diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,34104 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 5679,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.847953216374269e-08,
+      "loss": 2.369,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1695906432748539e-07,
+      "loss": 2.1433,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7543859649122808e-07,
+      "loss": 2.2322,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3391812865497077e-07,
+      "loss": 2.1588,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9239766081871344e-07,
+      "loss": 2.3003,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5087719298245616e-07,
+      "loss": 2.1944,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.093567251461988e-07,
+      "loss": 2.4212,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6783625730994155e-07,
+      "loss": 2.3647,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.263157894736843e-07,
+      "loss": 2.1969,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.847953216374269e-07,
+      "loss": 2.2765,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.432748538011696e-07,
+      "loss": 2.1124,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.017543859649123e-07,
+      "loss": 2.3799,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.60233918128655e-07,
+      "loss": 2.0603,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.187134502923977e-07,
+      "loss": 2.2584,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.771929824561404e-07,
+      "loss": 2.2411,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.356725146198831e-07,
+      "loss": 2.0055,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.941520467836258e-07,
+      "loss": 2.175,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0526315789473685e-06,
+      "loss": 1.8468,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.111111111111111e-06,
+      "loss": 2.1071,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1695906432748538e-06,
+      "loss": 2.1696,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2280701754385965e-06,
+      "loss": 1.7789,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2865497076023392e-06,
+      "loss": 1.7899,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.345029239766082e-06,
+      "loss": 2.0566,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4035087719298246e-06,
+      "loss": 1.7263,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4619883040935674e-06,
+      "loss": 2.0541,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.52046783625731e-06,
+      "loss": 1.7005,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5789473684210526e-06,
+      "loss": 1.8504,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6374269005847953e-06,
+      "loss": 1.9644,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.695906432748538e-06,
+      "loss": 1.6037,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7543859649122807e-06,
+      "loss": 1.8533,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8128654970760235e-06,
+      "loss": 1.6177,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8713450292397662e-06,
+      "loss": 1.8398,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.929824561403509e-06,
+      "loss": 1.6446,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9883040935672516e-06,
+      "loss": 1.6193,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0467836257309943e-06,
+      "loss": 1.8499,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.105263157894737e-06,
+      "loss": 1.8674,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1637426900584798e-06,
+      "loss": 1.5362,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.222222222222222e-06,
+      "loss": 1.75,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.280701754385965e-06,
+      "loss": 1.5441,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3391812865497075e-06,
+      "loss": 1.6981,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3976608187134502e-06,
+      "loss": 1.5221,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.456140350877193e-06,
+      "loss": 1.7808,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5146198830409357e-06,
+      "loss": 1.1133,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5730994152046784e-06,
+      "loss": 1.7822,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.631578947368421e-06,
+      "loss": 1.4951,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.690058479532164e-06,
+      "loss": 1.7371,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7485380116959066e-06,
+      "loss": 1.4405,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8070175438596493e-06,
+      "loss": 1.6977,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.865497076023392e-06,
+      "loss": 1.7163,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9239766081871347e-06,
+      "loss": 1.497,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9824561403508774e-06,
+      "loss": 1.6342,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.04093567251462e-06,
+      "loss": 1.4855,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0994152046783624e-06,
+      "loss": 1.7146,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.157894736842105e-06,
+      "loss": 1.4486,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.216374269005848e-06,
+      "loss": 1.7034,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2748538011695906e-06,
+      "loss": 1.6943,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 1.4447,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.391812865497076e-06,
+      "loss": 1.6693,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4502923976608188e-06,
+      "loss": 1.5027,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5087719298245615e-06,
+      "loