,
+      "step": 287
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9181632136928384e-05,
+      "loss": 0.4441,
+      "step": 288
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9174054664849336e-05,
+      "loss": 0.397,
+      "step": 289
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.916644301294899e-05,
+      "loss": 0.439,
+      "step": 290
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.915879719479594e-05,
+      "loss": 0.429,
+      "step": 291
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.915111722401968e-05,
+      "loss": 0.4271,
+      "step": 292
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.914340311431059e-05,
+      "loss": 0.4472,
+      "step": 293
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.913565487941991e-05,
+      "loss": 0.4484,
+      "step": 294
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.912787253315969e-05,
+      "loss": 0.4244,
+      "step": 295
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.912005608940281e-05,
+      "loss": 0.3716,
+      "step": 296
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.911220556208292e-05,
+      "loss": 0.4074,
+      "step": 297
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.910432096519444e-05,
+      "loss": 0.4177,
+      "step": 298
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.909640231279251e-05,
+      "loss": 0.4159,
+      "step": 299
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.908844961899298e-05,
+      "loss": 0.4068,
+      "step": 300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9080462897972386e-05,
+      "loss": 0.3828,
+      "step": 301
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.907244216396792e-05,
+      "loss": 0.4219,
+      "step": 302
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9064387431277406e-05,
+      "loss": 0.3993,
+      "step": 303
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9056298714259274e-05,
+      "loss": 0.4307,
+      "step": 304
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9048176027332545e-05,
+      "loss": 0.4151,
+      "step": 305
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9040019384976763e-05,
+      "loss": 0.4145,
+      "step": 306
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9031828801732044e-05,
+      "loss": 0.4217,
+      "step": 307
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.902360429219897e-05,
+      "loss": 0.4486,
+      "step": 308
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9015345871038636e-05,
+      "loss": 0.445,
+      "step": 309
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.900705355297255e-05,
+      "loss": 0.4611,
+      "step": 310
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8998727352782674e-05,
+      "loss": 0.416,
+      "step": 311
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.899036728531135e-05,
+      "loss": 0.4575,
+      "step": 312
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.89819733654613e-05,
+      "loss": 0.4161,
+      "step": 313
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8973545608195615e-05,
+      "loss": 0.4246,
+      "step": 314
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.896508402853766e-05,
+      "loss": 0.4185,
+      "step": 315
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8956588641571106e-05,
+      "loss": 0.4074,
+      "step": 316
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8948059462439916e-05,
+      "loss": 0.4191,
+      "step": 317
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8939496506348264e-05,
+      "loss": 0.387,
+      "step": 318
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.893089978856054e-05,
+      "loss": 0.3883,
+      "step": 319
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8922269324401334e-05,
+      "loss": 0.4127,
+      "step": 320
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8913605129255364e-05,
+      "loss": 0.4054,
+      "step": 321
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.890490721856751e-05,
+      "loss": 0.4292,
+      "step": 322
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8896175607842715e-05,
+      "loss": 0.4042,
+      "step": 323
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.888741031264603e-05,
+      "loss": 0.4076,
+      "step": 324
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.887861134860253e-05,
+      "loss": 0.3938,
+      "step": 325
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8869778731397336e-05,
+      "loss": 0.4382,
+      "step": 326
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8860912476775526e-05,
+      "loss": 0.3933,
+      "step": 327
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8852012600542165e-05,
+      "loss": 0.3892,
+      "step": 328
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.884307911856223e-05,
+      "loss": 0.3933,
+      "step": 329
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8834112046760624e-05,
+      "loss": 0.3929,
+      "step": 330
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.882511140112212e-05,
+      "loss": 0.4574,
+      "step": 331
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8816077197691343e-05,
+      "loss": 0.418,
+      "step": 332
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8807009452572734e-05,
+      "loss": 0.4007,
+      "step": 333
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.879790818193052e-05,
+      "loss": 0.3948,
+      "step": 334
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8788773401988714e-05,
+      "loss": 0.4243,
+      "step": 335
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.877960512903103e-05,
+      "loss": 0.4116,
+      "step": 336
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.87704033794009e-05,
+      "loss": 0.3809,
+      "step": 337
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.876116816950146e-05,
+      "loss": 0.4024,
+      "step": 338
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.875189951579545e-05,
+      "loss": 0.3869,
+      "step": 339
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.874259743480525e-05,
+      "loss": 0.4159,
+      "step": 340
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.873326194311281e-05,
+      "loss": 0.4143,
+      "step": 341
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.872389305735967e-05,
+      "loss": 0.3993,
+      "step": 342
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.871449079424689e-05,
+      "loss": 0.4197,
+      "step": 343
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8705055170534986e-05,
+      "loss": 0.4274,
+      "step": 344
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8695586203043987e-05,
+      "loss": 0.3984,
+      "step": 345
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.868608390865336e-05,
+      "loss": 0.4016,
+      "step": 346
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8676548304301966e-05,
+      "loss": 0.4137,
+      "step": 347
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.866697940698804e-05,
+      "loss": 0.423,
+      "step": 348
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.865737723376918e-05,
+      "loss": 0.4104,
+      "step": 349
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.86477418017623e-05,
+      "loss": 0.4387,
+      "step": 350
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8638073128143594e-05,
+      "loss": 0.4194,
+      "step": 351
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8628371230148506e-05,
+      "loss": 0.4162,
+      "step": 352
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.861863612507173e-05,
+      "loss": 0.4348,
+      "step": 353
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8608867830267135e-05,
+      "loss": 0.4013,
+      "step": 354
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8599066363147766e-05,
+      "loss": 0.3889,
+      "step": 355
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8589231741185784e-05,
+      "loss": 0.4047,
+      "step": 356
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.857936398191248e-05,
+      "loss": 0.4265,
+      "step": 357
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.856946310291818e-05,
+      "loss": 0.4089,
+      "step": 358
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8559529121852284e-05,
+      "loss": 0.3904,
+      "step": 359
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.854956205642318e-05,
+      "loss": 0.4358,
+      "step": 360
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.853956192439824e-05,
+      "loss": 0.4304,
+      "step": 361
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.852952874360378e-05,
+      "loss": 0.4107,
+      "step": 362
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.851946253192502e-05,
+      "loss": 0.4094,
+      "step": 363
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8509363307306074e-05,
+      "loss": 0.4124,
+      "step": 364
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.849923108774989e-05,
+      "loss": 0.3983,
+      "step": 365
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8489065891318264e-05,
+      "loss": 0.4009,
+      "step": 366
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8478867736131735e-05,
+      "loss": 0.4035,
+      "step": 367
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.846863664036962e-05,
+      "loss": 0.3922,
+      "step": 368
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.845837262226996e-05,
+      "loss": 0.4297,
+      "step": 369
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.844807570012946e-05,
+      "loss": 0.3703,
+      "step": 370
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.843774589230351e-05,
+      "loss": 0.3527,
+      "step": 371
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.842738321720609e-05,
+      "loss": 0.396,
+      "step": 372
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.841698769330981e-05,
+      "loss": 0.4086,
+      "step": 373
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8406559339145785e-05,
+      "loss": 0.4155,
+      "step": 374
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8396098173303704e-05,
+      "loss": 0.407,
+      "step": 375
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8385604214431715e-05,
+      "loss": 0.4263,
+      "step": 376
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.837507748123643e-05,
+      "loss": 0.4253,
+      "step": 377
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.836451799248289e-05,
+      "loss": 0.3868,
+      "step": 378
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.835392576699452e-05,
+      "loss": 0.3715,
+      "step": 379
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.834330082365311e-05,
+      "loss": 0.4107,
+      "step": 380
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.833264318139877e-05,
+      "loss": 0.3834,
+      "step": 381
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.832195285922988e-05,
+      "loss": 0.3987,
+      "step": 382
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8311229876203115e-05,
+      "loss": 0.3868,
+      "step": 383
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.830047425143334e-05,
+      "loss": 0.4411,
+      "step": 384
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.828968600409361e-05,
+      "loss": 0.4104,
+      "step": 385
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.827886515341515e-05,
+      "loss": 0.3876,
+      "step": 386
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.826801171868729e-05,
+      "loss": 0.3926,
+      "step": 387
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.825712571925745e-05,
+      "loss": 0.3702,
+      "step": 388
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.824620717453109e-05,
+      "loss": 0.3927,
+      "step": 389
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.82352561039717e-05,
+      "loss": 0.4058,
+      "step": 390
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.822427252710075e-05,
+      "loss": 0.4268,
+      "step": 391
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.821325646349765e-05,
+      "loss": 0.3906,
+      "step": 392
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.820220793279971e-05,
+      "loss": 0.3925,
+      "step": 393
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.819112695470215e-05,
+      "loss": 0.4048,
+      "step": 394
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.818001354895798e-05,
+      "loss": 0.4156,
+      "step": 395
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.816886773537807e-05,
+      "loss": 0.4015,
+      "step": 396
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.815768953383103e-05,
+      "loss": 0.3773,
+      "step": 397
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8146478964243196e-05,
+      "loss": 0.432,
+      "step": 398
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8135236046598636e-05,
+      "loss": 0.3992,
+      "step": 399
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.812396080093905e-05,
+      "loss": 0.3615,
+      "step": 400
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.811265324736379e-05,
+      "loss": 0.3893,
+      "step": 401
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8101313406029786e-05,
+      "loss": 0.4279,
+      "step": 402
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8089941297151526e-05,
+      "loss": 0.4092,
+      "step": 403
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.807853694100103e-05,
+      "loss": 0.3996,
+      "step": 404
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.806710035790778e-05,
+      "loss": 0.3896,
+      "step": 405
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.805563156825873e-05,
+      "loss": 0.4054,
+      "step": 406
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.804413059249823e-05,
+      "loss": 0.3894,
+      "step": 407
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8032597451128024e-05,
+      "loss": 0.3854,
+      "step": 408
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.802103216470716e-05,
+      "loss": 0.4116,
+      "step": 409
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8009434753852014e-05,
+      "loss": 0.3575,
+      "step": 410
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7997805239236234e-05,
+      "loss": 0.4189,
+      "step": 411
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.798614364159067e-05,
+      "loss": 0.4076,
+      "step": 412
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7974449981703394e-05,
+      "loss": 0.3758,
+      "step": 413
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7962724280419595e-05,
+      "loss": 0.3814,
+      "step": 414
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.795096655864162e-05,
+      "loss": 0.3825,
+      "step": 415
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.793917683732887e-05,
+      "loss": 0.4125,
+      "step": 416
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.79273551374978e-05,
+      "loss": 0.3937,
+      "step": 417
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.791550148022185e-05,
+      "loss": 0.4365,
+      "step": 418
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.790361588663146e-05,
+      "loss": 0.4013,
+      "step": 419
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.789169837791397e-05,
+      "loss": 0.419,
+      "step": 420
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.787974897531364e-05,
+      "loss": 0.3872,
+      "step": 421
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7867767700131556e-05,
+      "loss": 0.3656,
+      "step": 422
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.785575457372564e-05,
+      "loss": 0.3948,
+      "step": 423
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.784370961751059e-05,
+      "loss": 0.3718,
+      "step": 424
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.783163285295783e-05,
+      "loss": 0.4237,
+      "step": 425
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7819524301595505e-05,
+      "loss": 0.3892,
+      "step": 426
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.780738398500841e-05,
+      "loss": 0.408,
+      "step": 427
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.779521192483796e-05,
+      "loss": 0.3867,
+      "step": 428
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7783008142782185e-05,
+      "loss": 0.423,
+      "step": 429
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.777077266059563e-05,
+      "loss": 0.4313,
+      "step": 430
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.775850550008937e-05,
+      "loss": 0.3993,
+      "step": 431
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.774620668313094e-05,
+      "loss": 0.3897,
+      "step": 432
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7733876231644296e-05,
+      "loss": 0.4082,
+      "step": 433
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7721514167609815e-05,
+      "loss": 0.3717,
+      "step": 434
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7709120513064196e-05,
+      "loss": 0.3971,
+      "step": 435
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7696695290100484e-05,
+      "loss": 0.3859,
+      "step": 436
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.768423852086797e-05,
+      "loss": 0.4025,
+      "step": 437
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7671750227572177e-05,
+      "loss": 0.4078,
+      "step": 438
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.765923043247485e-05,
+      "loss": 0.3863,
+      "step": 439
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.764667915789387e-05,
+      "loss": 0.3637,
+      "step": 440
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.763409642620324e-05,
+      "loss": 0.3848,
+      "step": 441
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.762148225983303e-05,
+      "loss": 0.3797,
+      "step": 442
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.760883668126936e-05,
+      "loss": 0.3903,
+      "step": 443
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.759615971305435e-05,
+      "loss": 0.4303,
+      "step": 444
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7583451377786035e-05,
+      "loss": 0.3905,
+      "step": 445
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.757071169811841e-05,
+      "loss": 0.4095,
+      "step": 446
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.755794069676134e-05,
+      "loss": 0.4087,
+      "step": 447
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.754513839648049e-05,
+      "loss": 0.4348,
+      "step": 448
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.753230482009735e-05,
+      "loss": 0.4,
+      "step": 449
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.751943999048916e-05,
+      "loss": 0.3916,
+      "step": 450
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.750654393058887e-05,
+      "loss": 0.372,
+      "step": 451
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7493616663385085e-05,
+      "loss": 0.4157,
+      "step": 452
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.748065821192206e-05,
+      "loss": 0.3933,
+      "step": 453
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7467668599299635e-05,
+      "loss": 0.389,
+      "step": 454
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.74546478486732e-05,
+      "loss": 0.3738,
+      "step": 455
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.744159598325364e-05,
+      "loss": 0.3921,
+      "step": 456
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.74285130263073e-05,
+      "loss": 0.402,
+      "step": 457
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7415399001155995e-05,
+      "loss": 0.4058,
+      "step": 458
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7402253931176866e-05,
+      "loss": 0.3949,
+      "step": 459
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.738907783980243e-05,
+      "loss": 0.3858,
+      "step": 460
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.737587075052049e-05,
+      "loss": 0.4093,
+      "step": 461
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7362632686874095e-05,
+      "loss": 0.3941,
+      "step": 462
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.734936367246154e-05,
+      "loss": 0.3928,
+      "step": 463
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.733606373093626e-05,
+      "loss": 0.3975,
+      "step": 464
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.732273288600685e-05,
+      "loss": 0.3762,
+      "step": 465
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7309371161436964e-05,
+      "loss": 0.3914,
+      "step": 466
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.729597858104533e-05,
+      "loss": 0.4141,
+      "step": 467
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7282555168705664e-05,
+      "loss": 0.3896,
+      "step": 468
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7269100948346645e-05,
+      "loss": 0.4142,
+      "step": 469
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.725561594395188e-05,
+      "loss": 0.3938,
+      "step": 470
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7242100179559825e-05,
+      "loss": 0.3647,
+      "step": 471
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7228553679263804e-05,
+      "loss": 0.3979,
+      "step": 472
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.721497646721192e-05,
+      "loss": 0.3968,
+      "step": 473
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7201368567607e-05,
+      "loss": 0.3937,
+      "step": 474
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7187730004706615e-05,
+      "loss": 0.3757,
+      "step": 475
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.717406080282296e-05,
+      "loss": 0.4071,
+      "step": 476
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.716036098632287e-05,
+      "loss": 0.4244,
+      "step": 477
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.714663057962773e-05,
+      "loss": 0.3486,
+      "step": 478
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.71328696072135e-05,
+      "loss": 0.3823,
+      "step": 479
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.711907809361058e-05,
+      "loss": 0.3715,
+      "step": 480
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.710525606340382e-05,
+      "loss": 0.4122,
+      "step": 481
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.709140354123251e-05,
+      "loss": 0.391,
+      "step": 482
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7077520551790235e-05,
+      "loss": 0.3815,
+      "step": 483
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.706360711982494e-05,
+      "loss": 0.3999,
+      "step": 484
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.704966327013882e-05,
+      "loss": 0.4244,
+      "step": 485
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.703568902758829e-05,
+      "loss": 0.412,
+      "step": 486
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7021684417083924e-05,
+      "loss": 0.3622,
+      "step": 487
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.700764946359048e-05,
+      "loss": 0.3649,
+      "step": 488
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.699358419212676e-05,
+      "loss": 0.4059,
+      "step": 489
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6979488627765636e-05,
+      "loss": 0.448,
+      "step": 490
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.696536279563397e-05,
+      "loss": 0.3637,
+      "step": 491
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.695120672091258e-05,
+      "loss": 0.3802,
+      "step": 492
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.69370204288362e-05,
+      "loss": 0.3819,
+      "step": 493
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.692280394469343e-05,
+      "loss": 0.4026,
+      "step": 494
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.690855729382667e-05,
+      "loss": 0.3812,
+      "step": 495
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.689428050163213e-05,
+      "loss": 0.3978,
+      "step": 496
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.687997359355973e-05,
+      "loss": 0.3874,
+      "step": 497
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.686563659511307e-05,
+      "loss": 0.4132,
+      "step": 498
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.68512695318494e-05,
+      "loss": 0.3885,
+      "step": 499
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.683687242937956e-05,
+      "loss": 0.3909,
+      "step": 500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6822445313367935e-05,
+      "loss": 0.417,
+      "step": 501
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.680798820953243e-05,
+      "loss": 0.3985,
+      "step": 502
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6793501143644376e-05,
+      "loss": 0.3643,
+      "step": 503
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.677898414152854e-05,
+      "loss": 0.3954,
+      "step": 504
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6764437229063036e-05,
+      "loss": 0.3879,
+      "step": 505
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6749860432179326e-05,
+      "loss": 0.385,
+      "step": 506
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.67352537768621e-05,
+      "loss": 0.3873,
+      "step": 507
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.672061728914931e-05,
+      "loss": 0.3671,
+      "step": 508
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6705950995132064e-05,
+      "loss": 0.3999,
+      "step": 509
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6691254920954624e-05,
+      "loss": 0.3712,
+      "step": 510
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.667652909281432e-05,
+      "loss": 0.3929,
+      "step": 511
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.666177353696154e-05,
+      "loss": 0.3595,
+      "step": 512
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.664698827969965e-05,
+      "loss": 0.3781,
+      "step": 513
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.663217334738495e-05,
+      "loss": 0.3895,
+      "step": 514
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6617328766426686e-05,
+      "loss": 0.3992,
+      "step": 515
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6602454563286906e-05,
+      "loss": 0.3794,
+      "step": 516
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.658755076448048e-05,
+      "loss": 0.3967,
+      "step": 517
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.657261739657505e-05,
+      "loss": 0.3498,
+      "step": 518
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.655765448619094e-05,
+      "loss": 0.4164,
+      "step": 519
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.654266206000116e-05,
+      "loss": 0.3896,
+      "step": 520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.652764014473133e-05,
+      "loss": 0.3581,
+      "step": 521
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.651258876715964e-05,
+      "loss": 0.3815,
+      "step": 522
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.649750795411677e-05,
+      "loss": 0.3766,
+      "step": 523
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.648239773248593e-05,
+      "loss": 0.3755,
+      "step": 524
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.646725812920269e-05,
+      "loss": 0.4137,
+      "step": 525
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.645208917125505e-05,
+      "loss": 0.374,
+      "step": 526
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.643689088568331e-05,
+      "loss": 0.3983,
+      "step": 527
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.642166329958003e-05,
+      "loss": 0.3856,
+      "step": 528
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6406406440090056e-05,
+      "loss": 0.3955,
+      "step": 529
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6391120334410375e-05,
+      "loss": 0.4118,
+      "step": 530
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.637580500979012e-05,
+      "loss": 0.4246,
+      "step": 531
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6360460493530505e-05,
+      "loss": 0.3807,
+      "step": 532
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6345086812984785e-05,
+      "loss": 0.3806,
+      "step": 533
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6329683995558215e-05,
+      "loss": 0.3965,
+      "step": 534
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6314252068707966e-05,
+      "loss": 0.4035,
+      "step": 535
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6298791059943133e-05,
+      "loss": 0.3862,
+      "step": 536
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6283300996824604e-05,
+      "loss": 0.3801,
+      "step": 537
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.626778190696511e-05,
+      "loss": 0.3699,
+      "step": 538
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.625223381802909e-05,
+      "loss": 0.3596,
+      "step": 539
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6236656757732694e-05,
+      "loss": 0.4144,
+      "step": 540
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6221050753843725e-05,
+      "loss": 0.3745,
+      "step": 541
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.620541583418155e-05,
+      "loss": 0.3691,
+      "step": 542
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6189752026617115e-05,
+      "loss": 0.3788,
+      "step": 543
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6174059359072836e-05,
+      "loss": 0.3792,
+      "step": 544
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6158337859522594e-05,
+      "loss": 0.3779,
+      "step": 545
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6142587555991654e-05,
+      "loss": 0.3825,
+      "step": 546
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.612680847655663e-05,
+      "loss": 0.3714,
+      "step": 547
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.611100064934544e-05,
+      "loss": 0.3783,
+      "step": 548
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.609516410253723e-05,
+      "loss": 0.4128,
+      "step": 549
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6079298864362374e-05,
+      "loss": 0.391,
+      "step": 550
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.606340496310234e-05,
+      "loss": 0.3885,
+      "step": 551
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6047482427089754e-05,
+      "loss": 0.4198,
+      "step": 552
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6031531284708245e-05,
+      "loss": 0.3922,
+      "step": 553
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.601555156439243e-05,
+      "loss": 0.3787,
+      "step": 554
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.59995432946279e-05,
+      "loss": 0.3875,
+      "step": 555
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.598350650395112e-05,
+      "loss": 0.3554,
+      "step": 556
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5967441220949394e-05,
+      "loss": 0.338,
+      "step": 557
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5951347474260826e-05,
+      "loss": 0.3891,
+      "step": 558
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.593522529257427e-05,
+      "loss": 0.3845,
+      "step": 559
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.591907470462923e-05,
+      "loss": 0.3796,
+      "step": 560
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.590289573921588e-05,
+      "loss": 0.3772,
+      "step": 561
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.588668842517498e-05,
+      "loss": 0.3598,
+      "step": 562
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.587045279139781e-05,
+      "loss": 0.3584,
+      "step": 563
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5854188866826136e-05,
+      "loss": 0.3843,
+      "step": 564
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.583789668045216e-05,
+      "loss": 0.3857,
+      "step": 565
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.582157626131848e-05,
+      "loss": 0.3752,
+      "step": 566
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.580522763851797e-05,
+      "loss": 0.3714,
+      "step": 567
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5788850841193836e-05,
+      "loss": 0.357,
+      "step": 568
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.577244589853948e-05,
+      "loss": 0.386,
+      "step": 569
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5756012839798494e-05,
+      "loss": 0.4127,
+      "step": 570
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.573955169426457e-05,
+      "loss": 0.3587,
+      "step": 571
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.572306249128147e-05,
+      "loss": 0.4074,
+      "step": 572
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.570654526024298e-05,
+      "loss": 0.36,
+      "step": 573
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.569000003059286e-05,
+      "loss": 0.3472,
+      "step": 574
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.567342683182474e-05,
+      "loss": 0.3818,
+      "step": 575
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.565682569348216e-05,
+      "loss": 0.359,
+      "step": 576
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.564019664515843e-05,
+      "loss": 0.3886,
+      "step": 577
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.562353971649662e-05,
+      "loss": 0.375,
+      "step": 578
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5606854937189515e-05,
+      "loss": 0.362,
+      "step": 579
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.559014233697952e-05,
+      "loss": 0.3744,
+      "step": 580
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.557340194565866e-05,
+      "loss": 0.4042,
+      "step": 581
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5556633793068486e-05,
+      "loss": 0.3592,
+      "step": 582
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5539837909100035e-05,
+      "loss": 0.3709,
+      "step": 583
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.552301432369378e-05,
+      "loss": 0.39,
+      "step": 584
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.550616306683958e-05,
+      "loss": 0.3622,
+      "step": 585
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5489284168576614e-05,
+      "loss": 0.3717,
+      "step": 586
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.547237765899336e-05,
+      "loss": 0.3391,
+      "step": 587
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.545544356822747e-05,
+      "loss": 0.339,
+      "step": 588
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5438481926465804e-05,
+      "loss": 0.3708,
+      "step": 589
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.542149276394431e-05,
+      "loss": 0.3866,
+      "step": 590
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.540447611094801e-05,
+      "loss": 0.4041,
+      "step": 591
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5387431997810915e-05,
+      "loss": 0.3889,
+      "step": 592
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5370360454916005e-05,
+      "loss": 0.3654,
+      "step": 593
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.535326151269515e-05,
+      "loss": 0.365,
+      "step": 594
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.533613520162904e-05,
+      "loss": 0.3802,
+      "step": 595
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.53189815522472e-05,
+      "loss": 0.3816,
+      "step": 596
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.530180059512783e-05,
+      "loss": 0.3742,
+      "step": 597
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.528459236089787e-05,
+      "loss": 0.3832,
+      "step": 598
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.526735688023284e-05,
+      "loss": 0.366,
+      "step": 599
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.525009418385683e-05,
+      "loss": 0.3839,
+      "step": 600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.523280430254248e-05,
+      "loss": 0.3994,
+      "step": 601
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5215487267110864e-05,
+      "loss": 0.3585,
+      "step": 602
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.519814310843146e-05,
+      "loss": 0.3672,
+      "step": 603
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.518077185742212e-05,
+      "loss": 0.389,
+      "step": 604
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.516337354504895e-05,
+      "loss": 0.3789,
+      "step": 605
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.514594820232633e-05,
+      "loss": 0.3766,
+      "step": 606
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.512849586031681e-05,
+      "loss": 0.3612,
+      "step": 607
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.511101655013108e-05,
+      "loss": 0.3628,
+      "step": 608
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.509351030292789e-05,
+      "loss": 0.3949,
+      "step": 609
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.507597714991401e-05,
+      "loss": 0.3766,
+      "step": 610
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.505841712234419e-05,
+      "loss": 0.3682,
+      "step": 611
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.504083025152107e-05,
+      "loss": 0.3562,
+      "step": 612
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.502321656879513e-05,
+      "loss": 0.4008,
+      "step": 613
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.500557610556468e-05,
+      "loss": 0.3332,
+      "step": 614
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.498790889327574e-05,
+      "loss": 0.3605,
+      "step": 615
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.497021496342203e-05,
+      "loss": 0.3873,
+      "step": 616
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.4952494347544874e-05,
+      "loss": 0.3577,
+      "step": 617
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.493474707723319e-05,
+      "loss": 0.3334,
+      "step": 618
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4916973184123405e-05,
+      "loss": 0.3667,
+      "step": 619
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.489917269989939e-05,
+      "loss": 0.3672,
+      "step": 620
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.488134565629245e-05,
+      "loss": 0.3994,
+      "step": 621
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4863492085081194e-05,
+      "loss": 0.4052,
+      "step": 622
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.484561201809155e-05,
+      "loss": 0.3668,
+      "step": 623
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.482770548719666e-05,
+      "loss": 0.3647,
+      "step": 624
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.480977252431685e-05,
+      "loss": 0.3588,
+      "step": 625
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.479181316141956e-05,
+      "loss": 0.395,
+      "step": 626
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.47738274305193e-05,
+      "loss": 0.3811,
+      "step": 627
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.475581536367756e-05,
+      "loss": 0.3551,
+      "step": 628
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.473777699300281e-05,
+      "loss": 0.3557,
+      "step": 629
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.471971235065038e-05,
+      "loss": 0.3695,
+      "step": 630
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.470162146882245e-05,
+      "loss": 0.3632,
+      "step": 631
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4683504379767966e-05,
+      "loss": 0.36,
+      "step": 632
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4665361115782596e-05,
+      "loss": 0.3443,
+      "step": 633
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.464719170920867e-05,
+      "loss": 0.3577,
+      "step": 634
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4628996192435104e-05,
+      "loss": 0.377,
+      "step": 635
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4610774597897375e-05,
+      "loss": 0.3836,
+      "step": 636
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.459252695807745e-05,
+      "loss": 0.3767,
+      "step": 637
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.45742533055037e-05,
+      "loss": 0.3815,
+      "step": 638
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.45559536727509e-05,
+      "loss": 0.3799,
+      "step": 639
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.453762809244011e-05,
+      "loss": 0.3779,
+      "step": 640
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.451927659723865e-05,
+      "loss": 0.3924,
+      "step": 641
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4500899219860043e-05,
+      "loss": 0.3935,
+      "step": 642
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.448249599306396e-05,
+      "loss": 0.3805,
+      "step": 643
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.446406694965612e-05,
+      "loss": 0.3504,
+      "step": 644
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4445612122488297e-05,
+      "loss": 0.3968,
+      "step": 645
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4427131544458195e-05,
+      "loss": 0.377,
+      "step": 646
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4408625248509454e-05,
+      "loss": 0.3722,
+      "step": 647
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4390093267631534e-05,
+      "loss": 0.3911,
+      "step": 648
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4371535634859695e-05,
+      "loss": 0.3911,
+      "step": 649
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.435295238327491e-05,
+      "loss": 0.3856,
+      "step": 650
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.433434354600385e-05,
+      "loss": 0.3853,
+      "step": 651
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.431570915621876e-05,
+      "loss": 0.3688,
+      "step": 652
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.429704924713745e-05,
+      "loss": 0.3601,
+      "step": 653
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.427836385202323e-05,
+      "loss": 0.3859,
+      "step": 654
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.425965300418482e-05,
+      "loss": 0.3875,
+      "step": 655
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.424091673697635e-05,
+      "loss": 0.3848,
+      "step": 656
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.42221550837972e-05,
+      "loss": 0.3827,
+      "step": 657
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.420336807809208e-05,
+      "loss": 0.3624,
+      "step": 658
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.418455575335083e-05,
+      "loss": 0.3824,
+      "step": 659
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.416571814310845e-05,
+      "loss": 0.3746,
+      "step": 660
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.414685528094502e-05,
+      "loss": 0.3216,
+      "step": 661
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.412796720048564e-05,
+      "loss": 0.3556,
+      "step": 662
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.410905393540032e-05,
+      "loss": 0.3577,
+      "step": 663
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4090115519404014e-05,
+      "loss": 0.3183,
+      "step": 664
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.40711519862565e-05,
+      "loss": 0.3853,
+      "step": 665
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4052163369762304e-05,
+      "loss": 0.3827,
+      "step": 666
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.40331497037707e-05,
+      "loss": 0.3545,
+      "step": 667
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.401411102217559e-05,
+      "loss": 0.3607,
+      "step": 668
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.399504735891548e-05,
+      "loss": 0.366,
+      "step": 669
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.39759587479734e-05,
+      "loss": 0.3645,
+      "step": 670
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.395684522337686e-05,
+      "loss": 0.3696,
+      "step": 671
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3937706819197807e-05,
+      "loss": 0.3998,
+      "step": 672
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.391854356955247e-05,
+      "loss": 0.3678,
+      "step": 673
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.389935550860144e-05,
+      "loss": 0.3643,
+      "step": 674
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.388014267054949e-05,
+      "loss": 0.3269,
+      "step": 675
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.386090508964558e-05,
+      "loss": 0.3659,
+      "step": 676
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3841642800182775e-05,
+      "loss": 0.3572,
+      "step": 677
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.382235583649819e-05,
+      "loss": 0.3709,
+      "step": 678
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3803044232972904e-05,
+      "loss": 0.3409,
+      "step": 679
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.378370802403195e-05,
+      "loss": 0.3414,
+      "step": 680
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.37643472441442e-05,
+      "loss": 0.3827,
+      "step": 681
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3744961927822336e-05,
+      "loss": 0.3838,
+      "step": 682
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.372555210962278e-05,
+      "loss": 0.3479,
+      "step": 683
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.37061178241456e-05,
+      "loss": 0.369,
+      "step": 684
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.368665910603454e-05,
+      "loss": 0.3441,
+      "step": 685
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.366717598997685e-05,
+      "loss": 0.3717,
+      "step": 686
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.364766851070329e-05,
+      "loss": 0.3553,
+      "step": 687
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3628136702988045e-05,
+      "loss": 0.3744,
+      "step": 688
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.360858060164867e-05,
+      "loss": 0.3458,
+      "step": 689
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3589000241546014e-05,
+      "loss": 0.3602,
+      "step": 690
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.356939565758421e-05,
+      "loss": 0.3639,
+      "step": 691
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.354976688471052e-05,
+      "loss": 0.357,
+      "step": 692
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3530113957915355e-05,
+      "loss": 0.3812,
+      "step": 693
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3510436912232186e-05,
+      "loss": 0.3783,
+      "step": 694
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.349073578273745e-05,
+      "loss": 0.3742,
+      "step": 695
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3471010604550555e-05,
+      "loss": 0.3726,
+      "step": 696
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3451261412833765e-05,
+      "loss": 0.4058,
+      "step": 697
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.343148824279212e-05,
+      "loss": 0.3428,
+      "step": 698
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.341169112967346e-05,
+      "loss": 0.3165,
+      "step": 699
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.339187010876824e-05,
+      "loss": 0.3551,
+      "step": 700
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.337202521540959e-05,
+      "loss": 0.3397,
+      "step": 701
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.335215648497318e-05,
+      "loss": 0.374,
+      "step": 702
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.333226395287715e-05,
+      "loss": 0.3647,
+      "step": 703
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.33123476545821e-05,
+      "loss": 0.3808,
+      "step": 704
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.329240762559097e-05,
+      "loss": 0.336,
+      "step": 705
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3272443901449e-05,
+      "loss": 0.3633,
+      "step": 706
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.325245651774371e-05,
+      "loss": 0.3655,
+      "step": 707
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3232445510104756e-05,
+      "loss": 0.3738,
+      "step": 708
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3212410914203915e-05,
+      "loss": 0.359,
+      "step": 709
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.319235276575501e-05,
+      "loss": 0.3843,
+      "step": 710
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.317227110051388e-05,
+      "loss": 0.3826,
+      "step": 711
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.315216595427823e-05,
+      "loss": 0.339,
+      "step": 712
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.313203736288767e-05,
+      "loss": 0.3897,
+      "step": 713
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.311188536222359e-05,
+      "loss": 0.3628,
+      "step": 714
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.309170998820908e-05,
+      "loss": 0.342,
+      "step": 715
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.307151127680896e-05,
+      "loss": 0.3438,
+      "step": 716
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.305128926402957e-05,
+      "loss": 0.3924,
+      "step": 717
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.303104398591886e-05,
+      "loss": 0.4017,
+      "step": 718
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.301077547856622e-05,
+      "loss": 0.355,
+      "step": 719
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.299048377810244e-05,
+      "loss": 0.3295,
+      "step": 720
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.297016892069967e-05,
+      "loss": 0.3881,
+      "step": 721
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2949830942571335e-05,
+      "loss": 0.3472,
+      "step": 722
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.292946987997208e-05,
+      "loss": 0.3819,
+      "step": 723
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.290908576919768e-05,
+      "loss": 0.3318,
+      "step": 724
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.288867864658503e-05,
+      "loss": 0.3485,
+      "step": 725
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2868248548512024e-05,
+      "loss": 0.3462,
+      "step": 726
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.284779551139751e-05,
+      "loss": 0.3793,
+      "step": 727
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.282731957170125e-05,
+      "loss": 0.3699,
+      "step": 728
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.280682076592379e-05,
+      "loss": 0.3587,
+      "step": 729
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.278629913060648e-05,
+      "loss": 0.3552,
+      "step": 730
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2765754702331346e-05,
+      "loss": 0.3427,
+      "step": 731
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2745187517721056e-05,
+      "loss": 0.3575,
+      "step": 732
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.272459761343882e-05,
+      "loss": 0.34,
+      "step": 733
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2703985026188364e-05,
+      "loss": 0.3465,
+      "step": 734
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.268334979271387e-05,
+      "loss": 0.3614,
+      "step": 735
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.266269194979985e-05,
+      "loss": 0.3618,
+      "step": 736
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.264201153427114e-05,
+      "loss": 0.3518,
+      "step": 737
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2621308582992815e-05,
+      "loss": 0.34,
+      "step": 738
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.260058313287012e-05,
+      "loss": 0.3773,
+      "step": 739
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.257983522084841e-05,
+      "loss": 0.3352,
+      "step": 740
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2559064883913066e-05,
+      "loss": 0.3365,
+      "step": 741
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.253827215908946e-05,
+      "loss": 0.3547,
+      "step": 742
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.251745708344287e-05,
+      "loss": 0.36,
+      "step": 743
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2496619694078416e-05,
+      "loss": 0.3579,
+      "step": 744
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2475760028140985e-05,
+      "loss": 0.3299,
+      "step": 745
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.245487812281518e-05,
+      "loss": 0.3468,
+      "step": 746
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2433974015325276e-05,
+      "loss": 0.3071,
+      "step": 747
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.241304774293506e-05,
+      "loss": 0.3521,
+      "step": 748
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.239209934294791e-05,
+      "loss": 0.3425,
+      "step": 749
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.237112885270658e-05,
+      "loss": 0.3185,
+      "step": 750
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2350136309593245e-05,
+      "loss": 0.3562,
+      "step": 751
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.232912175102938e-05,
+      "loss": 0.3207,
+      "step": 752
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.230808521447568e-05,
+      "loss": 0.385,
+      "step": 753
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2287026737432083e-05,
+      "loss": 0.3571,
+      "step": 754
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.226594635743756e-05,
+      "loss": 0.3397,
+      "step": 755
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.224484411207018e-05,
+      "loss": 0.37,
+      "step": 756
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.222372003894697e-05,
+      "loss": 0.3287,
+      "step": 757
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.220257417572387e-05,
+      "loss": 0.3397,
+      "step": 758
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.218140656009565e-05,
+      "loss": 0.351,
+      "step": 759
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.216021722979589e-05,
+      "loss": 0.3638,
+      "step": 760
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2139006222596845e-05,
+      "loss": 0.3704,
+      "step": 761
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2117773576309425e-05,
+      "loss": 0.3754,
+      "step": 762
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2096519328783115e-05,
+      "loss": 0.3783,
+      "step": 763
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.20752435179059e-05,
+      "loss": 0.3384,
+      "step": 764
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.205394618160422e-05,
+      "loss": 0.3435,
+      "step": 765
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.203262735784286e-05,
+      "loss": 0.3415,
+      "step": 766
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.201128708462493e-05,
+      "loss": 0.3486,
+      "step": 767
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.198992539999178e-05,
+      "loss": 0.365,
+      "step": 768
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.19685423420229e-05,
+      "loss": 0.3724,
+      "step": 769
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.194713794883591e-05,
+      "loss": 0.3498,
+      "step": 770
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1925712258586444e-05,
+      "loss": 0.3675,
+      "step": 771
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1904265309468105e-05,
+      "loss": 0.3405,
+      "step": 772
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.188279713971241e-05,
+      "loss": 0.3881,
+      "step": 773
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.186130778758867e-05,
+      "loss": 0.3268,
+      "step": 774
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.183979729140398e-05,
+      "loss": 0.3404,
+      "step": 775
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.181826568950312e-05,
+      "loss": 0.3891,
+      "step": 776
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.179671302026849e-05,
+      "loss": 0.3421,
+      "step": 777
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.177513932212006e-05,
+      "loss": 0.3518,
+      "step": 778
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.175354463351526e-05,
+      "loss": 0.3342,
+      "step": 779
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1731928992948965e-05,
+      "loss": 0.3194,
+      "step": 780
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1710292438953376e-05,
+      "loss": 0.3679,
+      "step": 781
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.168863501009799e-05,
+      "loss": 0.3443,
+      "step": 782
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.166695674498952e-05,
+      "loss": 0.3654,
+      "step": 783
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1645257682271794e-05,
+      "loss": 0.3568,
+      "step": 784
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.162353786062575e-05,
+      "loss": 0.3698,
+      "step": 785
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.16017973187693e-05,
+      "loss": 0.3584,
+      "step": 786
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.158003609545731e-05,
+      "loss": 0.3495,
+      "step": 787
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1558254229481513e-05,
+      "loss": 0.3745,
+      "step": 788
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.153645175967043e-05,
+      "loss": 0.3284,
+      "step": 789
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.151462872488933e-05,
+      "loss": 0.3424,
+      "step": 790
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.14927851640401e-05,
+      "loss": 0.3448,
+      "step": 791
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.147092111606126e-05,
+      "loss": 0.344,
+      "step": 792
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1449036619927844e-05,
+      "loss": 0.3675,
+      "step": 793
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1427131714651314e-05,
+      "loss": 0.3798,
+      "step": 794
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1405206439279525e-05,
+      "loss": 0.3439,
+      "step": 795
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1383260832896664e-05,
+      "loss": 0.3269,
+      "step": 796
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.136129493462312e-05,
+      "loss": 0.3572,
+      "step": 797
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.133930878361549e-05,
+      "loss": 0.3486,
+      "step": 798
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.131730241906645e-05,
+      "loss": 0.3889,
+      "step": 799
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.129527588020474e-05,
+      "loss": 0.3594,
+      "step": 800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.127322920629503e-05,
+      "loss": 0.338,
+      "step": 801
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.125116243663788e-05,
+      "loss": 0.3495,
+      "step": 802
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.122907561056971e-05,
+      "loss": 0.3412,
+      "step": 803
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.120696876746266e-05,
+      "loss": 0.3525,
+      "step": 804
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.118484194672456e-05,
+      "loss": 0.3739,
+      "step": 805
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.116269518779888e-05,
+      "loss": 0.3562,
+      "step": 806
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.114052853016458e-05,
+      "loss": 0.3269,
+      "step": 807
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.111834201333611e-05,
+      "loss": 0.3635,
+      "step": 808
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.109613567686337e-05,
+      "loss": 0.3315,
+      "step": 809
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1073909560331524e-05,
+      "loss": 0.3576,
+      "step": 810
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.105166370336102e-05,
+      "loss": 0.3454,
+      "step": 811
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.102939814560751e-05,
+      "loss": 0.3582,
+      "step": 812
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.100711292676174e-05,
+      "loss": 0.3525,
+      "step": 813
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.098480808654954e-05,
+      "loss": 0.3464,
+      "step": 814
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.096248366473167e-05,
+      "loss": 0.3448,
+      "step": 815
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.094013970110383e-05,
+      "loss": 0.3243,
+      "step": 816
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.091777623549655e-05,
+      "loss": 0.3377,
+      "step": 817
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0895393307775123e-05,
+      "loss": 0.3788,
+      "step": 818
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0872990957839514e-05,
+      "loss": 0.3151,
+      "step": 819
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.085056922562436e-05,
+      "loss": 0.3757,
+      "step": 820
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0828128151098786e-05,
+      "loss": 0.3425,
+      "step": 821
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0805667774266446e-05,
+      "loss": 0.3585,
+      "step": 822
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0783188135165385e-05,
+      "loss": 0.3552,
+      "step": 823
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.076068927386797e-05,
+      "loss": 0.3652,
+      "step": 824
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0738171230480874e-05,
+      "loss": 0.3612,
+      "step": 825
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.071563404514492e-05,
+      "loss": 0.3175,
+      "step": 826
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.069307775803508e-05,
+      "loss": 0.3771,
+      "step": 827
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0670502409360374e-05,
+      "loss": 0.3623,
+      "step": 828
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0647908039363786e-05,
+      "loss": 0.3325,
+      "step": 829
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.062529468832222e-05,
+      "loss": 0.3189,
+      "step": 830
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0602662396546416e-05,
+      "loss": 0.3599,
+      "step": 831
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0580011204380883e-05,
+      "loss": 0.3337,
+      "step": 832
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.055734115220381e-05,
+      "loss": 0.3427,
+      "step": 833
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.053465228042701e-05,
+      "loss": 0.3199,
+      "step": 834
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0511944629495844e-05,
+      "loss": 0.3269,
+      "step": 835
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.048921823988915e-05,
+      "loss": 0.3573,
+      "step": 836
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0466473152119172e-05,
+      "loss": 0.3488,
+      "step": 837
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.044370940673148e-05,
+      "loss": 0.3626,
+      "step": 838
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0420927044304918e-05,
+      "loss": 0.3085,
+      "step": 839
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0398126105451493e-05,
+      "loss": 0.3718,
+      "step": 840
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0375306630816332e-05,
+      "loss": 0.3427,
+      "step": 841
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0352468661077635e-05,
+      "loss": 0.3455,
+      "step": 842
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0329612236946518e-05,
+      "loss": 0.3159,
+      "step": 843
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0306737399167045e-05,
+      "loss": 0.352,
+      "step": 844
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.028384418851608e-05,
+      "loss": 0.3592,
+      "step": 845
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0260932645803226e-05,
+      "loss": 0.3254,
+      "step": 846
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0238002811870797e-05,
+      "loss": 0.3828,
+      "step": 847
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0215054727593687e-05,
+      "loss": 0.3449,
+      "step": 848
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0192088433879338e-05,
+      "loss": 0.3196,
+      "step": 849
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0169103971667642e-05,
+      "loss": 0.3308,
+      "step": 850
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.014610138193088e-05,
+      "loss": 0.3607,
+      "step": 851
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.012308070567366e-05,
+      "loss": 0.332,
+      "step": 852
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.010004198393281e-05,
+      "loss": 0.3372,
+      "step": 853
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.007698525777735e-05,
+      "loss": 0.3137,
+      "step": 854
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0053910568308375e-05,
+      "loss": 0.3657,
+      "step": 855
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0030817956659004e-05,
+      "loss": 0.3435,
+      "step": 856
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0007707463994315e-05,
+      "loss": 0.3444,
+      "step": 857
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9984579131511257e-05,
+      "loss": 0.3725,
+      "step": 858
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.996143300043857e-05,
+      "loss": 0.3188,
+      "step": 859
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.993826911203674e-05,
+      "loss": 0.3227,
+      "step": 860
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.991508750759789e-05,
+      "loss": 0.3303,
+      "step": 861
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9891888228445733e-05,
+      "loss": 0.3455,
+      "step": 862
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.986867131593548e-05,
+      "loss": 0.3288,
+      "step": 863
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.98454368114538e-05,
+      "loss": 0.3138,
+      "step": 864
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9822184756418698e-05,
+      "loss": 0.338,
+      "step": 865
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9798915192279455e-05,
+      "loss": 0.3768,
+      "step": 866
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9775628160516606e-05,
+      "loss": 0.3831,
+      "step": 867
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9752323702641774e-05,
+      "loss": 0.3581,
+      "step": 868
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9729001860197688e-05,
+      "loss": 0.3354,
+      "step": 869
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9705662674758032e-05,
+      "loss": 0.3445,
+      "step": 870
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.968230618792743e-05,
+      "loss": 0.3272,
+      "step": 871
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9658932441341353e-05,
+      "loss": 0.3597,
+      "step": 872
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9635541476666002e-05,
+      "loss": 0.3252,
+      "step": 873
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.961213333559831e-05,
+      "loss": 0.3485,
+      "step": 874
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9588708059865814e-05,
+      "loss": 0.3823,
+      "step": 875
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9565265691226594e-05,
+      "loss": 0.3088,
+      "step": 876
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.95418062714692e-05,
+      "loss": 0.3397,
+      "step": 877
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9518329842412573e-05,
+      "loss": 0.2965,
+      "step": 878
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9494836445905993e-05,
+      "loss": 0.3211,
+      "step": 879
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9471326123828964e-05,
+      "loss": 0.3097,
+      "step": 880
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.944779891809118e-05,
+      "loss": 0.3491,
+      "step": 881
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.942425487063242e-05,
+      "loss": 0.3666,
+      "step": 882
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.940069402342249e-05,
+      "loss": 0.3243,
+      "step": 883
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.937711641846114e-05,
+      "loss": 0.3806,
+      "step": 884
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9353522097777997e-05,
+      "loss": 0.33,
+      "step": 885
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.932991110343248e-05,
+      "loss": 0.3257,
+      "step": 886
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9306283477513738e-05,
+      "loss": 0.3718,
+      "step": 887
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9282639262140563e-05,
+      "loss": 0.3624,
+      "step": 888
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9258978499461318e-05,
+      "loss": 0.3509,
+      "step": 889
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.923530123165387e-05,
+      "loss": 0.3371,
+      "step": 890
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9211607500925496e-05,
+      "loss": 0.3172,
+      "step": 891
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9187897349512835e-05,
+      "loss": 0.3646,
+      "step": 892
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.916417081968178e-05,
+      "loss": 0.3552,
+      "step": 893
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.914042795372745e-05,
+      "loss": 0.3198,
+      "step": 894
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9116668793974042e-05,
+      "loss": 0.3378,
+      "step": 895
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9092893382774837e-05,
+      "loss": 0.3384,
+      "step": 896
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.906910176251207e-05,
+      "loss": 0.3556,
+      "step": 897
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.904529397559686e-05,
+      "loss": 0.3405,
+      "step": 898
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9021470064469173e-05,
+      "loss": 0.3321,
+      "step": 899
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8997630071597683e-05,
+      "loss": 0.3452,
+      "step": 900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8973774039479764e-05,
+      "loss": 0.2981,
+      "step": 901
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8949902010641367e-05,
+      "loss": 0.346,
+      "step": 902
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8926014027636942e-05,
+      "loss": 0.3072,
+      "step": 903
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.890211013304941e-05,
+      "loss": 0.359,
+      "step": 904
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.887819036949005e-05,
+      "loss": 0.3435,
+      "step": 905
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8854254779598405e-05,
+      "loss": 0.3505,
+      "step": 906
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8830303406042254e-05,
+      "loss": 0.2695,
+      "step": 907
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8806336291517506e-05,
+      "loss": 0.3394,
+      "step": 908
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8782353478748123e-05,
+      "loss": 0.3387,
+      "step": 909
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8758355010486067e-05,
+      "loss": 0.3334,
+      "step": 910
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8734340929511188e-05,
+      "loss": 0.3253,
+      "step": 911
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8710311278631175e-05,
+      "loss": 0.3132,
+      "step": 912
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.868626610068148e-05,
+      "loss": 0.3447,
+      "step": 913
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8662205438525225e-05,
+      "loss": 0.3592,
+      "step": 914
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.863812933505313e-05,
+      "loss": 0.3043,
+      "step": 915
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8614037833183463e-05,
+      "loss": 0.3639,
+      "step": 916
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.858993097586191e-05,
+      "loss": 0.3225,
+      "step": 917
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8565808806061546e-05,
+      "loss": 0.3615,
+      "step": 918
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8541671366782755e-05,
+      "loss": 0.3411,
+      "step": 919
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8517518701053112e-05,
+      "loss": 0.3244,
+      "step": 920
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8493350851927365e-05,
+      "loss": 0.3699,
+      "step": 921
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8469167862487295e-05,
+      "loss": 0.3529,
+      "step": 922
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8444969775841696e-05,
+      "loss": 0.3539,
+      "step": 923
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.842075663512628e-05,
+      "loss": 0.3186,
+      "step": 924
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.839652848350357e-05,
+      "loss": 0.3329,
+      "step": 925
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8372285364162864e-05,
+      "loss": 0.3422,
+      "step": 926
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8348027320320143e-05,
+      "loss": 0.3512,
+      "step": 927
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8323754395217973e-05,
+      "loss": 0.3452,
+      "step": 928
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.829946663212548e-05,
+      "loss": 0.3342,
+      "step": 929
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8275164074338208e-05,
+      "loss": 0.3263,
+      "step": 930
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.82508467651781e-05,
+      "loss": 0.3663,
+      "step": 931
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.822651474799338e-05,
+      "loss": 0.3398,
+      "step": 932
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8202168066158482e-05,
+      "loss": 0.3231,
+      "step": 933
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8177806763074014e-05,
+      "loss": 0.3199,
+      "step": 934
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8153430882166613e-05,
+      "loss": 0.3504,
+      "step": 935
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.812904046688892e-05,
+      "loss": 0.3562,
+      "step": 936
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.810463556071948e-05,
+      "loss": 0.3595,
+      "step": 937
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.808021620716268e-05,
+      "loss": 0.3727,
+      "step": 938
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8055782449748648e-05,
+      "loss": 0.3109,
+      "step": 939
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8031334332033185e-05,
+      "loss": 0.3666,
+      "step": 940
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8006871897597706e-05,
+      "loss": 0.3364,
+      "step": 941
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7982395190049146e-05,
+      "loss": 0.3377,
+      "step": 942
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.795790425301986e-05,
+      "loss": 0.3539,
+      "step": 943
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7933399130167605e-05,
+      "loss": 0.3331,
+      "step": 944
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7908879865175386e-05,
+      "loss": 0.3239,
+      "step": 945
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7884346501751465e-05,
+      "loss": 0.35,
+      "step": 946
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7859799083629187e-05,
+      "loss": 0.3133,
+      "step": 947
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.783523765456698e-05,
+      "loss": 0.3182,
+      "step": 948
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7810662258348248e-05,
+      "loss": 0.3599,
+      "step": 949
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7786072938781274e-05,
+      "loss": 0.3342,
+      "step": 950
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7761469739699186e-05,
+      "loss": 0.3486,
+      "step": 951
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7736852704959827e-05,
+      "loss": 0.3398,
+      "step": 952
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.771222187844573e-05,
+      "loss": 0.3634,
+      "step": 953
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7687577304064004e-05,
+      "loss": 0.3311,
+      "step": 954
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.766291902574624e-05,
+      "loss": 0.3273,
+      "step": 955
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7638247087448506e-05,
+      "loss": 0.3796,
+      "step": 956
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.761356153315118e-05,
+      "loss": 0.3446,
+      "step": 957
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7588862406858922e-05,
+      "loss": 0.3326,
+      "step": 958
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7564149752600605e-05,
+      "loss": 0.316,
+      "step": 959
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7539423614429188e-05,
+      "loss": 0.3383,
+      "step": 960
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7514684036421684e-05,
+      "loss": 0.3897,
+      "step": 961
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7489931062679063e-05,
+      "loss": 0.3186,
+      "step": 962
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7465164737326172e-05,
+      "loss": 0.3502,
+      "step": 963
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.744038510451165e-05,
+      "loss": 0.3546,
+      "step": 964
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7415592208407876e-05,
+      "loss": 0.3208,
+      "step": 965
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.739078609321086e-05,
+      "loss": 0.3319,
+      "step": 966
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7365966803140173e-05,
+      "loss": 0.3421,
+      "step": 967
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7341134382438884e-05,
+      "loss": 0.3294,
+      "step": 968
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.731628887537346e-05,
+      "loss": 0.3231,
+      "step": 969
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7291430326233692e-05,
+      "loss": 0.3279,
+      "step": 970
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7266558779332642e-05,
+      "loss": 0.319,
+      "step": 971
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7241674279006506e-05,
+      "loss": 0.3277,
+      "step": 972
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7216776869614613e-05,
+      "loss": 0.3236,
+      "step": 973
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7191866595539263e-05,
+      "loss": 0.3413,
+      "step": 974
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7166943501185712e-05,
+      "loss": 0.3338,
+      "step": 975
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7142007630982074e-05,
+      "loss": 0.3397,
+      "step": 976
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7117059029379212e-05,
+      "loss": 0.3396,
+      "step": 977
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7092097740850712e-05,
+      "loss": 0.3282,
+      "step": 978
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.706712380989277e-05,
+      "loss": 0.3127,
+      "step": 979
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.704213728102409e-05,
+      "loss": 0.326,
+      "step": 980
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7017138198785883e-05,
+      "loss": 0.3391,
+      "step": 981
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6992126607741687e-05,
+      "loss": 0.3387,
+      "step": 982
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.696710255247738e-05,
+      "loss": 0.3309,
+      "step": 983
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6942066077601023e-05,
+      "loss": 0.3352,
+      "step": 984
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6917017227742853e-05,
+      "loss": 0.3675,
+      "step": 985
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6891956047555132e-05,
+      "loss": 0.3463,
+      "step": 986
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6866882581712127e-05,
+      "loss": 0.32,
+      "step": 987
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6841796874909988e-05,
+      "loss": 0.3241,
+      "step": 988
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6816698971866703e-05,
+      "loss": 0.3404,
+      "step": 989
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.679158891732199e-05,
+      "loss": 0.3495,
+      "step": 990
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6766466756037227e-05,
+      "loss": 0.2915,
+      "step": 991
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6741332532795374e-05,
+      "loss": 0.306,
+      "step": 992
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6716186292400908e-05,
+      "loss": 0.3256,
+      "step": 993
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6691028079679707e-05,
+      "loss": 0.3493,
+      "step": 994
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6665857939479e-05,
+      "loss": 0.294,
+      "step": 995
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.664067591666728e-05,
+      "loss": 0.3251,
+      "step": 996
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6615482056134215e-05,
+      "loss": 0.3347,
+      "step": 997
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6590276402790593e-05,
+      "loss": 0.3335,
+      "step": 998
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6565059001568195e-05,
+      "loss": 0.3376,
+      "step": 999
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.653982989741976e-05,
+      "loss": 0.3107,
+      "step": 1000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6514589135318906e-05,
+      "loss": 0.3281,
+      "step": 1001
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6489336760259994e-05,
+      "loss": 0.343,
+      "step": 1002
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6464072817258123e-05,
+      "loss": 0.3105,
+      "step": 1003
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6438797351348982e-05,
+      "loss": 0.3374,
+      "step": 1004
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.641351040758883e-05,
+      "loss": 0.3263,
+      "step": 1005
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.638821203105437e-05,
+      "loss": 0.3267,
+      "step": 1006
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6362902266842683e-05,
+      "loss": 0.3451,
+      "step": 1007
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6337581160071156e-05,
+      "loss": 0.3229,
+      "step": 1008
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.63122487558774e-05,
+      "loss": 0.3248,
+      "step": 1009
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6286905099419148e-05,
+      "loss": 0.311,
+      "step": 1010
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.626155023587421e-05,
+      "loss": 0.3096,
+      "step": 1011
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6236184210440363e-05,
+      "loss": 0.3118,
+      "step": 1012
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.621080706833529e-05,
+      "loss": 0.3217,
+      "step": 1013
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6185418854796473e-05,
+      "loss": 0.3119,
+      "step": 1014
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6160019615081152e-05,
+      "loss": 0.3222,
+      "step": 1015
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.613460939446621e-05,
+      "loss": 0.2972,
+      "step": 1016
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6109188238248106e-05,
+      "loss": 0.328,
+      "step": 1017
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6083756191742796e-05,
+      "loss": 0.3087,
+      "step": 1018
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6058313300285644e-05,
+      "loss": 0.3047,
+      "step": 1019
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.603285960923135e-05,
+      "loss": 0.3097,
+      "step": 1020
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.600739516395387e-05,
+      "loss": 0.3264,
+      "step": 1021
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5981920009846308e-05,
+      "loss": 0.332,
+      "step": 1022
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.59564341923209e-05,
+      "loss": 0.3052,
+      "step": 1023
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5930937756808848e-05,
+      "loss": 0.3764,
+      "step": 1024
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5905430748760313e-05,
+      "loss": 0.3147,
+      "step": 1025
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.587991321364427e-05,
+      "loss": 0.3391,
+      "step": 1026
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.585438519694849e-05,
+      "loss": 0.3308,
+      "step": 1027
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5828846744179422e-05,
+      "loss": 0.3167,
+      "step": 1028
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5803297900862104e-05,
+      "loss": 0.361,
+      "step": 1029
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.577773871254011e-05,
+      "loss": 0.3481,
+      "step": 1030
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5752169224775455e-05,
+      "loss": 0.3368,
+      "step": 1031
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5726589483148496e-05,
+      "loss": 0.3143,
+      "step": 1032
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.57009995332579e-05,
+      "loss": 0.3315,
+      "step": 1033
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5675399420720508e-05,
+      "loss": 0.345,
+      "step": 1034
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5649789191171273e-05,
+      "loss": 0.3183,
+      "step": 1035
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.562416889026321e-05,
+      "loss": 0.3219,
+      "step": 1036
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5598538563667253e-05,
+      "loss": 0.3359,
+      "step": 1037
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5572898257072238e-05,
+      "loss": 0.332,
+      "step": 1038
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5547248016184766e-05,
+      "loss": 0.3126,
+      "step": 1039
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5521587886729164e-05,
+      "loss": 0.3395,
+      "step": 1040
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5495917914447382e-05,
+      "loss": 0.3513,
+      "step": 1041
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.547023814509892e-05,
+      "loss": 0.355,
+      "step": 1042
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.544454862446073e-05,
+      "loss": 0.324,
+      "step": 1043
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5418849398327156e-05,
+      "loss": 0.2994,
+      "step": 1044
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.539314051250985e-05,
+      "loss": 0.3562,
+      "step": 1045
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5367422012837666e-05,
+      "loss": 0.3454,
+      "step": 1046
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5341693945156604e-05,
+      "loss": 0.2955,
+      "step": 1047
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5315956355329724e-05,
+      "loss": 0.3148,
+      "step": 1048
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5290209289237054e-05,
+      "loss": 0.3336,
+      "step": 1049
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5264452792775524e-05,
+      "loss": 0.289,
+      "step": 1050
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5238686911858863e-05,
+      "loss": 0.3464,
+      "step": 1051
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.521291169241753e-05,
+      "loss": 0.319,
+      "step": 1052
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5187127180398646e-05,
+      "loss": 0.3342,
+      "step": 1053
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5161333421765873e-05,
+      "loss": 0.2891,
+      "step": 1054
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5135530462499375e-05,
+      "loss": 0.3169,
+      "step": 1055
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.510971834859572e-05,
+      "loss": 0.3221,
+      "step": 1056
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5083897126067766e-05,
+      "loss": 0.3434,
+      "step": 1057
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5058066840944647e-05,
+      "loss": 0.3145,
+      "step": 1058
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5032227539271618e-05,
+      "loss": 0.3482,
+      "step": 1059
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5006379267110035e-05,
+      "loss": 0.3309,
+      "step": 1060
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4980522070537227e-05,
+      "loss": 0.3068,
+      "step": 1061
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4954655995646437e-05,
+      "loss": 0.3249,
+      "step": 1062
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4928781088546733e-05,
+      "loss": 0.2972,
+      "step": 1063
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.490289739536293e-05,
+      "loss": 0.2961,
+      "step": 1064
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4877004962235507e-05,
+      "loss": 0.3489,
+      "step": 1065
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4851103835320513e-05,
+      "loss": 0.3289,
+      "step": 1066
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4825194060789506e-05,
+      "loss": 0.3244,
+      "step": 1067
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4799275684829455e-05,
+      "loss": 0.2948,
+      "step": 1068
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4773348753642656e-05,
+      "loss": 0.3421,
+      "step": 1069
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4747413313446672e-05,
+      "loss": 0.3468,
+      "step": 1070
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4721469410474215e-05,
+      "loss": 0.3035,
+      "step": 1071
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4695517090973094e-05,
+      "loss": 0.3128,
+      "step": 1072
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.466955640120612e-05,
+      "loss": 0.3588,
+      "step": 1073
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4643587387451017e-05,
+      "loss": 0.3295,
+      "step": 1074
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4617610096000367e-05,
+      "loss": 0.3063,
+      "step": 1075
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4591624573161482e-05,
+      "loss": 0.3184,
+      "step": 1076
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4565630865256366e-05,
+      "loss": 0.3246,
+      "step": 1077
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.453962901862161e-05,
+      "loss": 0.3295,
+      "step": 1078
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.45136190796083e-05,
+      "loss": 0.3322,
+      "step": 1079
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4487601094581978e-05,
+      "loss": 0.3833,
+      "step": 1080
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.446157510992249e-05,
+      "loss": 0.3478,
+      "step": 1081
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4435541172023974e-05,
+      "loss": 0.35,
+      "step": 1082
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.440949932729474e-05,
+      "loss": 0.3504,
+      "step": 1083
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.438344962215716e-05,
+      "loss": 0.3257,
+      "step": 1084
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.435739210304767e-05,
+      "loss": 0.3299,
+      "step": 1085
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.43313268164166e-05,
+      "loss": 0.3375,
+      "step": 1086
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4305253808728133e-05,
+      "loss": 0.3035,
+      "step": 1087
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4279173126460226e-05,
+      "loss": 0.3091,
+      "step": 1088
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4253084816104494e-05,
+      "loss": 0.3222,
+      "step": 1089
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.422698892416618e-05,
+      "loss": 0.296,
+      "step": 1090
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4200885497164026e-05,
+      "loss": 0.3148,
+      "step": 1091
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.41747745816302e-05,
+      "loss": 0.305,
+      "step": 1092
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.414865622411023e-05,
+      "loss": 0.3026,
+      "step": 1093
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4122530471162907e-05,
+      "loss": 0.3346,
+      "step": 1094
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4096397369360206e-05,
+      "loss": 0.3168,
+      "step": 1095
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.40702569652872e-05,
+      "loss": 0.3315,
+      "step": 1096
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.404410930554197e-05,
+      "loss": 0.3325,
+      "step": 1097
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4017954436735555e-05,
+      "loss": 0.3001,
+      "step": 1098
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3991792405491824e-05,
+      "loss": 0.3146,
+      "step": 1099
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.396562325844742e-05,
+      "loss": 0.3173,
+      "step": 1100
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3939447042251672e-05,
+      "loss": 0.2793,
+      "step": 1101
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3913263803566515e-05,
+      "loss": 0.3295,
+      "step": 1102
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.388707358906639e-05,
+      "loss": 0.335,
+      "step": 1103
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3860876445438176e-05,
+      "loss": 0.3398,
+      "step": 1104
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3834672419381124e-05,
+      "loss": 0.3378,
+      "step": 1105
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.380846155760672e-05,
+      "loss": 0.2989,
+      "step": 1106
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3782243906838672e-05,
+      "loss": 0.3552,
+      "step": 1107
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3756019513812757e-05,
+      "loss": 0.3032,
+      "step": 1108
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3729788425276786e-05,
+      "loss": 0.339,
+      "step": 1109
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3703550687990515e-05,
+      "loss": 0.3151,
+      "step": 1110
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3677306348725525e-05,
+      "loss": 0.3472,
+      "step": 1111
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.36510554542652e-05,
+      "loss": 0.3145,
+      "step": 1112
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3624798051404582e-05,
+      "loss": 0.3642,
+      "step": 1113
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3598534186950323e-05,
+      "loss": 0.3172,
+      "step": 1114
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.35722639077206e-05,
+      "loss": 0.3268,
+      "step": 1115
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3545987260545023e-05,
+      "loss": 0.3299,
+      "step": 1116
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3519704292264543e-05,
+      "loss": 0.3285,
+      "step": 1117
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3493415049731394e-05,
+      "loss": 0.3147,
+      "step": 1118
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3467119579808984e-05,
+      "loss": 0.3306,
+      "step": 1119
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3440817929371825e-05,
+      "loss": 0.3754,
+      "step": 1120
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3414510145305448e-05,
+      "loss": 0.3268,
+      "step": 1121
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3388196274506312e-05,
+      "loss": 0.339,
+      "step": 1122
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3361876363881737e-05,
+      "loss": 0.3176,
+      "step": 1123
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3335550460349804e-05,
+      "loss": 0.3668,
+      "step": 1124
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.330921861083927e-05,
+      "loss": 0.3059,
+      "step": 1125
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.32828808622895e-05,
+      "loss": 0.3162,
+      "step": 1126
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3256537261650386e-05,
+      "loss": 0.324,
+      "step": 1127
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.323018785588222e-05,
+      "loss": 0.2932,
+      "step": 1128
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.320383269195567e-05,
+      "loss": 0.3514,
+      "step": 1129
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3177471816851656e-05,
+      "loss": 0.3084,
+      "step": 1130
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3151105277561282e-05,
+      "loss": 0.304,
+      "step": 1131
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3124733121085755e-05,
+      "loss": 0.3082,
+      "step": 1132
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3098355394436277e-05,
+      "loss": 0.3405,
+      "step": 1133
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.307197214463401e-05,
+      "loss": 0.2966,
+      "step": 1134
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3045583418709933e-05,
+      "loss": 0.3552,
+      "step": 1135
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.301918926370479e-05,
+      "loss": 0.323,
+      "step": 1136
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.299278972666902e-05,
+      "loss": 0.3108,
+      "step": 1137
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2966384854662648e-05,
+      "loss": 0.3337,
+      "step": 1138
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2939974694755203e-05,
+      "loss": 0.3335,
+      "step": 1139
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2913559294025653e-05,
+      "loss": 0.3209,
+      "step": 1140
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2887138699562275e-05,
+      "loss": 0.2871,
+      "step": 1141
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.286071295846266e-05,
+      "loss": 0.2755,
+      "step": 1142
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2834282117833525e-05,
+      "loss": 0.3018,
+      "step": 1143
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.28078462247907e-05,
+      "loss": 0.3139,
+      "step": 1144
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.278140532645902e-05,
+      "loss": 0.3148,
+      "step": 1145
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2754959469972232e-05,
+      "loss": 0.3298,
+      "step": 1146
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2728508702472937e-05,
+      "loss": 0.298,
+      "step": 1147
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.270205307111247e-05,
+      "loss": 0.3049,
+      "step": 1148
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.267559262305086e-05,
+      "loss": 0.3254,
+      "step": 1149
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.264912740545671e-05,
+      "loss": 0.3484,
+      "step": 1150
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2622657465507117e-05,
+      "loss": 0.3308,
+      "step": 1151
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2596182850387614e-05,
+      "loss": 0.3215,
+      "step": 1152
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.256970360729205e-05,
+      "loss": 0.332,
+      "step": 1153
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2543219783422548e-05,
+      "loss": 0.3449,
+      "step": 1154
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2516731425989368e-05,
+      "loss": 0.3175,
+      "step": 1155
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2490238582210865e-05,
+      "loss": 0.3027,
+      "step": 1156
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2463741299313405e-05,
+      "loss": 0.311,
+      "step": 1157
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2437239624531232e-05,
+      "loss": 0.3239,
+      "step": 1158
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2410733605106462e-05,
+      "loss": 0.3117,
+      "step": 1159
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2384223288288932e-05,
+      "loss": 0.318,
+      "step": 1160
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.235770872133612e-05,
+      "loss": 0.3187,
+      "step": 1161
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2331189951513134e-05,
+      "loss": 0.3257,
+      "step": 1162
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2304667026092518e-05,
+      "loss": 0.267,
+      "step": 1163
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.227813999235426e-05,
+      "loss": 0.3004,
+      "step": 1164
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2251608897585663e-05,
+      "loss": 0.3696,
+      "step": 1165
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2225073789081248e-05,
+      "loss": 0.3066,
+      "step": 1166
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.219853471414273e-05,
+      "loss": 0.3322,
+      "step": 1167
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2171991720078866e-05,
+      "loss": 0.3207,
+      "step": 1168
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2145444854205406e-05,
+      "loss": 0.3261,
+      "step": 1169
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2118894163845e-05,
+      "loss": 0.3096,
+      "step": 1170
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2092339696327123e-05,
+      "loss": 0.3169,
+      "step": 1171
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2065781498987976e-05,
+      "loss": 0.3293,
+      "step": 1172
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.203921961917041e-05,
+      "loss": 0.3296,
+      "step": 1173
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.201265410422385e-05,
+      "loss": 0.3314,
+      "step": 1174
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.1986085001504183e-05,
+      "loss": 0.2997,
+      "step": 1175
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.1959512358373702e-05,
+      "loss": 0.3239,
+      "step": 1176
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.193293622220102e-05,
+      "loss": 0.3185,
+      "step": 1177
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.190635664036095e-05,
+      "loss": 0.2955,
+      "step": 1178
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1879773660234487e-05,
+      "loss": 0.3093,
+      "step": 1179
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1853187329208646e-05,
+      "loss": 0.3226,
+      "step": 1180
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1826597694676427e-05,
+      "loss": 0.315,
+      "step": 1181
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1800004804036744e-05,
+      "loss": 0.2928,
+      "step": 1182
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1773408704694267e-05,
+      "loss": 0.3536,
+      "step": 1183
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1746809444059433e-05,
+      "loss": 0.3217,
+      "step": 1184
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1720207069548288e-05,
+      "loss": 0.298,
+      "step": 1185
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1693601628582423e-05,
+      "loss": 0.3247,
+      "step": 1186
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.166699316858893e-05,
+      "loss": 0.3412,
+      "step": 1187
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.164038173700024e-05,
+      "loss": 0.3114,
+      "step": 1188
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.161376738125411e-05,
+      "loss": 0.3322,
+      "step": 1189
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1587150148793503e-05,
+      "loss": 0.3224,
+      "step": 1190
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1560530087066503e-05,
+      "loss": 0.3274,
+      "step": 1191
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1533907243526248e-05,
+      "loss": 0.3283,
+      "step": 1192
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1507281665630814e-05,
+      "loss": 0.3415,
+      "step": 1193
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.148065340084318e-05,
+      "loss": 0.3631,
+      "step": 1194
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1454022496631106e-05,
+      "loss": 0.3264,
+      "step": 1195
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.142738900046704e-05,
+      "loss": 0.3421,
+      "step": 1196
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1400752959828068e-05,
+      "loss": 0.3181,
+      "step": 1197
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1374114422195804e-05,
+      "loss": 0.2968,
+      "step": 1198
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.134747343505632e-05,
+      "loss": 0.2974,
+      "step": 1199
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1320830045900048e-05,
+      "loss": 0.3088,
+      "step": 1200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1294184302221704e-05,
+      "loss": 0.3501,
+      "step": 1201
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1267536251520194e-05,
+      "loss": 0.3338,
+      "step": 1202
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1240885941298557e-05,
+      "loss": 0.3199,
+      "step": 1203
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.121423341906384e-05,
+      "loss": 0.3062,
+      "step": 1204
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1187578732327037e-05,
+      "loss": 0.3033,
+      "step": 1205
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1160921928603003e-05,
+      "loss": 0.3052,
+      "step": 1206
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1134263055410375e-05,
+      "loss": 0.2971,
+      "step": 1207
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1107602160271457e-05,
+      "loss": 0.2939,
+      "step": 1208
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.108093929071219e-05,
+      "loss": 0.2977,
+      "step": 1209
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1054274494262e-05,
+      "loss": 0.3109,
+      "step": 1210
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1027607818453765e-05,
+      "loss": 0.3174,
+      "step": 1211
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1000939310823725e-05,
+      "loss": 0.3193,
+      "step": 1212
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0974269018911358e-05,
+      "loss": 0.3244,
+      "step": 1213
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.094759699025935e-05,
+      "loss": 0.3085,
+      "step": 1214
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.092092327241345e-05,
+      "loss": 0.3259,
+      "step": 1215
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0894247912922457e-05,
+      "loss": 0.2901,
+      "step": 1216
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0867570959338076e-05,
+      "loss": 0.3386,
+      "step": 1217
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.084089245921484e-05,
+      "loss": 0.3459,
+      "step": 1218
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.081421246011007e-05,
+      "loss": 0.3263,
+      "step": 1219
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0787531009583736e-05,
+      "loss": 0.2939,
+      "step": 1220
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0760848155198405e-05,
+      "loss": 0.3273,
+      "step": 1221
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0734163944519142e-05,
+      "loss": 0.323,
+      "step": 1222
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0707478425113433e-05,
+      "loss": 0.3107,
+      "step": 1223
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0680791644551096e-05,
+      "loss": 0.2814,
+      "step": 1224
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0654103650404197e-05,
+      "loss": 0.3297,
+      "step": 1225
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0627414490246966e-05,
+      "loss": 0.3285,
+      "step": 1226
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0600724211655705e-05,
+      "loss": 0.3044,
+      "step": 1227
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.057403286220872e-05,
+      "loss": 0.291,
+      "step": 1228
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.054734048948622e-05,
+      "loss": 0.3154,
+      "step": 1229
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.052064714107024e-05,
+      "loss": 0.2862,
+      "step": 1230
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.049395286454456e-05,
+      "loss": 0.2944,
+      "step": 1231
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.04672577074946e-05,
+      "loss": 0.3324,
+      "step": 1232
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.044056171750735e-05,
+      "loss": 0.327,
+      "step": 1233
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0413864942171315e-05,
+      "loss": 0.3379,
+      "step": 1234
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0387167429076354e-05,
+      "loss": 0.3034,
+      "step": 1235
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0360469225813686e-05,
+      "loss": 0.3348,
+      "step": 1236
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0333770379975727e-05,
+      "loss": 0.2992,
+      "step": 1237
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0307070939156043e-05,
+      "loss": 0.292,
+      "step": 1238
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.028037095094929e-05,
+      "loss": 0.2803,
+      "step": 1239
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0253670462951058e-05,
+      "loss": 0.3021,
+      "step": 1240
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.022696952275786e-05,
+      "loss": 0.3151,
+      "step": 1241
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0200268177966992e-05,
+      "loss": 0.3369,
+      "step": 1242
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0173566476176488e-05,
+      "loss": 0.3118,
+      "step": 1243
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.014686446498502e-05,
+      "loss": 0.3306,
+      "step": 1244
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.012016219199179e-05,
+      "loss": 0.2872,
+      "step": 1245
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.009345970479649e-05,
+      "loss": 0.3121,
+      "step": 1246
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.006675705099918e-05,
+      "loss": 0.2956,
+      "step": 1247
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.004005427820023e-05,
+      "loss": 0.328,
+      "step": 1248
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.00133514340002e-05,
+      "loss": 0.3206,
+      "step": 1249
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.99866485659998e-05,
+      "loss": 0.3108,
+      "step": 1250
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9959945721799775e-05,
+      "loss": 0.3346,
+      "step": 1251
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9933242949000822e-05,
+      "loss": 0.2974,
+      "step": 1252
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.990654029520351e-05,
+      "loss": 0.3056,
+      "step": 1253
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9879837808008212e-05,
+      "loss": 0.3192,
+      "step": 1254
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9853135535014985e-05,
+      "loss": 0.3132,
+      "step": 1255
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.982643352382351e-05,
+      "loss": 0.312,
+      "step": 1256
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9799731822033015e-05,
+      "loss": 0.2871,
+      "step": 1257
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9773030477242148e-05,
+      "loss": 0.3264,
+      "step": 1258
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.974632953704894e-05,
+      "loss": 0.2957,
+      "step": 1259
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9719629049050712e-05,
+      "loss": 0.3045,
+      "step": 1260
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.969292906084396e-05,
+      "loss": 0.2935,
+      "step": 1261
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9666229620024283e-05,
+      "loss": 0.2993,
+      "step": 1262
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9639530774186317e-05,
+      "loss": 0.3055,
+      "step": 1263
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.961283257092365e-05,
+      "loss": 0.3029,
+      "step": 1264
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9586135057828696e-05,
+      "loss": 0.2898,
+      "step": 1265
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9559438282492652e-05,
+      "loss": 0.3006,
+      "step": 1266
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.953274229250541e-05,
+      "loss": 0.358,
+      "step": 1267
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.950604713545545e-05,
+      "loss": 0.3176,
+      "step": 1268
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9479352858929763e-05,
+      "loss": 0.3032,
+      "step": 1269
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9452659510513783e-05,
+      "loss": 0.3124,
+      "step": 1270
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.942596713779129e-05,
+      "loss": 0.3263,
+      "step": 1271
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9399275788344302e-05,
+      "loss": 0.296,
+      "step": 1272
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.937258550975304e-05,
+      "loss": 0.329,
+      "step": 1273
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9345896349595806e-05,
+      "loss": 0.2997,
+      "step": 1274
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.93192083554489e-05,
+      "loss": 0.3267,
+      "step": 1275
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.929252157488657e-05,
+      "loss": 0.3212,
+      "step": 1276
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.926583605548086e-05,
+      "loss": 0.3027,
+      "step": 1277
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9239151844801594e-05,
+      "loss": 0.3212,
+      "step": 1278
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9212468990416267e-05,
+      "loss": 0.2824,
+      "step": 1279
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9185787539889933e-05,
+      "loss": 0.2885,
+      "step": 1280
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.915910754078516e-05,
+      "loss": 0.3117,
+      "step": 1281
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.913242904066193e-05,
+      "loss": 0.3294,
+      "step": 1282
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.910575208707755e-05,
+      "loss": 0.3377,
+      "step": 1283
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9079076727586556e-05,
+      "loss": 0.3065,
+      "step": 1284
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9052403009740658e-05,
+      "loss": 0.2843,
+      "step": 1285
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.902573098108865e-05,
+      "loss": 0.2912,
+      "step": 1286
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8999060689176282e-05,
+      "loss": 0.3412,
+      "step": 1287
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8972392181546235e-05,
+      "loss": 0.2945,
+      "step": 1288
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8945725505738008e-05,
+      "loss": 0.2746,
+      "step": 1289
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.891906070928782e-05,
+      "loss": 0.3244,
+      "step": 1290
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8892397839728542e-05,
+      "loss": 0.3226,
+      "step": 1291
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.886573694458963e-05,
+      "loss": 0.2906,
+      "step": 1292
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8839078071397007e-05,
+      "loss": 0.3092,
+      "step": 1293
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.881242126767297e-05,
+      "loss": 0.3145,
+      "step": 1294
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8785766580936167e-05,
+      "loss": 0.2872,
+      "step": 1295
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.875911405870145e-05,
+      "loss": 0.2934,
+      "step": 1296
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.873246374847981e-05,
+      "loss": 0.3307,
+      "step": 1297
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8705815697778306e-05,
+      "loss": 0.2893,
+      "step": 1298
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.867916995409996e-05,
+      "loss": 0.3111,
+      "step": 1299
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.865252656494368e-05,
+      "loss": 0.3134,
+      "step": 1300
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.86258855778042e-05,
+      "loss": 0.342,
+      "step": 1301
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.859924704017194e-05,
+      "loss": 0.3237,
+      "step": 1302
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8572610999532972e-05,
+      "loss": 0.314,
+      "step": 1303
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.85459775033689e-05,
+      "loss": 0.3087,
+      "step": 1304
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8519346599156822e-05,
+      "loss": 0.2748,
+      "step": 1305
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8492718334369196e-05,
+      "loss": 0.3022,
+      "step": 1306
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8466092756473762e-05,
+      "loss": 0.2976,
+      "step": 1307
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8439469912933507e-05,
+      "loss": 0.3548,
+      "step": 1308
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8412849851206504e-05,
+      "loss": 0.3019,
+      "step": 1309
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.838623261874589e-05,
+      "loss": 0.2928,
+      "step": 1310
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8359618262999768e-05,
+      "loss": 0.3038,
+      "step": 1311
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.833300683141108e-05,
+      "loss": 0.3021,
+      "step": 1312
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8306398371417577e-05,
+      "loss": 0.3168,
+      "step": 1313
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.827979293045172e-05,
+      "loss": 0.3222,
+      "step": 1314
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8253190555940574e-05,
+      "loss": 0.3442,
+      "step": 1315
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8226591295305733e-05,
+      "loss": 0.3232,
+      "step": 1316
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8199995195963263e-05,
+      "loss": 0.3133,
+      "step": 1317
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8173402305323576e-05,
+      "loss": 0.3486,
+      "step": 1318
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8146812670791358e-05,
+      "loss": 0.3332,
+      "step": 1319
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8120226339765516e-05,
+      "loss": 0.2731,
+      "step": 1320
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8093643359639056e-05,
+      "loss": 0.3128,
+      "step": 1321
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8067063777798982e-05,
+      "loss": 0.2901,
+      "step": 1322
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.80404876416263e-05,
+      "loss": 0.279,
+      "step": 1323
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8013914998495823e-05,
+      "loss": 0.2924,
+      "step": 1324
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.798734589577616e-05,
+      "loss": 0.304,
+      "step": 1325
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7960780380829592e-05,
+      "loss": 0.3343,
+      "step": 1326
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7934218501012027e-05,
+      "loss": 0.306,
+      "step": 1327
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7907660303672887e-05,
+      "loss": 0.3043,
+      "step": 1328
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7881105836155006e-05,
+      "loss": 0.334,
+      "step": 1329
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.78545551457946e-05,
+      "loss": 0.2868,
+      "step": 1330
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7828008279921138e-05,
+      "loss": 0.3121,
+      "step": 1331
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.780146528585727e-05,
+      "loss": 0.3173,
+      "step": 1332
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7774926210918755e-05,
+      "loss": 0.2932,
+      "step": 1333
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7748391102414347e-05,
+      "loss": 0.3011,
+      "step": 1334
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.772186000764574e-05,
+      "loss": 0.3122,
+      "step": 1335
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.769533297390749e-05,
+      "loss": 0.3048,
+      "step": 1336
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7668810048486876e-05,
+      "loss": 0.2962,
+      "step": 1337
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.764229127866388e-05,
+      "loss": 0.3047,
+      "step": 1338
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7615776711711078e-05,
+      "loss": 0.3301,
+      "step": 1339
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.758926639489354e-05,
+      "loss": 0.2899,
+      "step": 1340
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7562760375468768e-05,
+      "loss": 0.296,
+      "step": 1341
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7536258700686605e-05,
+      "loss": 0.3222,
+      "step": 1342
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.750976141778914e-05,
+      "loss": 0.3081,
+      "step": 1343
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7483268574010635e-05,
+      "loss": 0.2742,
+      "step": 1344
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.745678021657746e-05,
+      "loss": 0.3116,
+      "step": 1345
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7430296392707955e-05,
+      "loss": 0.3023,
+      "step": 1346
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7403817149612396e-05,
+      "loss": 0.3127,
+      "step": 1347
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7377342534492886e-05,
+      "loss": 0.3428,
+      "step": 1348
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.73508725945433e-05,
+      "loss": 0.298,
+      "step": 1349
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.732440737694915e-05,
+      "loss": 0.2911,
+      "step": 1350
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7297946928887533e-05,
+      "loss": 0.3064,
+      "step": 1351
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7271491297527073e-05,
+      "loss": 0.3066,
+      "step": 1352
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7245040530027778e-05,
+      "loss": 0.3047,
+      "step": 1353
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7218594673540984e-05,
+      "loss": 0.295,
+      "step": 1354
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7192153775209303e-05,
+      "loss": 0.3147,
+      "step": 1355
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.716571788216648e-05,
+      "loss": 0.2953,
+      "step": 1356
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7139287041537342e-05,
+      "loss": 0.3253,
+      "step": 1357
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7112861300437728e-05,
+      "loss": 0.3139,
+      "step": 1358
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7086440705974357e-05,
+      "loss": 0.2788,
+      "step": 1359
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7060025305244797e-05,
+      "loss": 0.2758,
+      "step": 1360
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7033615145337355e-05,
+      "loss": 0.2973,
+      "step": 1361
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7007210273330982e-05,
+      "loss": 0.2994,
+      "step": 1362
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6980810736295213e-05,
+      "loss": 0.2806,
+      "step": 1363
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6954416581290074e-05,
+      "loss": 0.3263,
+      "step": 1364
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6928027855365998e-05,
+      "loss": 0.3256,
+      "step": 1365
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6901644605563727e-05,
+      "loss": 0.3013,
+      "step": 1366
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.687526687891425e-05,
+      "loss": 0.3038,
+      "step": 1367
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6848894722438728e-05,
+      "loss": 0.3019,
+      "step": 1368
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.682252818314835e-05,
+      "loss": 0.2927,
+      "step": 1369
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6796167308044334e-05,
+      "loss": 0.2905,
+      "step": 1370
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6769812144117783e-05,
+      "loss": 0.2873,
+      "step": 1371
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6743462738349624e-05,
+      "loss": 0.3129,
+      "step": 1372
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.67171191377105e-05,
+      "loss": 0.2643,
+      "step": 1373
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6690781389160733e-05,
+      "loss": 0.2965,
+      "step": 1374
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6664449539650207e-05,
+      "loss": 0.316,
+      "step": 1375
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6638123636118266e-05,
+      "loss": 0.2844,
+      "step": 1376
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6611803725493695e-05,
+      "loss": 0.3372,
+      "step": 1377
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6585489854694566e-05,
+      "loss": 0.2943,
+      "step": 1378
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.655918207062818e-05,
+      "loss": 0.2869,
+      "step": 1379
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6532880420191023e-05,
+      "loss": 0.2775,
+      "step": 1380
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6506584950268613e-05,
+      "loss": 0.2838,
+      "step": 1381
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6480295707735457e-05,
+      "loss": 0.3027,
+      "step": 1382
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6454012739454984e-05,
+      "loss": 0.3248,
+      "step": 1383
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6427736092279405e-05,
+      "loss": 0.3149,
+      "step": 1384
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.640146581304968e-05,
+      "loss": 0.3088,
+      "step": 1385
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6375201948595428e-05,
+      "loss": 0.2889,
+      "step": 1386
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.634894454573481e-05,
+      "loss": 0.2676,
+      "step": 1387
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.632269365127448e-05,
+      "loss": 0.291,
+      "step": 1388
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.629644931200949e-05,
+      "loss": 0.3005,
+      "step": 1389
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.627021157472322e-05,
+      "loss": 0.2896,
+      "step": 1390
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6243980486187253e-05,
+      "loss": 0.3212,
+      "step": 1391
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.621775609316133e-05,
+      "loss": 0.3338,
+      "step": 1392
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6191538442393284e-05,
+      "loss": 0.3179,
+      "step": 1393
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6165327580618883e-05,
+      "loss": 0.3257,
+      "step": 1394
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6139123554561824e-05,
+      "loss": 0.3206,
+      "step": 1395
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6112926410933617e-05,
+      "loss": 0.3156,
+      "step": 1396
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6086736196433498e-05,
+      "loss": 0.3133,
+      "step": 1397
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.606055295774833e-05,
+      "loss": 0.2885,
+      "step": 1398
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6034376741552583e-05,
+      "loss": 0.278,
+      "step": 1399
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6008207594508186e-05,
+      "loss": 0.2991,
+      "step": 1400
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.598204556326445e-05,
+      "loss": 0.2825,
+      "step": 1401
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5955890694458034e-05,
+      "loss": 0.2856,
+      "step": 1402
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.592974303471281e-05,
+      "loss": 0.316,
+      "step": 1403
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5903602630639797e-05,
+      "loss": 0.304,
+      "step": 1404
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5877469528837096e-05,
+      "loss": 0.3095,
+      "step": 1405
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5851343775889775e-05,
+      "loss": 0.3296,
+      "step": 1406
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.58252254183698e-05,
+      "loss": 0.3259,
+      "step": 1407
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.579911450283598e-05,
+      "loss": 0.294,
+      "step": 1408
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5773011075833823e-05,
+      "loss": 0.3229,
+      "step": 1409
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5746915183895516e-05,
+      "loss": 0.3216,
+      "step": 1410
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.572082687353978e-05,
+      "loss": 0.2723,
+      "step": 1411
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5694746191271874e-05,
+      "loss": 0.3071,
+      "step": 1412
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5668673183583408e-05,
+      "loss": 0.2805,
+      "step": 1413
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.564260789695233e-05,
+      "loss": 0.2984,
+      "step": 1414
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5616550377842843e-05,
+      "loss": 0.324,
+      "step": 1415
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5590500672705272e-05,
+      "loss": 0.3359,
+      "step": 1416
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5564458827976022e-05,
+      "loss": 0.3256,
+      "step": 1417
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5538424890077514e-05,
+      "loss": 0.3045,
+      "step": 1418
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5512398905418032e-05,
+      "loss": 0.2787,
+      "step": 1419
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5486380920391703e-05,
+      "loss": 0.3128,
+      "step": 1420
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5460370981378397e-05,
+      "loss": 0.2777,
+      "step": 1421
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5434369134743644e-05,
+      "loss": 0.2939,
+      "step": 1422
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5408375426838525e-05,
+      "loss": 0.2857,
+      "step": 1423
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.538238990399964e-05,
+      "loss": 0.2934,
+      "step": 1424
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.535641261254899e-05,
+      "loss": 0.2721,
+      "step": 1425
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5330443598793888e-05,
+      "loss": 0.3247,
+      "step": 1426
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5304482909026912e-05,
+      "loss": 0.3125,
+      "step": 1427
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5278530589525792e-05,
+      "loss": 0.2978,
+      "step": 1428
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5252586686553337e-05,
+      "loss": 0.3165,
+      "step": 1429
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5226651246357346e-05,
+      "loss": 0.3168,
+      "step": 1430
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5200724315170552e-05,
+      "loss": 0.2999,
+      "step": 1431
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5174805939210504e-05,
+      "loss": 0.3046,
+      "step": 1432
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5148896164679494e-05,
+      "loss": 0.2821,
+      "step": 1433
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.51229950377645e-05,
+      "loss": 0.2962,
+      "step": 1434
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.509710260463708e-05,
+      "loss": 0.3273,
+      "step": 1435
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5071218911453269e-05,
+      "loss": 0.306,
+      "step": 1436
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.504534400435357e-05,
+      "loss": 0.2975,
+      "step": 1437
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5019477929462778e-05,
+      "loss": 0.3262,
+      "step": 1438
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4993620732889967e-05,
+      "loss": 0.3375,
+      "step": 1439
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4967772460728387e-05,
+      "loss": 0.2914,
+      "step": 1440
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4941933159055361e-05,
+      "loss": 0.3077,
+      "step": 1441
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4916102873932237e-05,
+      "loss": 0.2763,
+      "step": 1442
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4890281651404288e-05,
+      "loss": 0.2807,
+      "step": 1443
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.486446953750063e-05,
+      "loss": 0.2894,
+      "step": 1444
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.483866657823413e-05,
+      "loss": 0.2776,
+      "step": 1445
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.481287281960136e-05,
+      "loss": 0.3123,
+      "step": 1446
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4787088307582478e-05,
+      "loss": 0.3199,
+      "step": 1447
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4761313088141145e-05,
+      "loss": 0.2961,
+      "step": 1448
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4735547207224481e-05,
+      "loss": 0.2931,
+      "step": 1449
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4709790710762954e-05,
+      "loss": 0.2752,
+      "step": 1450
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4684043644670288e-05,
+      "loss": 0.3297,
+      "step": 1451
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4658306054843401e-05,
+      "loss": 0.2923,
+      "step": 1452
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4632577987162343e-05,
+      "loss": 0.2943,
+      "step": 1453
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4606859487490158e-05,
+      "loss": 0.2692,
+      "step": 1454
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4581150601672844e-05,
+      "loss": 0.2883,
+      "step": 1455
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4555451375539273e-05,
+      "loss": 0.3064,
+      "step": 1456
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4529761854901088e-05,
+      "loss": 0.3126,
+      "step": 1457
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.450408208555262e-05,
+      "loss": 0.2881,
+      "step": 1458
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4478412113270843e-05,
+      "loss": 0.3105,
+      "step": 1459
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4452751983815244e-05,
+      "loss": 0.2797,
+      "step": 1460
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4427101742927767e-05,
+      "loss": 0.3049,
+      "step": 1461
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.440146143633275e-05,
+      "loss": 0.3217,
+      "step": 1462
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4375831109736798e-05,
+      "loss": 0.2927,
+      "step": 1463
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4350210808828723e-05,
+      "loss": 0.2851,
+      "step": 1464
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.43246005792795e-05,
+      "loss": 0.2767,
+      "step": 1465
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4299000466742103e-05,
+      "loss": 0.3188,
+      "step": 1466
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4273410516851502e-05,
+      "loss": 0.3107,
+      "step": 1467
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.424783077522455e-05,
+      "loss": 0.2948,
+      "step": 1468
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4222261287459898e-05,
+      "loss": 0.2939,
+      "step": 1469
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.41967020991379e-05,
+      "loss": 0.2761,
+      "step": 1470
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4171153255820583e-05,
+      "loss": 0.2875,
+      "step": 1471
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4145614803051518e-05,
+      "loss": 0.2802,
+      "step": 1472
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4120086786355738e-05,
+      "loss": 0.3111,
+      "step": 1473
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4094569251239695e-05,
+      "loss": 0.3075,
+      "step": 1474
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4069062243191157e-05,
+      "loss": 0.3023,
+      "step": 1475
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4043565807679107e-05,
+      "loss": 0.3006,
+      "step": 1476
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.401807999015369e-05,
+      "loss": 0.3213,
+      "step": 1477
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3992604836046137e-05,
+      "loss": 0.293,
+      "step": 1478
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3967140390768659e-05,
+      "loss": 0.3179,
+      "step": 1479
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.394168669971436e-05,
+      "loss": 0.3032,
+      "step": 1480
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.391624380825721e-05,
+      "loss": 0.2814,
+      "step": 1481
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.38908117617519e-05,
+      "loss": 0.3255,
+      "step": 1482
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.386539060553379e-05,
+      "loss": 0.2825,
+      "step": 1483
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3839980384918854e-05,
+      "loss": 0.2799,
+      "step": 1484
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3814581145203532e-05,
+      "loss": 0.2862,
+      "step": 1485
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3789192931664715e-05,
+      "loss": 0.2955,
+      "step": 1486
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3763815789559642e-05,
+      "loss": 0.3003,
+      "step": 1487
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3738449764125796e-05,
+      "loss": 0.2904,
+      "step": 1488
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3713094900580855e-05,
+      "loss": 0.303,
+      "step": 1489
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3687751244122609e-05,
+      "loss": 0.2983,
+      "step": 1490
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3662418839928849e-05,
+      "loss": 0.3082,
+      "step": 1491
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3637097733157323e-05,
+      "loss": 0.2928,
+      "step": 1492
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.361178796894563e-05,
+      "loss": 0.2976,
+      "step": 1493
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3586489592411173e-05,
+      "loss": 0.3208,
+      "step": 1494
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3561202648651021e-05,
+      "loss": 0.3023,
+      "step": 1495
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3535927182741884e-05,
+      "loss": 0.3152,
+      "step": 1496
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3510663239740014e-05,
+      "loss": 0.3147,
+      "step": 1497
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3485410864681106e-05,
+      "loss": 0.3243,
+      "step": 1498
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3460170102580242e-05,
+      "loss": 0.2918,
+      "step": 1499
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3434940998431811e-05,
+      "loss": 0.2953,
+      "step": 1500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3409723597209418e-05,
+      "loss": 0.291,
+      "step": 1501
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3384517943865784e-05,
+      "loss": 0.3015,
+      "step": 1502
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3359324083332723e-05,
+      "loss": 0.3089,
+      "step": 1503
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3334142060521007e-05,
+      "loss": 0.3241,
+      "step": 1504
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3308971920320298e-05,
+      "loss": 0.3327,
+      "step": 1505
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3283813707599093e-05,
+      "loss": 0.2868,
+      "step": 1506
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3258667467204631e-05,
+      "loss": 0.306,
+      "step": 1507
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3233533243962775e-05,
+      "loss": 0.2743,
+      "step": 1508
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3208411082678016e-05,
+      "loss": 0.3041,
+      "step": 1509
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3183301028133299e-05,
+      "loss": 0.3071,
+      "step": 1510
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.315820312509001e-05,
+      "loss": 0.2848,
+      "step": 1511
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.313311741828788e-05,
+      "loss": 0.2979,
+      "step": 1512
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3108043952444873e-05,
+      "loss": 0.269,
+      "step": 1513
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3082982772257157e-05,
+      "loss": 0.2726,
+      "step": 1514
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.305793392239898e-05,
+      "loss": 0.3501,
+      "step": 1515
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.303289744752263e-05,
+      "loss": 0.3211,
+      "step": 1516
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3007873392258316e-05,
+      "loss": 0.2597,
+      "step": 1517
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2982861801214122e-05,
+      "loss": 0.2894,
+      "step": 1518
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2957862718975914e-05,
+      "loss": 0.2547,
+      "step": 1519
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.293287619010724e-05,
+      "loss": 0.3089,
+      "step": 1520
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2907902259149287e-05,
+      "loss": 0.3266,
+      "step": 1521
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2882940970620795e-05,
+      "loss": 0.2656,
+      "step": 1522
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2857992369017938e-05,
+      "loss": 0.2745,
+      "step": 1523
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2833056498814291e-05,
+      "loss": 0.2775,
+      "step": 1524
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2808133404460743e-05,
+      "loss": 0.273,
+      "step": 1525
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2783223130385395e-05,
+      "loss": 0.2797,
+      "step": 1526
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2758325720993492e-05,
+      "loss": 0.2797,
+      "step": 1527
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2733441220667365e-05,
+      "loss": 0.2885,
+      "step": 1528
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2708569673766313e-05,
+      "loss": 0.2797,
+      "step": 1529
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2683711124626548e-05,
+      "loss": 0.3002,
+      "step": 1530
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2658865617561123e-05,
+      "loss": 0.3046,
+      "step": 1531
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2634033196859834e-05,
+      "loss": 0.2708,
+      "step": 1532
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2609213906789145e-05,
+      "loss": 0.2647,
+      "step": 1533
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.258440779159213e-05,
+      "loss": 0.2835,
+      "step": 1534
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2559614895488354e-05,
+      "loss": 0.2618,
+      "step": 1535
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.253483526267384e-05,
+      "loss": 0.3022,
+      "step": 1536
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.251006893732094e-05,
+      "loss": 0.2929,
+      "step": 1537
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.248531596357832e-05,
+      "loss": 0.2801,
+      "step": 1538
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2460576385570822e-05,
+      "loss": 0.3147,
+      "step": 1539
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2435850247399398e-05,
+      "loss": 0.2945,
+      "step": 1540
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2411137593141081e-05,
+      "loss": 0.2977,
+      "step": 1541
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2386438466848829e-05,
+      "loss": 0.2922,
+      "step": 1542
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2361752912551497e-05,
+      "loss": 0.2993,
+      "step": 1543
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2337080974253763e-05,
+      "loss": 0.2782,
+      "step": 1544
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2312422695936008e-05,
+      "loss": 0.2574,
+      "step": 1545
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.228777812155427e-05,
+      "loss": 0.296,
+      "step": 1546
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2263147295040178e-05,
+      "loss": 0.2938,
+      "step": 1547
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2238530260300824e-05,
+      "loss": 0.3017,
+      "step": 1548
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.221392706121873e-05,
+      "loss": 0.2994,
+      "step": 1549
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2189337741651759e-05,
+      "loss": 0.2916,
+      "step": 1550
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2164762345433028e-05,
+      "loss": 0.308,
+      "step": 1551
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2140200916370816e-05,
+      "loss": 0.2704,
+      "step": 1552
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.211565349824854e-05,
+      "loss": 0.2888,
+      "step": 1553
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2091120134824616e-05,
+      "loss": 0.2975,
+      "step": 1554
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2066600869832405e-05,
+      "loss": 0.2941,
+      "step": 1555
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2042095746980142e-05,
+      "loss": 0.3007,
+      "step": 1556
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2017604809950862e-05,
+      "loss": 0.3,
+      "step": 1557
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1993128102402301e-05,
+      "loss": 0.2828,
+      "step": 1558
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1968665667966822e-05,
+      "loss": 0.2903,
+      "step": 1559
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.194421755025136e-05,
+      "loss": 0.2704,
+      "step": 1560
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1919783792837327e-05,
+      "loss": 0.3044,
+      "step": 1561
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1895364439280522e-05,
+      "loss": 0.3071,
+      "step": 1562
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1870959533111083e-05,
+      "loss": 0.2912,
+      "step": 1563
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1846569117833396e-05,
+      "loss": 0.2897,
+      "step": 1564
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1822193236925987e-05,
+      "loss": 0.2992,
+      "step": 1565
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.179783193384152e-05,
+      "loss": 0.2786,
+      "step": 1566
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.177348525200663e-05,
+      "loss": 0.2834,
+      "step": 1567
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.17491532348219e-05,
+      "loss": 0.2833,
+      "step": 1568
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1724835925661795e-05,
+      "loss": 0.2589,
+      "step": 1569
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1700533367874526e-05,
+      "loss": 0.305,
+      "step": 1570
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1676245604782029e-05,
+      "loss": 0.2864,
+      "step": 1571
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1651972679679863e-05,
+      "loss": 0.2775,
+      "step": 1572
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1627714635837139e-05,
+      "loss": 0.3075,
+      "step": 1573
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1603471516496435e-05,
+      "loss": 0.2683,
+      "step": 1574
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1579243364873723e-05,
+      "loss": 0.294,
+      "step": 1575
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1555030224158304e-05,
+      "loss": 0.2847,
+      "step": 1576
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1530832137512718e-05,
+      "loss": 0.2755,
+      "step": 1577
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1506649148072642e-05,
+      "loss": 0.3175,
+      "step": 1578
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.148248129894689e-05,
+      "loss": 0.2978,
+      "step": 1579
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1458328633217257e-05,
+      "loss": 0.281,
+      "step": 1580
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1434191193938453e-05,
+      "loss": 0.2756,
+      "step": 1581
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1410069024138096e-05,
+      "loss": 0.314,
+      "step": 1582
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1385962166816544e-05,
+      "loss": 0.2764,
+      "step": 1583
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1361870664946873e-05,
+      "loss": 0.2672,
+      "step": 1584
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1337794561474782e-05,
+      "loss": 0.286,
+      "step": 1585
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1313733899318526e-05,
+      "loss": 0.2905,
+      "step": 1586
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1289688721368832e-05,
+      "loss": 0.2925,
+      "step": 1587
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.126565907048882e-05,
+      "loss": 0.3007,
+      "step": 1588
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.124164498951394e-05,
+      "loss": 0.2424,
+      "step": 1589
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1217646521251878e-05,
+      "loss": 0.2696,
+      "step": 1590
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.11936637084825e-05,
+      "loss": 0.2884,
+      "step": 1591
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1169696593957751e-05,
+      "loss": 0.3357,
+      "step": 1592
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1145745220401602e-05,
+      "loss": 0.3003,
+      "step": 1593
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1121809630509959e-05,
+      "loss": 0.3149,
+      "step": 1594
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.109788986695059e-05,
+      "loss": 0.2778,
+      "step": 1595
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1073985972363058e-05,
+      "loss": 0.287,
+      "step": 1596
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1050097989358646e-05,
+      "loss": 0.2842,
+      "step": 1597
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.102622596052024e-05,
+      "loss": 0.2572,
+      "step": 1598
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.100236992840232e-05,
+      "loss": 0.2752,
+      "step": 1599
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0978529935530835e-05,
+      "loss": 0.3057,
+      "step": 1600
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0954706024403143e-05,
+      "loss": 0.3065,
+      "step": 1601
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0930898237487937e-05,
+      "loss": 0.277,
+      "step": 1602
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.090710661722517e-05,
+      "loss": 0.2628,
+      "step": 1603
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0883331206025964e-05,
+      "loss": 0.3024,
+      "step": 1604
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.085957204627256e-05,
+      "loss": 0.2825,
+      "step": 1605
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0835829180318215e-05,
+      "loss": 0.2967,
+      "step": 1606
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0812102650487173e-05,
+      "loss": 0.264,
+      "step": 1607
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0788392499074511e-05,
+      "loss": 0.3113,
+      "step": 1608
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0764698768346134e-05,
+      "loss": 0.2813,
+      "step": 1609
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0741021500538689e-05,
+      "loss": 0.3122,
+      "step": 1610
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0717360737859446e-05,
+      "loss": 0.2991,
+      "step": 1611
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0693716522486262e-05,
+      "loss": 0.2776,
+      "step": 1612
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.067008889656752e-05,
+      "loss": 0.2752,
+      "step": 1613
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0646477902222015e-05,
+      "loss": 0.3306,
+      "step": 1614
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0622883581538864e-05,
+      "loss": 0.2689,
+      "step": 1615
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0599305976577511e-05,
+      "loss": 0.2958,
+      "step": 1616
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0575745129367587e-05,
+      "loss": 0.3006,
+      "step": 1617
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0552201081908828e-05,
+      "loss": 0.259,
+      "step": 1618
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0528673876171036e-05,
+      "loss": 0.2777,
+      "step": 1619
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0505163554094018e-05,
+      "loss": 0.2854,
+      "step": 1620
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0481670157587435e-05,
+      "loss": 0.318,
+      "step": 1621
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0458193728530807e-05,
+      "loss": 0.3022,
+      "step": 1622
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0434734308773411e-05,
+      "loss": 0.2891,
+      "step": 1623
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0411291940134196e-05,
+      "loss": 0.2892,
+      "step": 1624
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.038786666440169e-05,
+      "loss": 0.2947,
+      "step": 1625
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0364458523334e-05,
+      "loss": 0.2928,
+      "step": 1626
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0341067558658659e-05,
+      "loss": 0.2917,
+      "step": 1627
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0317693812072567e-05,
+      "loss": 0.279,
+      "step": 1628
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.029433732524197e-05,
+      "loss": 0.3009,
+      "step": 1629
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.027099813980232e-05,
+      "loss": 0.2998,
+      "step": 1630
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0247676297358229e-05,
+      "loss": 0.2623,
+      "step": 1631
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.02243718394834e-05,
+      "loss": 0.2957,
+      "step": 1632
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0201084807720545e-05,
+      "loss": 0.2837,
+      "step": 1633
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.017781524358131e-05,
+      "loss": 0.2887,
+      "step": 1634
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.01545631885462e-05,
+      "loss": 0.3131,
+      "step": 1635
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0131328684064519e-05,
+      "loss": 0.3018,
+      "step": 1636
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0108111771554274e-05,
+      "loss": 0.3181,
+      "step": 1637
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.008491249240212e-05,
+      "loss": 0.2888,
+      "step": 1638
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0061730887963268e-05,
+      "loss": 0.3236,
+      "step": 1639
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0038566999561434e-05,
+      "loss": 0.2801,
+      "step": 1640
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.001542086848875e-05,
+      "loss": 0.2893,
+      "step": 1641
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.99229253600569e-06,
+      "loss": 0.2974,
+      "step": 1642
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.969182043341e-06,
+      "loss": 0.2708,
+      "step": 1643
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.946089431691632e-06,
+      "loss": 0.2954,
+      "step": 1644
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.923014742222654e-06,
+      "loss": 0.3046,
+      "step": 1645
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.899958016067192e-06,
+      "loss": 0.2836,
+      "step": 1646
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.876919294326347e-06,
+      "loss": 0.2623,
+      "step": 1647
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.853898618069123e-06,
+      "loss": 0.2962,
+      "step": 1648
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.830896028332366e-06,
+      "loss": 0.267,
+      "step": 1649
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.807911566120669e-06,
+      "loss": 0.2854,
+      "step": 1650
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.78494527240632e-06,
+      "loss": 0.2608,
+      "step": 1651
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.761997188129208e-06,
+      "loss": 0.287,
+      "step": 1652
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.739067354196772e-06,
+      "loss": 0.268,
+      "step": 1653
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.716155811483929e-06,
+      "loss": 0.2737,
+      "step": 1654
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.69326260083296e-06,
+      "loss": 0.2788,
+      "step": 1655
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.670387763053479e-06,
+      "loss": 0.2586,
+      "step": 1656
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.647531338922379e-06,
+      "loss": 0.296,
+      "step": 1657
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.624693369183673e-06,
+      "loss": 0.2747,
+      "step": 1658
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.60187389454851e-06,
+      "loss": 0.2902,
+      "step": 1659
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.57907295569509e-06,
+      "loss": 0.285,
+      "step": 1660
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.556290593268522e-06,
+      "loss": 0.2909,
+      "step": 1661
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.533526847880835e-06,
+      "loss": 0.2521,
+      "step": 1662
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.51078176011085e-06,
+      "loss": 0.277,
+      "step": 1663
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.488055370504166e-06,
+      "loss": 0.2763,
+      "step": 1664
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.465347719573e-06,
+      "loss": 0.2769,
+      "step": 1665
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.442658847796194e-06,
+      "loss": 0.2909,
+      "step": 1666
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.419988795619127e-06,
+      "loss": 0.2812,
+      "step": 1667
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.397337603453592e-06,
+      "loss": 0.2766,
+      "step": 1668
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.374705311677783e-06,
+      "loss": 0.293,
+      "step": 1669
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.35209196063622e-06,
+      "loss": 0.3046,
+      "step": 1670
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.329497590639638e-06,
+      "loss": 0.2784,
+      "step": 1671
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.30692224196492e-06,
+      "loss": 0.2833,
+      "step": 1672
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.28436595485508e-06,
+      "loss": 0.2768,
+      "step": 1673
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.261828769519134e-06,
+      "loss": 0.2593,
+      "step": 1674
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.239310726132027e-06,
+      "loss": 0.3007,
+      "step": 1675
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.21681186483462e-06,
+      "loss": 0.2786,
+      "step": 1676
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.194332225733557e-06,
+      "loss": 0.2864,
+      "step": 1677
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.171871848901217e-06,
+      "loss": 0.2922,
+      "step": 1678
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.149430774375646e-06,
+      "loss": 0.2604,
+      "step": 1679
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.127009042160484e-06,
+      "loss": 0.3176,
+      "step": 1680
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.104606692224892e-06,
+      "loss": 0.2697,
+      "step": 1681
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.082223764503451e-06,
+      "loss": 0.294,
+      "step": 1682
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.05986029889617e-06,
+      "loss": 0.2754,
+      "step": 1683
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.037516335268344e-06,
+      "loss": 0.3109,
+      "step": 1684
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.015191913450468e-06,
+      "loss": 0.2863,
+      "step": 1685
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.992887073238261e-06,
+      "loss": 0.2758,
+      "step": 1686
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.970601854392498e-06,
+      "loss": 0.2871,
+      "step": 1687
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.948336296638987e-06,
+      "loss": 0.2833,
+      "step": 1688
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.926090439668487e-06,
+      "loss": 0.3074,
+      "step": 1689
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.903864323136635e-06,
+      "loss": 0.3121,
+      "step": 1690
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.881657986663888e-06,
+      "loss": 0.2892,
+      "step": 1691
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.85947146983543e-06,
+      "loss": 0.2843,
+      "step": 1692
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.837304812201128e-06,
+      "loss": 0.259,
+      "step": 1693
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.815158053275437e-06,
+      "loss": 0.2698,
+      "step": 1694
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.793031232537344e-06,
+      "loss": 0.2951,
+      "step": 1695
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.770924389430296e-06,
+      "loss": 0.2642,
+      "step": 1696
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.748837563362123e-06,
+      "loss": 0.286,
+      "step": 1697
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.726770793704982e-06,
+      "loss": 0.2946,
+      "step": 1698
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.704724119795265e-06,
+      "loss": 0.2997,
+      "step": 1699
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.68269758093355e-06,
+      "loss": 0.2678,
+      "step": 1700
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.660691216384516e-06,
+      "loss": 0.2759,
+      "step": 1701
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.638705065376887e-06,
+      "loss": 0.2914,
+      "step": 1702
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.616739167103348e-06,
+      "loss": 0.279,
+      "step": 1703
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.594793560720478e-06,
+      "loss": 0.2889,
+      "step": 1704
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.572868285348694e-06,
+      "loss": 0.2729,
+      "step": 1705
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.550963380072164e-06,
+      "loss": 0.2624,
+      "step": 1706
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.529078883938745e-06,
+      "loss": 0.2669,
+      "step": 1707
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.50721483595991e-06,
+      "loss": 0.2843,
+      "step": 1708
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.485371275110683e-06,
+      "loss": 0.2906,
+      "step": 1709
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.463548240329567e-06,
+      "loss": 0.3207,
+      "step": 1710
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.441745770518495e-06,
+      "loss": 0.3003,
+      "step": 1711
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.419963904542696e-06,
+      "loss": 0.2884,
+      "step": 1712
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.398202681230702e-06,
+      "loss": 0.282,
+      "step": 1713
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.376462139374262e-06,
+      "loss": 0.2817,
+      "step": 1714
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.354742317728213e-06,
+      "loss": 0.2816,
+      "step": 1715
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.333043255010484e-06,
+      "loss": 0.289,
+      "step": 1716
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.311364989902006e-06,
+      "loss": 0.2973,
+      "step": 1717
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.289707561046632e-06,
+      "loss": 0.2583,
+      "step": 1718
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.26807100705104e-06,
+      "loss": 0.2733,
+      "step": 1719
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.24645536648474e-06,
+      "loss": 0.3237,
+      "step": 1720
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.224860677879951e-06,
+      "loss": 0.2717,
+      "step": 1721
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.20328697973151e-06,
+      "loss": 0.3108,
+      "step": 1722
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.181734310496886e-06,
+      "loss": 0.2951,
+      "step": 1723
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.160202708596033e-06,
+      "loss": 0.2839,
+      "step": 1724
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.13869221241134e-06,
+      "loss": 0.2938,
+      "step": 1725
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.117202860287595e-06,
+      "loss": 0.2841,
+      "step": 1726
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.095734690531891e-06,
+      "loss": 0.3184,
+      "step": 1727
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.074287741413564e-06,
+      "loss": 0.2912,
+      "step": 1728
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.052862051164095e-06,
+      "loss": 0.2889,
+      "step": 1729
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.031457657977103e-06,
+      "loss": 0.2815,
+      "step": 1730
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.010074600008231e-06,
+      "loss": 0.2698,
+      "step": 1731
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.988712915375067e-06,
+      "loss": 0.3059,
+      "step": 1732
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.967372642157142e-06,
+      "loss": 0.302,
+      "step": 1733
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.946053818395785e-06,
+      "loss": 0.2769,
+      "step": 1734
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.924756482094103e-06,
+      "loss": 0.3028,
+      "step": 1735
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.90348067121689e-06,
+      "loss": 0.2923,
+      "step": 1736
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.88222642369058e-06,
+      "loss": 0.2611,
+      "step": 1737
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.86099377740316e-06,
+      "loss": 0.2944,
+      "step": 1738
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.839782770204116e-06,
+      "loss": 0.249,
+      "step": 1739
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.818593439904351e-06,
+      "loss": 0.3006,
+      "step": 1740
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.79742582427614e-06,
+      "loss": 0.279,
+      "step": 1741
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.776279961053038e-06,
+      "loss": 0.2764,
+      "step": 1742
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.755155887929826e-06,
+      "loss": 0.2847,
+      "step": 1743
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.734053642562444e-06,
+      "loss": 0.2884,
+      "step": 1744
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.712973262567927e-06,
+      "loss": 0.2578,
+      "step": 1745
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.69191478552432e-06,
+      "loss": 0.3055,
+      "step": 1746
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.670878248970632e-06,
+      "loss": 0.2817,
+      "step": 1747
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.649863690406762e-06,
+      "loss": 0.2721,
+      "step": 1748
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.628871147293429e-06,
+      "loss": 0.2653,
+      "step": 1749
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.607900657052101e-06,
+      "loss": 0.2801,
+      "step": 1750
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.5869522570649415e-06,
+      "loss": 0.3177,
+      "step": 1751
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.566025984674734e-06,
+      "loss": 0.3032,
+      "step": 1752
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.54512187718482e-06,
+      "loss": 0.2589,
+      "step": 1753
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.524239971859022e-06,
+      "loss": 0.2967,
+      "step": 1754
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.503380305921592e-06,
+      "loss": 0.2952,
+      "step": 1755
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.482542916557136e-06,
+      "loss": 0.264,
+      "step": 1756
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.4617278409105395e-06,
+      "loss": 0.2747,
+      "step": 1757
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.440935116086942e-06,
+      "loss": 0.2948,
+      "step": 1758
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.420164779151599e-06,
+      "loss": 0.2406,
+      "step": 1759
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.399416867129878e-06,
+      "loss": 0.2698,
+      "step": 1760
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.378691417007191e-06,
+      "loss": 0.2783,
+      "step": 1761
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.357988465728867e-06,
+      "loss": 0.2549,
+      "step": 1762
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.3373080502001516e-06,
+      "loss": 0.3228,
+      "step": 1763
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.316650207286138e-06,
+      "loss": 0.2833,
+      "step": 1764
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.296014973811638e-06,
+      "loss": 0.2535,
+      "step": 1765
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.27540238656119e-06,
+      "loss": 0.3028,
+      "step": 1766
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.254812482278948e-06,
+      "loss": 0.2764,
+      "step": 1767
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.234245297668656e-06,
+      "loss": 0.312,
+      "step": 1768
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.213700869393525e-06,
+      "loss": 0.2723,
+      "step": 1769
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.193179234076208e-06,
+      "loss": 0.2728,
+      "step": 1770
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.172680428298759e-06,
+      "loss": 0.2545,
+      "step": 1771
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.152204488602492e-06,
+      "loss": 0.2721,
+      "step": 1772
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.131751451487976e-06,
+      "loss": 0.2441,
+      "step": 1773
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.111321353414968e-06,
+      "loss": 0.2655,
+      "step": 1774
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.0909142308023286e-06,
+      "loss": 0.2604,
+      "step": 1775
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.0705301200279295e-06,
+      "loss": 0.2762,
+      "step": 1776
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.05016905742867e-06,
+      "loss": 0.2769,
+      "step": 1777
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.029831079300342e-06,
+      "loss": 0.2578,
+      "step": 1778
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.009516221897565e-06,
+      "loss": 0.2852,
+      "step": 1779
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.989224521433784e-06,
+      "loss": 0.3015,
+      "step": 1780
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.968956014081145e-06,
+      "loss": 0.2689,
+      "step": 1781
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.94871073597043e-06,
+      "loss": 0.3012,
+      "step": 1782
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.928488723191049e-06,
+      "loss": 0.317,
+      "step": 1783
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.908290011790919e-06,
+      "loss": 0.2776,
+      "step": 1784
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.88811463777642e-06,
+      "loss": 0.2811,
+      "step": 1785
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.8679626371123335e-06,
+      "loss": 0.2692,
+      "step": 1786
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.8478340457217725e-06,
+      "loss": 0.3092,
+      "step": 1787
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.827728899486135e-06,
+      "loss": 0.2874,
+      "step": 1788
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.80764723424499e-06,
+      "loss": 0.2635,
+      "step": 1789
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.787589085796092e-06,
+      "loss": 0.3089,
+      "step": 1790
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.767554489895252e-06,
+      "loss": 0.2935,
+      "step": 1791
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.747543482256294e-06,
+      "loss": 0.2674,
+      "step": 1792
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.727556098551e-06,
+      "loss": 0.2698,
+      "step": 1793
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.70759237440904e-06,
+      "loss": 0.2732,
+      "step": 1794
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.6876523454179055e-06,
+      "loss": 0.2878,
+      "step": 1795
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.667736047122848e-06,
+      "loss": 0.2918,
+      "step": 1796
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.647843515026822e-06,
+      "loss": 0.2867,
+      "step": 1797
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.627974784590407e-06,
+      "loss": 0.2723,
+      "step": 1798
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.608129891231763e-06,
+      "loss": 0.2768,
+      "step": 1799
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.588308870326549e-06,
+      "loss": 0.278,
+      "step": 1800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.568511757207881e-06,
+      "loss": 0.2568,
+      "step": 1801
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.548738587166241e-06,
+      "loss": 0.3071,
+      "step": 1802
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.528989395449445e-06,
+      "loss": 0.2776,
+      "step": 1803
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.509264217262555e-06,
+      "loss": 0.2866,
+      "step": 1804
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.489563087767827e-06,
+      "loss": 0.25,
+      "step": 1805
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.469886042084655e-06,
+      "loss": 0.2529,
+      "step": 1806
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.45023311528949e-06,
+      "loss": 0.2955,
+      "step": 1807
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.430604342415801e-06,
+      "loss": 0.2749,
+      "step": 1808
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.4109997584539905e-06,
+      "loss": 0.2842,
+      "step": 1809
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.391419398351344e-06,
+      "loss": 0.251,
+      "step": 1810
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.371863297011966e-06,
+      "loss": 0.3003,
+      "step": 1811
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.352331489296719e-06,
+      "loss": 0.272,
+      "step": 1812
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.332824010023157e-06,
+      "loss": 0.3154,
+      "step": 1813
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.313340893965463e-06,
+      "loss": 0.2706,
+      "step": 1814
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.293882175854407e-06,
+      "loss": 0.2999,
+      "step": 1815
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.2744478903772374e-06,
+      "loss": 0.3017,
+      "step": 1816
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.255038072177664e-06,
+      "loss": 0.2617,
+      "step": 1817
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.235652755855803e-06,
+      "loss": 0.2587,
+      "step": 1818
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.216291975968054e-06,
+      "loss": 0.2633,
+      "step": 1819
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.196955767027093e-06,
+      "loss": 0.3255,
+      "step": 1820
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.177644163501819e-06,
+      "loss": 0.2813,
+      "step": 1821
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.1583571998172306e-06,
+      "loss": 0.289,
+      "step": 1822
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.139094910354422e-06,
+      "loss": 0.2917,
+      "step": 1823
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.119857329450513e-06,
+      "loss": 0.2579,
+      "step": 1824
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.100644491398571e-06,
+      "loss": 0.2761,
+      "step": 1825
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.08145643044753e-06,
+      "loss": 0.2503,
+      "step": 1826
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.062293180802199e-06,
+      "loss": 0.2413,
+      "step": 1827
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.043154776623141e-06,
+      "loss": 0.2728,
+      "step": 1828
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.024041252026611e-06,
+      "loss": 0.2599,
+      "step": 1829
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.004952641084529e-06,
+      "loss": 0.2873,
+      "step": 1830
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.985888977824417e-06,
+      "loss": 0.297,
+      "step": 1831
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.966850296229312e-06,
+      "loss": 0.309,
+      "step": 1832
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.947836630237698e-06,
+      "loss": 0.2925,
+      "step": 1833
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.928848013743505e-06,
+      "loss": 0.2555,
+      "step": 1834
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.909884480595993e-06,
+      "loss": 0.3032,
+      "step": 1835
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.890946064599685e-06,
+      "loss": 0.2721,
+      "step": 1836
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.872032799514371e-06,
+      "loss": 0.2958,
+      "step": 1837
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.853144719054986e-06,
+      "loss": 0.2514,
+      "step": 1838
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.834281856891555e-06,
+      "loss": 0.2747,
+      "step": 1839
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.815444246649178e-06,
+      "loss": 0.2579,
+      "step": 1840
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.796631921907927e-06,
+      "loss": 0.2482,
+      "step": 1841
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.777844916202799e-06,
+      "loss": 0.2905,
+      "step": 1842
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.759083263023659e-06,
+      "loss": 0.2547,
+      "step": 1843
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.740346995815178e-06,
+      "loss": 0.2769,
+      "step": 1844
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.721636147976774e-06,
+      "loss": 0.2848,
+      "step": 1845
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.702950752862553e-06,
+      "loss": 0.2646,
+      "step": 1846
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.684290843781244e-06,
+      "loss": 0.294,
+      "step": 1847
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.665656453996153e-06,
+      "loss": 0.3034,
+      "step": 1848
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.6470476167250876e-06,
+      "loss": 0.2933,
+      "step": 1849
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.628464365140309e-06,
+      "loss": 0.2626,
+      "step": 1850
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.609906732368471e-06,
+      "loss": 0.2784,
+      "step": 1851
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.591374751490551e-06,
+      "loss": 0.2875,
+      "step": 1852
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.572868455541809e-06,
+      "loss": 0.2614,
+      "step": 1853
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.554387877511711e-06,
+      "loss": 0.2772,
+      "step": 1854
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.535933050343883e-06,
+      "loss": 0.2969,
+      "step": 1855
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.5175040069360456e-06,
+      "loss": 0.283,
+      "step": 1856
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.499100780139956e-06,
+      "loss": 0.2833,
+      "step": 1857
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.480723402761356e-06,
+      "loss": 0.2657,
+      "step": 1858
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.462371907559898e-06,
+      "loss": 0.2661,
+      "step": 1859
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.444046327249106e-06,
+      "loss": 0.2872,
+      "step": 1860
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4257466944963035e-06,
+      "loss": 0.2711,
+      "step": 1861
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.407473041922557e-06,
+      "loss": 0.2747,
+      "step": 1862
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.3892254021026295e-06,
+      "loss": 0.308,
+      "step": 1863
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.371003807564899e-06,
+      "loss": 0.2658,
+      "step": 1864
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.3528082907913406e-06,
+      "loss": 0.2522,
+      "step": 1865
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.334638884217409e-06,
+      "loss": 0.2756,
+      "step": 1866
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.316495620232034e-06,
+      "loss": 0.2745,
+      "step": 1867
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.298378531177557e-06,
+      "loss": 0.2627,
+      "step": 1868
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.280287649349627e-06,
+      "loss": 0.2695,
+      "step": 1869
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.262223006997198e-06,
+      "loss": 0.2511,
+      "step": 1870
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.24418463632244e-06,
+      "loss": 0.2942,
+      "step": 1871
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.226172569480712e-06,
+      "loss": 0.2721,
+      "step": 1872
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.2081868385804465e-06,
+      "loss": 0.2678,
+      "step": 1873
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.190227475683154e-06,
+      "loss": 0.2806,
+      "step": 1874
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.172294512803349e-06,
+      "loss": 0.2717,
+      "step": 1875
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.154387981908459e-06,
+      "loss": 0.2619,
+      "step": 1876
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.136507914918807e-06,
+      "loss": 0.2875,
+      "step": 1877
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.118654343707559e-06,
+      "loss": 0.2652,
+      "step": 1878
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.100827300100615e-06,
+      "loss": 0.2974,
+      "step": 1879
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.083026815876602e-06,
+      "loss": 0.2529,
+      "step": 1880
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.065252922766814e-06,
+      "loss": 0.2879,
+      "step": 1881
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.047505652455138e-06,
+      "loss": 0.2743,
+      "step": 1882
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.029785036577976e-06,
+      "loss": 0.2768,
+      "step": 1883
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.012091106724262e-06,
+      "loss": 0.2685,
+      "step": 1884
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.994423894435327e-06,
+      "loss": 0.3099,
+      "step": 1885
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.97678343120487e-06,
+      "loss": 0.2884,
+      "step": 1886
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.959169748478938e-06,
+      "loss": 0.2546,
+      "step": 1887
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.941582877655813e-06,
+      "loss": 0.2937,
+      "step": 1888
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.924022850085992e-06,
+      "loss": 0.2789,
+      "step": 1889
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.906489697072116e-06,
+      "loss": 0.2614,
+      "step": 1890
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.888983449868927e-06,
+      "loss": 0.3095,
+      "step": 1891
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.871504139683199e-06,
+      "loss": 0.2542,
+      "step": 1892
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.854051797673676e-06,
+      "loss": 0.2669,
+      "step": 1893
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.836626454951056e-06,
+      "loss": 0.2844,
+      "step": 1894
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.8192281425778876e-06,
+      "loss": 0.2624,
+      "step": 1895
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.801856891568537e-06,
+      "loss": 0.2977,
+      "step": 1896
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.784512732889137e-06,
+      "loss": 0.2816,
+      "step": 1897
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.767195697457523e-06,
+      "loss": 0.2686,
+      "step": 1898
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.749905816143174e-06,
+      "loss": 0.2978,
+      "step": 1899
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.732643119767173e-06,
+      "loss": 0.3001,
+      "step": 1900
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.715407639102136e-06,
+      "loss": 0.2651,
+      "step": 1901
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.698199404872172e-06,
+      "loss": 0.3281,
+      "step": 1902
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.68101844775281e-06,
+      "loss": 0.2575,
+      "step": 1903
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6638647983709605e-06,
+      "loss": 0.2663,
+      "step": 1904
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6467384873048585e-06,
+      "loss": 0.2856,
+      "step": 1905
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.629639545083997e-06,
+      "loss": 0.27,
+      "step": 1906
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.612568002189086e-06,
+      "loss": 0.2508,
+      "step": 1907
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.595523889051996e-06,
+      "loss": 0.2761,
+      "step": 1908
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.578507236055694e-06,
+      "loss": 0.3073,
+      "step": 1909
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.561518073534201e-06,
+      "loss": 0.2762,
+      "step": 1910
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5445564317725285e-06,
+      "loss": 0.2634,
+      "step": 1911
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.527622341006648e-06,
+      "loss": 0.2968,
+      "step": 1912
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.510715831423387e-06,
+      "loss": 0.2736,
+      "step": 1913
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.493836933160429e-06,
+      "loss": 0.2583,
+      "step": 1914
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.476985676306231e-06,
+      "loss": 0.2776,
+      "step": 1915
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.4601620908999755e-06,
+      "loss": 0.2537,
+      "step": 1916
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.443366206931523e-06,
+      "loss": 0.2663,
+      "step": 1917
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.426598054341344e-06,
+      "loss": 0.2948,
+      "step": 1918
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.409857663020482e-06,
+      "loss": 0.249,
+      "step": 1919
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.393145062810491e-06,
+      "loss": 0.2522,
+      "step": 1920
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.376460283503377e-06,
+      "loss": 0.2527,
+      "step": 1921
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.359803354841577e-06,
+      "loss": 0.2565,
+      "step": 1922
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.343174306517848e-06,
+      "loss": 0.2868,
+      "step": 1923
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.326573168175261e-06,
+      "loss": 0.2667,
+      "step": 1924
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.309999969407155e-06,
+      "loss": 0.2909,
+      "step": 1925
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.293454739757026e-06,
+      "loss": 0.2392,
+      "step": 1926
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.276937508718533e-06,
+      "loss": 0.2775,
+      "step": 1927
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.260448305735434e-06,
+      "loss": 0.2875,
+      "step": 1928
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.24398716020151e-06,
+      "loss": 0.2627,
+      "step": 1929
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.227554101460513e-06,
+      "loss": 0.3135,
+      "step": 1930
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.211149158806163e-06,
+      "loss": 0.2856,
+      "step": 1931
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.194772361482038e-06,
+      "loss": 0.2642,
+      "step": 1932
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.178423738681534e-06,
+      "loss": 0.2816,
+      "step": 1933
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1621033195478366e-06,
+      "loss": 0.2876,
+      "step": 1934
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.145811133173865e-06,
+      "loss": 0.3069,
+      "step": 1935
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.129547208602198e-06,
+      "loss": 0.3124,
+      "step": 1936
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.113311574825023e-06,
+      "loss": 0.2589,
+      "step": 1937
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.097104260784121e-06,
+      "loss": 0.2537,
+      "step": 1938
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.080925295370781e-06,
+      "loss": 0.289,
+      "step": 1939
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0647747074257385e-06,
+      "loss": 0.2715,
+      "step": 1940
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.048652525739172e-06,
+      "loss": 0.2886,
+      "step": 1941
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0325587790506126e-06,
+      "loss": 0.2778,
+      "step": 1942
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.016493496048885e-06,
+      "loss": 0.2268,
+      "step": 1943
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0004567053721025e-06,
+      "loss": 0.2864,
+      "step": 1944
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.984448435607573e-06,
+      "loss": 0.2803,
+      "step": 1945
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9684687152917625e-06,
+      "loss": 0.2678,
+      "step": 1946
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.952517572910246e-06,
+      "loss": 0.2754,
+      "step": 1947
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.936595036897657e-06,
+      "loss": 0.2768,
+      "step": 1948
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9207011356376345e-06,
+      "loss": 0.2814,
+      "step": 1949
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.904835897462771e-06,
+      "loss": 0.2761,
+      "step": 1950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.8889993506545675e-06,
+      "loss": 0.2873,
+      "step": 1951
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.873191523443376e-06,
+      "loss": 0.2883,
+      "step": 1952
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.857412444008353e-06,
+      "loss": 0.2694,
+      "step": 1953
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8416621404774135e-06,
+      "loss": 0.2808,
+      "step": 1954
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.825940640927168e-06,
+      "loss": 0.2872,
+      "step": 1955
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.81024797338289e-06,
+      "loss": 0.2581,
+      "step": 1956
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7945841658184533e-06,
+      "loss": 0.2772,
+      "step": 1957
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7789492461562806e-06,
+      "loss": 0.2982,
+      "step": 1958
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7633432422673054e-06,
+      "loss": 0.2452,
+      "step": 1959
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7477661819709153e-06,
+      "loss": 0.2648,
+      "step": 1960
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7322180930348985e-06,
+      "loss": 0.2613,
+      "step": 1961
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7166990031754034e-06,
+      "loss": 0.2817,
+      "step": 1962
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7012089400568796e-06,
+      "loss": 0.2679,
+      "step": 1963
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6857479312920363e-06,
+      "loss": 0.2833,
+      "step": 1964
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.670316004441792e-06,
+      "loss": 0.2637,
+      "step": 1965
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6549131870152187e-06,
+      "loss": 0.2721,
+      "step": 1966
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6395395064695027e-06,
+      "loss": 0.2661,
+      "step": 1967
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.624194990209882e-06,
+      "loss": 0.2522,
+      "step": 1968
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6088796655896286e-06,
+      "loss": 0.2894,
+      "step": 1969
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.593593559909947e-06,
+      "loss": 0.2478,
+      "step": 1970
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5783367004199664e-06,
+      "loss": 0.2485,
+      "step": 1971
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5631091143167006e-06,
+      "loss": 0.2695,
+      "step": 1972
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.547910828744954e-06,
+      "loss": 0.2794,
+      "step": 1973
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.532741870797307e-06,
+      "loss": 0.2726,
+      "step": 1974
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.517602267514075e-06,
+      "loss": 0.2662,
+      "step": 1975
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5024920458832345e-06,
+      "loss": 0.2963,
+      "step": 1976
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.4874112328403744e-06,
+      "loss": 0.2797,
+      "step": 1977
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4723598552686722e-06,
+      "loss": 0.2367,
+      "step": 1978
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4573379399988484e-06,
+      "loss": 0.2752,
+      "step": 1979
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.442345513809071e-06,
+      "loss": 0.295,
+      "step": 1980
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.427382603424958e-06,
+      "loss": 0.2824,
+      "step": 1981
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.412449235519528e-06,
+      "loss": 0.2596,
+      "step": 1982
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3975454367131032e-06,
+      "loss": 0.284,
+      "step": 1983
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.382671233573316e-06,
+      "loss": 0.2777,
+      "step": 1984
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.367826652615045e-06,
+      "loss": 0.2983,
+      "step": 1985
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.353011720300361e-06,
+      "loss": 0.2903,
+      "step": 1986
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.338226463038461e-06,
+      "loss": 0.266,
+      "step": 1987
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.32347090718568e-06,
+      "loss": 0.2748,
+      "step": 1988
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3087450790453835e-06,
+      "loss": 0.2877,
+      "step": 1989
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2940490048679386e-06,
+      "loss": 0.2959,
+      "step": 1990
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.279382710850696e-06,
+      "loss": 0.2954,
+      "step": 1991
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2647462231379045e-06,
+      "loss": 0.2667,
+      "step": 1992
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.250139567820678e-06,
+      "loss": 0.2899,
+      "step": 1993
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.235562770936962e-06,
+      "loss": 0.3007,
+      "step": 1994
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2210158584714635e-06,
+      "loss": 0.2458,
+      "step": 1995
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2064988563556334e-06,
+      "loss": 0.281,
+      "step": 1996
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1920117904675773e-06,
+      "loss": 0.2539,
+      "step": 1997
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.177554686632067e-06,
+      "loss": 0.279,
+      "step": 1998
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.163127570620452e-06,
+      "loss": 0.2745,
+      "step": 1999
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.148730468150607e-06,
+      "loss": 0.2731,
+      "step": 2000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1343634048869378e-06,
+      "loss": 0.277,
+      "step": 2001
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.1200264064402776e-06,
+      "loss": 0.2621,
+      "step": 2002
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.105719498367872e-06,
+      "loss": 0.2839,
+      "step": 2003
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.091442706173331e-06,
+      "loss": 0.2541,
+      "step": 2004
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0771960553065774e-06,
+      "loss": 0.2636,
+      "step": 2005
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.062979571163802e-06,
+      "loss": 0.2495,
+      "step": 2006
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.048793279087421e-06,
+      "loss": 0.2321,
+      "step": 2007
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0346372043660334e-06,
+      "loss": 0.2584,
+      "step": 2008
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.020511372234367e-06,
+      "loss": 0.2723,
+      "step": 2009
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.006415807873242e-06,
+      "loss": 0.2424,
+      "step": 2010
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9923505364095253e-06,
+      "loss": 0.2948,
+      "step": 2011
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9783155829160804e-06,
+      "loss": 0.2724,
+      "step": 2012
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.964310972411724e-06,
+      "loss": 0.288,
+      "step": 2013
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9503367298611853e-06,
+      "loss": 0.2715,
+      "step": 2014
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9363928801750563e-06,
+      "loss": 0.2529,
+      "step": 2015
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9224794482097672e-06,
+      "loss": 0.2734,
+      "step": 2016
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9085964587674987e-06,
+      "loss": 0.2765,
+      "step": 2017
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8947439365961826e-06,
+      "loss": 0.2971,
+      "step": 2018
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8809219063894313e-06,
+      "loss": 0.26,
+      "step": 2019
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.867130392786506e-06,
+      "loss": 0.2783,
+      "step": 2020
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8533694203722694e-06,
+      "loss": 0.2753,
+      "step": 2021
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.83963901367714e-06,
+      "loss": 0.2748,
+      "step": 2022
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8259391971770457e-06,
+      "loss": 0.2621,
+      "step": 2023
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.812269995293391e-06,
+      "loss": 0.2606,
+      "step": 2024
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.7986314323929973e-06,
+      "loss": 0.2494,
+      "step": 2025
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.785023532788089e-06,
+      "loss": 0.2604,
+      "step": 2026
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7714463207361997e-06,
+      "loss": 0.2614,
+      "step": 2027
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7578998204401773e-06,
+      "loss": 0.252,
+      "step": 2028
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7443840560481328e-06,
+      "loss": 0.2556,
+      "step": 2029
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7308990516533597e-06,
+      "loss": 0.2594,
+      "step": 2030
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7174448312943356e-06,
+      "loss": 0.2542,
+      "step": 2031
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.704021418954668e-06,
+      "loss": 0.2564,
+      "step": 2032
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6906288385630384e-06,
+      "loss": 0.2916,
+      "step": 2033
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.677267113993152e-06,
+      "loss": 0.2492,
+      "step": 2034
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.663936269063738e-06,
+      "loss": 0.2769,
+      "step": 2035
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.650636327538465e-06,
+      "loss": 0.2469,
+      "step": 2036
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6373673131259024e-06,
+      "loss": 0.2744,
+      "step": 2037
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6241292494795144e-06,
+      "loss": 0.2648,
+      "step": 2038
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.610922160197573e-06,
+      "loss": 0.2747,
+      "step": 2039
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5977460688231373e-06,
+      "loss": 0.2863,
+      "step": 2040
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5846009988440067e-06,
+      "loss": 0.2681,
+      "step": 2041
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5714869736926984e-06,
+      "loss": 0.2996,
+      "step": 2042
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.558404016746374e-06,
+      "loss": 0.2644,
+      "step": 2043
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5453521513268077e-06,
+      "loss": 0.2813,
+      "step": 2044
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5323314007003674e-06,
+      "loss": 0.2876,
+      "step": 2045
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.519341788077947e-06,
+      "loss": 0.3095,
+      "step": 2046
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.506383336614919e-06,
+      "loss": 0.2885,
+      "step": 2047
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.4934560694111355e-06,
+      "loss": 0.2569,
+      "step": 2048
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.480560009510842e-06,
+      "loss": 0.2537,
+      "step": 2049
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4676951799026518e-06,
+      "loss": 0.2626,
+      "step": 2050
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.454861603519516e-06,
+      "loss": 0.2741,
+      "step": 2051
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4420593032386686e-06,
+      "loss": 0.2682,
+      "step": 2052
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4292883018815893e-06,
+      "loss": 0.2969,
+      "step": 2053
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.41654862221397e-06,
+      "loss": 0.2749,
+      "step": 2054
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.40384028694566e-06,
+      "loss": 0.2886,
+      "step": 2055
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.391163318730638e-06,
+      "loss": 0.3018,
+      "step": 2056
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3785177401669724e-06,
+      "loss": 0.2959,
+      "step": 2057
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.365903573796766e-06,
+      "loss": 0.2824,
+      "step": 2058
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.353320842106137e-06,
+      "loss": 0.2755,
+      "step": 2059
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.340769567525156e-06,
+      "loss": 0.2572,
+      "step": 2060
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.328249772427831e-06,
+      "loss": 0.2813,
+      "step": 2061
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3157614791320415e-06,
+      "loss": 0.2908,
+      "step": 2062
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.303304709899521e-06,
+      "loss": 0.2464,
+      "step": 2063
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2908794869358044e-06,
+      "loss": 0.2965,
+      "step": 2064
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2784858323901914e-06,
+      "loss": 0.2593,
+      "step": 2065
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.266123768355708e-06,
+      "loss": 0.2574,
+      "step": 2066
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.253793316869071e-06,
+      "loss": 0.2778,
+      "step": 2067
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2414944999106346e-06,
+      "loss": 0.2716,
+      "step": 2068
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2292273394043718e-06,
+      "loss": 0.2671,
+      "step": 2069
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2169918572178185e-06,
+      "loss": 0.2755,
+      "step": 2070
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.204788075162041e-06,
+      "loss": 0.3187,
+      "step": 2071
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1926160149915955e-06,
+      "loss": 0.2876,
+      "step": 2072
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.180475698404503e-06,
+      "loss": 0.2806,
+      "step": 2073
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1683671470421765e-06,
+      "loss": 0.2737,
+      "step": 2074
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1562903824894123e-06,
+      "loss": 0.2819,
+      "step": 2075
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.144245426274363e-06,
+      "loss": 0.2608,
+      "step": 2076
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1322322998684485e-06,
+      "loss": 0.2457,
+      "step": 2077
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.120251024686364e-06,
+      "loss": 0.3047,
+      "step": 2078
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.108301622086035e-06,
+      "loss": 0.2594,
+      "step": 2079
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.09638411336855e-06,
+      "loss": 0.2611,
+      "step": 2080
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0844985197781577e-06,
+      "loss": 0.2807,
+      "step": 2081
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.072644862502209e-06,
+      "loss": 0.2563,
+      "step": 2082
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0608231626711374e-06,
+      "loss": 0.2468,
+      "step": 2083
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.049033441358386e-06,
+      "loss": 0.2845,
+      "step": 2084
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.037275719580407e-06,
+      "loss": 0.2731,
+      "step": 2085
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.025550018296618e-06,
+      "loss": 0.2921,
+      "step": 2086
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.013856358409336e-06,
+      "loss": 0.3076,
+      "step": 2087
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0021947607637714e-06,
+      "loss": 0.2881,
+      "step": 2088
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9905652461479864e-06,
+      "loss": 0.2746,
+      "step": 2089
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.978967835292849e-06,
+      "loss": 0.239,
+      "step": 2090
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.967402548871982e-06,
+      "loss": 0.294,
+      "step": 2091
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.955869407501767e-06,
+      "loss": 0.2383,
+      "step": 2092
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.944368431741275e-06,
+      "loss": 0.2662,
+      "step": 2093
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9328996420922206e-06,
+      "loss": 0.2656,
+      "step": 2094
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9214630589989756e-06,
+      "loss": 0.277,
+      "step": 2095
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9100587028484763e-06,
+      "loss": 0.2617,
+      "step": 2096
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8986865939702181e-06,
+      "loss": 0.2758,
+      "step": 2097
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8873467526362144e-06,
+      "loss": 0.2674,
+      "step": 2098
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8760391990609528e-06,
+      "loss": 0.2712,
+      "step": 2099
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8647639534013694e-06,
+      "loss": 0.2443,
+      "step": 2100
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8535210357568067e-06,
+      "loss": 0.2614,
+      "step": 2101
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8423104661689773e-06,
+      "loss": 0.2681,
+      "step": 2102
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8311322646219353e-06,
+      "loss": 0.255,
+      "step": 2103
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8199864510420218e-06,
+      "loss": 0.2575,
+      "step": 2104
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8088730452978586e-06,
+      "loss": 0.2895,
+      "step": 2105
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7977920672002901e-06,
+      "loss": 0.2462,
+      "step": 2106
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7867435365023555e-06,
+      "loss": 0.2687,
+      "step": 2107
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7757274728992535e-06,
+      "loss": 0.2715,
+      "step": 2108
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7647438960283026e-06,
+      "loss": 0.2719,
+      "step": 2109
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7537928254689163e-06,
+      "loss": 0.2653,
+      "step": 2110
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7428742807425592e-06,
+      "loss": 0.2403,
+      "step": 2111
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7319882813127176e-06,
+      "loss": 0.265,
+      "step": 2112
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7211348465848554e-06,
+      "loss": 0.2624,
+      "step": 2113
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.710313995906394e-06,
+      "loss": 0.2539,
+      "step": 2114
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6995257485666661e-06,
+      "loss": 0.2524,
+      "step": 2115
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.688770123796888e-06,
+      "loss": 0.2675,
+      "step": 2116
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6780471407701204e-06,
+      "loss": 0.2443,
+      "step": 2117
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.667356818601238e-06,
+      "loss": 0.2766,
+      "step": 2118
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6566991763468921e-06,
+      "loss": 0.2416,
+      "step": 2119
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6460742330054814e-06,
+      "loss": 0.2787,
+      "step": 2120
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6354820075171152e-06,
+      "loss": 0.2847,
+      "step": 2121
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6249225187635765e-06,
+      "loss": 0.2813,
+      "step": 2122
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6143957855682924e-06,
+      "loss": 0.2715,
+      "step": 2123
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6039018266963014e-06,
+      "loss": 0.2683,
+      "step": 2124
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5934406608542196e-06,
+      "loss": 0.2671,
+      "step": 2125
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5830123066901992e-06,
+      "loss": 0.2709,
+      "step": 2126
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.572616782793912e-06,
+      "loss": 0.2612,
+      "step": 2127
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5622541076964992e-06,
+      "loss": 0.2612,
+      "step": 2128
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.551924299870542e-06,
+      "loss": 0.2787,
+      "step": 2129
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5416273777300483e-06,
+      "loss": 0.2633,
+      "step": 2130
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5313633596303846e-06,
+      "loss": 0.2946,
+      "step": 2131
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5211322638682701e-06,
+      "loss": 0.2759,
+      "step": 2132
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.510934108681743e-06,
+      "loss": 0.2423,
+      "step": 2133
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5007689122501123e-06,
+      "loss": 0.2604,
+      "step": 2134
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.490636692693932e-06,
+      "loss": 0.2471,
+      "step": 2135
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.480537468074985e-06,
+      "loss": 0.2607,
+      "step": 2136
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4704712563962287e-06,
+      "loss": 0.2717,
+      "step": 2137
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.46043807560176e-06,
+      "loss": 0.2636,
+      "step": 2138
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4504379435768212e-06,
+      "loss": 0.2674,
+      "step": 2139
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4404708781477217e-06,
+      "loss": 0.2344,
+      "step": 2140
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4305368970818223e-06,
+      "loss": 0.3194,
+      "step": 2141
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4206360180875267e-06,
+      "loss": 0.2587,
+      "step": 2142
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4107682588142214e-06,
+      "loss": 0.2499,
+      "step": 2143
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.400933636852242e-06,
+      "loss": 0.2561,
+      "step": 2144
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3911321697328673e-06,
+      "loss": 0.2749,
+      "step": 2145
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3813638749282721e-06,
+      "loss": 0.266,
+      "step": 2146
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3716287698514985e-06,
+      "loss": 0.2705,
+      "step": 2147
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3619268718564138e-06,
+      "loss": 0.3086,
+      "step": 2148
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3522581982377036e-06,
+      "loss": 0.268,
+      "step": 2149
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3426227662308233e-06,
+      "loss": 0.2536,
+      "step": 2150
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3330205930119623e-06,
+      "loss": 0.2672,
+      "step": 2151
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3234516956980391e-06,
+      "loss": 0.2454,
+      "step": 2152
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3139160913466409e-06,
+      "loss": 0.278,
+      "step": 2153
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3044137969560143e-06,
+      "loss": 0.2755,
+      "step": 2154
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2949448294650213e-06,
+      "loss": 0.265,
+      "step": 2155
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2855092057531193e-06,
+      "loss": 0.2617,
+      "step": 2156
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.276106942640325e-06,
+      "loss": 0.2795,
+      "step": 2157
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2667380568871868e-06,
+      "loss": 0.2623,
+      "step": 2158
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2574025651947586e-06,
+      "loss": 0.2633,
+      "step": 2159
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2481004842045553e-06,
+      "loss": 0.2728,
+      "step": 2160
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2388318304985436e-06,
+      "loss": 0.2714,
+      "step": 2161
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2295966205990984e-06,
+      "loss": 0.265,
+      "step": 2162
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2203948709689773e-06,
+      "loss": 0.253,
+      "step": 2163
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2112265980112925e-06,
+      "loss": 0.2736,
+      "step": 2164
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2020918180694796e-06,
+      "loss": 0.2637,
+      "step": 2165
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1929905474272707e-06,
+      "loss": 0.2596,
+      "step": 2166
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1839228023086592e-06,
+      "loss": 0.2638,
+      "step": 2167
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1748885988778814e-06,
+      "loss": 0.2595,
+      "step": 2168
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1658879532393797e-06,
+      "loss": 0.2639,
+      "step": 2169
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1569208814377775e-06,
+      "loss": 0.2765,
+      "step": 2170
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.147987399457844e-06,
+      "loss": 0.3036,
+      "step": 2171
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1390875232244781e-06,
+      "loss": 0.2905,
+      "step": 2172
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1302212686026693e-06,
+      "loss": 0.2752,
+      "step": 2173
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1213886513974704e-06,
+      "loss": 0.2839,
+      "step": 2174
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1125896873539777e-06,
+      "loss": 0.2564,
+      "step": 2175
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1038243921572888e-06,
+      "loss": 0.2868,
+      "step": 2176
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0950927814324985e-06,
+      "loss": 0.228,
+      "step": 2177
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0863948707446359e-06,
+      "loss": 0.2758,
+      "step": 2178
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.077730675598665e-06,
+      "loss": 0.2724,
+      "step": 2179
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0691002114394577e-06,
+      "loss": 0.2952,
+      "step": 2180
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0605034936517388e-06,
+      "loss": 0.2355,
+      "step": 2181
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0519405375600855e-06,
+      "loss": 0.2847,
+      "step": 2182
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0434113584288964e-06,
+      "loss": 0.2783,
+      "step": 2183
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0349159714623492e-06,
+      "loss": 0.2659,
+      "step": 2184
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.02645439180439e-06,
+      "loss": 0.2342,
+      "step": 2185
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0180266345386936e-06,
+      "loss": 0.2429,
+      "step": 2186
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0096327146886553e-06,
+      "loss": 0.2652,
+      "step": 2187
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0012726472173351e-06,
+      "loss": 0.2777,
+      "step": 2188
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.929464470274541e-07,
+      "loss": 0.2595,
+      "step": 2189
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.846541289613709e-07,
+      "loss": 0.2416,
+      "step": 2190
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.763957078010321e-07,
+      "loss": 0.274,
+      "step": 2191
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.681711982679597e-07,
+      "loss": 0.278,
+      "step": 2192
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.59980615023237e-07,
+      "loss": 0.2617,
+      "step": 2193
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.51823972667465e-07,
+      "loss": 0.2719,
+      "step": 2194
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.437012857407257e-07,
+      "loss": 0.2977,
+      "step": 2195
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.35612568722597e-07,
+      "loss": 0.2712,
+      "step": 2196
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.275578360320869e-07,
+      "loss": 0.269,
+      "step": 2197
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.19537102027619e-07,
+      "loss": 0.2976,
+      "step": 2198
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.11550381007027e-07,
+      "loss": 0.2654,
+      "step": 2199
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.035976872074981e-07,
+      "loss": 0.2847,
+      "step": 2200
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.956790348055633e-07,
+      "loss": 0.2368,
+      "step": 2201
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.8779443791708e-07,
+      "loss": 0.2743,
+      "step": 2202
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.799439105971941e-07,
+      "loss": 0.2909,
+      "step": 2203
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.721274668403134e-07,
+      "loss": 0.2877,
+      "step": 2204
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.643451205800946e-07,
+      "loss": 0.3083,
+      "step": 2205
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.565968856894091e-07,
+      "loss": 0.2737,
+      "step": 2206
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.488827759803242e-07,
+      "loss": 0.2641,
+      "step": 2207
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.412028052040644e-07,
+      "loss": 0.2524,
+      "step": 2208
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.335569870510162e-07,
+      "loss": 0.2732,
+      "step": 2209
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.259453351506686e-07,
+      "loss": 0.2793,
+      "step": 2210
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.183678630716185e-07,
+      "loss": 0.2741,
+      "step": 2211
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.108245843215213e-07,
+      "loss": 0.2676,
+      "step": 2212
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.033155123470893e-07,
+      "loss": 0.2811,
+      "step": 2213
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.958406605340507e-07,
+      "loss": 0.2821,
+      "step": 2214
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.884000422071381e-07,
+      "loss": 0.2673,
+      "step": 2215
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.809936706300569e-07,
+      "loss": 0.2742,
+      "step": 2216
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.736215590054596e-07,
+      "loss": 0.2563,
+      "step": 2217
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.66283720474934e-07,
+      "loss": 0.2243,
+      "step": 2218
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.589801681189679e-07,
+      "loss": 0.2938,
+      "step": 2219
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.517109149569313e-07,
+      "loss": 0.2431,
+      "step": 2220
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.44475973947052e-07,
+      "loss": 0.2781,
+      "step": 2221
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.372753579863956e-07,
+      "loss": 0.2794,
+      "step": 2222
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.301090799108323e-07,
+      "loss": 0.2765,
+      "step": 2223
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.229771524950302e-07,
+      "loss": 0.2913,
+      "step": 2224
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.158795884524172e-07,
+      "loss": 0.2737,
+      "step": 2225
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.088164004351661e-07,
+      "loss": 0.2758,
+      "step": 2226
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.017876010341784e-07,
+      "loss": 0.2666,
+      "step": 2227
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.947932027790382e-07,
+      "loss": 0.2886,
+      "step": 2228
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.87833218138021e-07,
+      "loss": 0.2629,
+      "step": 2229
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.809076595180486e-07,
+      "loss": 0.2698,
+      "step": 2230
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.740165392646747e-07,
+      "loss": 0.2584,
+      "step": 2231
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.671598696620684e-07,
+      "loss": 0.2555,
+      "step": 2232
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.603376629329772e-07,
+      "loss": 0.2581,
+      "step": 2233
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.535499312387283e-07,
+      "loss": 0.2903,
+      "step": 2234
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.46796686679183e-07,
+      "loss": 0.2688,
+      "step": 2235
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.400779412927227e-07,
+      "loss": 0.2931,
+      "step": 2236
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.333937070562468e-07,
+      "loss": 0.2915,
+      "step": 2237
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.267439958851151e-07,
+      "loss": 0.282,
+      "step": 2238
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.201288196331567e-07,
+      "loss": 0.3035,
+      "step": 2239
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.135481900926432e-07,
+      "loss": 0.2441,
+      "step": 2240
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.070021189942488e-07,
+      "loss": 0.2886,
+      "step": 2241
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.004906180070525e-07,
+      "loss": 0.2591,
+      "step": 2242
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.940136987385137e-07,
+      "loss": 0.2703,
+      "step": 2243
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.875713727344368e-07,
+      "loss": 0.3165,
+      "step": 2244
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.811636514789598e-07,
+      "loss": 0.266,
+      "step": 2245
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.747905463945392e-07,
+      "loss": 0.2836,
+      "step": 2246
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.684520688419271e-07,
+      "loss": 0.2593,
+      "step": 2247
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.621482301201364e-07,
+      "loss": 0.2969,
+      "step": 2248
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.558790414664405e-07,
+      "loss": 0.284,
+      "step": 2249
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.496445140563489e-07,
+      "loss": 0.2883,
+      "step": 2250
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.434446590035803e-07,
+      "loss": 0.3015,
+      "step": 2251
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.372794873600384e-07,
+      "loss": 0.2772,
+      "step": 2252
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.311490101158124e-07,
+      "loss": 0.2513,
+      "step": 2253
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.250532381991403e-07,
+      "loss": 0.2679,
+      "step": 2254
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.189921824763855e-07,
+      "loss": 0.2666,
+      "step": 2255
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.129658537520432e-07,
+      "loss": 0.2535,
+      "step": 2256
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.069742627686913e-07,
+      "loss": 0.2907,
+      "step": 2257
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.01017420206984e-07,
+      "loss": 0.2771,
+      "step": 2258
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.950953366856382e-07,
+      "loss": 0.2475,
+      "step": 2259
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.892080227614093e-07,
+      "loss": 0.2616,
+      "step": 2260
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.833554889290648e-07,
+      "loss": 0.2638,
+      "step": 2261
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.775377456213837e-07,
+      "loss": 0.2896,
+      "step": 2262
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.7175480320911457e-07,
+      "loss": 0.2649,
+      "step": 2263
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6600667200098036e-07,
+      "loss": 0.2595,
+      "step": 2264
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.602933622436445e-07,
+      "loss": 0.275,
+      "step": 2265
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.54614884121698e-07,
+      "loss": 0.268,
+      "step": 2266
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.489712477576391e-07,
+      "loss": 0.2645,
+      "step": 2267
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.433624632118627e-07,
+      "loss": 0.2708,
+      "step": 2268
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.377885404826265e-07,
+      "loss": 0.267,
+      "step": 2269
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.322494895060536e-07,
+      "loss": 0.2461,
+      "step": 2270
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.2674532015610113e-07,
+      "loss": 0.2646,
+      "step": 2271
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.2127604224454054e-07,
+      "loss": 0.2468,
+      "step": 2272
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.1584166552095073e-07,
+      "loss": 0.2847,
+      "step": 2273
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.1044219967269816e-07,
+      "loss": 0.268,
+      "step": 2274
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.050776543249102e-07,
+      "loss": 0.2624,
+      "step": 2275
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.997480390404684e-07,
+      "loss": 0.2707,
+      "step": 2276
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.944533633199865e-07,
+      "loss": 0.2913,
+      "step": 2277
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8919363660179675e-07,
+      "loss": 0.2454,
+      "step": 2278
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8396886826192804e-07,
+      "loss": 0.247,
+      "step": 2279
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.7877906761409233e-07,
+      "loss": 0.2624,
+      "step": 2280
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.7362424390967375e-07,
+      "loss": 0.2684,
+      "step": 2281
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.685044063376997e-07,
+      "loss": 0.2828,
+      "step": 2282
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.6341956402482946e-07,
+      "loss": 0.2783,
+      "step": 2283
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.5836972603535024e-07,
+      "loss": 0.3,
+      "step": 2284
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.533549013711368e-07,
+      "loss": 0.2557,
+      "step": 2285
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.483750989716561e-07,
+      "loss": 0.2501,
+      "step": 2286
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4343032771394726e-07,
+      "loss": 0.2536,
+      "step": 2287
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.385205964125948e-07,
+      "loss": 0.2937,
+      "step": 2288
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.336459138197246e-07,
+      "loss": 0.2511,
+      "step": 2289
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.288062886249854e-07,
+      "loss": 0.2959,
+      "step": 2290
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.2400172945553423e-07,
+      "loss": 0.274,
+      "step": 2291
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.1923224487601325e-07,
+      "loss": 0.2711,
+      "step": 2292
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.144978433885393e-07,
+      "loss": 0.2602,
+      "step": 2293
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.097985334327014e-07,
+      "loss": 0.2701,
+      "step": 2294
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0513432338552307e-07,
+      "loss": 0.2986,
+      "step": 2295
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.005052215614601e-07,
+      "loss": 0.2809,
+      "step": 2296
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.959112362123895e-07,
+      "loss": 0.2898,
+      "step": 2297
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.913523755275849e-07,
+      "loss": 0.2632,
+      "step": 2298
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8682864763370346e-07,
+      "loss": 0.2798,
+      "step": 2299
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.823400605947835e-07,
+      "loss": 0.2554,
+      "step": 2300
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7788662241221566e-07,
+      "loss": 0.2636,
+      "step": 2301
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7346834102472965e-07,
+      "loss": 0.2442,
+      "step": 2302
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.690852243083919e-07,
+      "loss": 0.277,
+      "step": 2303
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6473728007658126e-07,
+      "loss": 0.2748,
+      "step": 2304
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6042451607997343e-07,
+      "loss": 0.295,
+      "step": 2305
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5614694000654083e-07,
+      "loss": 0.2727,
+      "step": 2306
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.519045594815217e-07,
+      "loss": 0.2648,
+      "step": 2307
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.476973820674178e-07,
+      "loss": 0.2662,
+      "step": 2308
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4352541526397655e-07,
+      "loss": 0.2492,
+      "step": 2309
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.393886665081802e-07,
+      "loss": 0.2555,
+      "step": 2310
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3528714317422763e-07,
+      "loss": 0.2667,
+      "step": 2311
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.31220852573526e-07,
+      "loss": 0.2879,
+      "step": 2312
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2718980195467922e-07,
+      "loss": 0.2408,
+      "step": 2313
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2319399850347035e-07,
+      "loss": 0.2938,
+      "step": 2314
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.192334493428483e-07,
+      "loss": 0.2961,
+      "step": 2315
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.153081615329189e-07,
+      "loss": 0.2667,
+      "step": 2316
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.114181420709338e-07,
+      "loss": 0.2786,
+      "step": 2317
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0756339789127276e-07,
+      "loss": 0.2504,
+      "step": 2318
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0374393586542807e-07,
+      "loss": 0.2527,
+      "step": 2319
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.999597628020089e-07,
+      "loss": 0.2652,
+      "step": 2320
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.962108854467082e-07,
+      "loss": 0.2895,
+      "step": 2321
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.924973104823069e-07,
+      "loss": 0.2895,
+      "step": 2322
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8881904452864975e-07,
+      "loss": 0.3,
+      "step": 2323
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8517609414264724e-07,
+      "loss": 0.2671,
+      "step": 2324
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8156846581824706e-07,
+      "loss": 0.2962,
+      "step": 2325
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.779961659864382e-07,
+      "loss": 0.2669,
+      "step": 2326
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.744592010152313e-07,
+      "loss": 0.2719,
+      "step": 2327
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7095757720964723e-07,
+      "loss": 0.2689,
+      "step": 2328
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6749130081170628e-07,
+      "loss": 0.2348,
+      "step": 2329
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.640603780004235e-07,
+      "loss": 0.2791,
+      "step": 2330
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6066481489178665e-07,
+      "loss": 0.2601,
+      "step": 2331
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.573046175387538e-07,
+      "loss": 0.2507,
+      "step": 2332
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5397979193124024e-07,
+      "loss": 0.2542,
+      "step": 2333
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5069034399610493e-07,
+      "loss": 0.2895,
+      "step": 2334
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4743627959714624e-07,
+      "loss": 0.2534,
+      "step": 2335
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4421760453508182e-07,
+      "loss": 0.2918,
+      "step": 2336
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4103432454754873e-07,
+      "loss": 0.2651,
+      "step": 2337
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.378864453090878e-07,
+      "loss": 0.2566,
+      "step": 2338
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3477397243113256e-07,
+      "loss": 0.2754,
+      "step": 2339
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3169691146200037e-07,
+      "loss": 0.2658,
+      "step": 2340
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.28655267886888e-07,
+      "loss": 0.2915,
+      "step": 2341
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.256490471278493e-07,
+      "loss": 0.2679,
+      "step": 2342
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.226782545437999e-07,
+      "loss": 0.2337,
+      "step": 2343
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.197428954304991e-07,
+      "loss": 0.2837,
+      "step": 2344
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1684297502053688e-07,
+      "loss": 0.2596,
+      "step": 2345
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1397849848333587e-07,
+      "loss": 0.2702,
+      "step": 2346
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1114947092513817e-07,
+      "loss": 0.2361,
+      "step": 2347
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0835589738898755e-07,
+      "loss": 0.2833,
+      "step": 2348
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0559778285473165e-07,
+      "loss": 0.2348,
+      "step": 2349
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0287513223900424e-07,
+      "loss": 0.2448,
+      "step": 2350
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0018795039522744e-07,
+      "loss": 0.2612,
+      "step": 2351
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.753624211358947e-08,
+      "loss": 0.2449,
+      "step": 2352
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.492001212104917e-08,
+      "loss": 0.2496,
+      "step": 2353
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.233926508131818e-08,
+      "loss": 0.2924,
+      "step": 2354
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.979400559485652e-08,
+      "loss": 0.2444,
+      "step": 2355
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.728423819886367e-08,
+      "loss": 0.2457,
+      "step": 2356
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.480996736726977e-08,
+      "loss": 0.2376,
+      "step": 2357
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.237119751073108e-08,
+      "loss": 0.2546,
+      "step": 2358
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.996793297661676e-08,
+      "loss": 0.2832,
+      "step": 2359
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.760017804900655e-08,
+      "loss": 0.281,
+      "step": 2360
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.526793694867751e-08,
+      "loss": 0.263,
+      "step": 2361
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.29712138331018e-08,
+      "loss": 0.2615,
+      "step": 2362
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.071001279643774e-08,
+      "loss": 0.2591,
+      "step": 2363
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.84843378695188e-08,
+      "loss": 0.2622,
+      "step": 2364
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.629419301984908e-08,
+      "loss": 0.2833,
+      "step": 2365
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.413958215160109e-08,
+      "loss": 0.2568,
+      "step": 2366
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.202050910559809e-08,
+      "loss": 0.2955,
+      "step": 2367
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.993697765931839e-08,
+      "loss": 0.2773,
+      "step": 2368
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.78889915268821e-08,
+      "loss": 0.265,
+      "step": 2369
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.587655435904449e-08,
+      "loss": 0.2757,
+      "step": 2370
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.3899669743191496e-08,
+      "loss": 0.2602,
+      "step": 2371
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.19583412033331e-08,
+      "loss": 0.264,
+      "step": 2372
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.005257220009885e-08,
+      "loss": 0.2655,
+      "step": 2373
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.818236613072458e-08,
+      "loss": 0.2707,
+      "step": 2374
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6347726329059037e-08,
+      "loss": 0.2771,
+      "step": 2375
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.454865606554171e-08,
+      "loss": 0.2557,
+      "step": 2376
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.2785158547209436e-08,
+      "loss": 0.271,
+      "step": 2377
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.105723691768981e-08,
+      "loss": 0.2528,
+      "step": 2378
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.9364894257185595e-08,
+      "loss": 0.2432,
+      "step": 2379
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.7708133582483596e-08,
+      "loss": 0.25,
+      "step": 2380
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.608695784693694e-08,
+      "loss": 0.2362,
+      "step": 2381
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.450136994046504e-08,
+      "loss": 0.2677,
+      "step": 2382
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2951372689553615e-08,
+      "loss": 0.2526,
+      "step": 2383
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1436968857236904e-08,
+      "loss": 0.2684,
+      "step": 2384
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9958161143102125e-08,
+      "loss": 0.2868,
+      "step": 2385
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8514952183285038e-08,
+      "loss": 0.2701,
+      "step": 2386
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7107344550461046e-08,
+      "loss": 0.2907,
+      "step": 2387
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5735340753845206e-08,
+      "loss": 0.2723,
+      "step": 2388
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4398943239176687e-08,
+      "loss": 0.2534,
+      "step": 2389
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3098154388734305e-08,
+      "loss": 0.2597,
+      "step": 2390
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1832976521309888e-08,
+      "loss": 0.2811,
+      "step": 2391
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0603411892223814e-08,
+      "loss": 0.2366,
+      "step": 2392
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9409462693302794e-08,
+      "loss": 0.2678,
+      "step": 2393
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8251131052893222e-08,
+      "loss": 0.2739,
+      "step": 2394
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.712841903584561e-08,
+      "loss": 0.2812,
+      "step": 2395
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6041328643516817e-08,
+      "loss": 0.2689,
+      "step": 2396
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4989861813758945e-08,
+      "loss": 0.2685,
+      "step": 2397
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3974020420928214e-08,
+      "loss": 0.2645,
+      "step": 2398
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2993806275871656e-08,
+      "loss": 0.2591,
+      "step": 2399
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2049221125924881e-08,
+      "loss": 0.2551,
+      "step": 2400
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.11402666549143e-08,
+      "loss": 0.2607,
+      "step": 2401
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0266944483148244e-08,
+      "loss": 0.2398,
+      "step": 2402
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.429256167416967e-09,
+      "loss": 0.2797,
+      "step": 2403
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.62720320099264e-09,
+      "loss": 0.2447,
+      "step": 2404
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.860787013620475e-09,
+      "loss": 0.2509,
+      "step": 2405
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.13000897151872e-09,
+      "loss": 0.2879,
+      "step": 2406
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.434870377380886e-09,
+      "loss": 0.296,
+      "step": 2407
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.775372470362417e-09,
+      "loss": 0.2766,
+      "step": 2408
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.151516426091796e-09,
+      "loss": 0.2746,
+      "step": 2409
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5633033566616635e-09,
+      "loss": 0.2876,
+      "step": 2410
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0107343106221555e-09,
+      "loss": 0.2505,
+      "step": 2411
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4938102729875635e-09,
+      "loss": 0.2477,
+      "step": 2412
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.012532165231896e-09,
+      "loss": 0.2443,
+      "step": 2413
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.566900845286657e-09,
+      "loss": 0.2793,
+      "step": 2414
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1569171075341843e-09,
+      "loss": 0.262,
+      "step": 2415
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.782581682816531e-09,
+      "loss": 0.2629,
+      "step": 2416
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4438952384265846e-09,
+      "loss": 0.2579,
+      "step": 2417
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1408583781125082e-09,
+      "loss": 0.2567,
+      "step": 2418
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.734716420644162e-10,
+      "loss": 0.256,
+      "step": 2419
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.417355069321396e-10,
+      "loss": 0.2504,
+      "step": 2420
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.456503858096817e-10,
+      "loss": 0.2606,
+      "step": 2421
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8521662824188004e-10,
+      "loss": 0.2811,
+      "step": 2422
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6043452021330396e-10,
+      "loss": 0.2835,
+      "step": 2423
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.130428416823876e-11,
+      "loss": 0.2897,
+      "step": 2424
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7826078986260543e-11,
+      "loss": 0.2801,
+      "step": 2425
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.2827,
+      "step": 2426
+    },
+    {
+      "epoch": 1.0,
+      "step": 2426,
+      "total_flos": 3.709896226094138e+23,
+      "train_loss": 0.3340999293774489,
+      "train_runtime": 21329.7802,
+      "train_samples_per_second": 14.556,
+      "train_steps_per_second": 0.114
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 2426,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 3.709896226094138e+23,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}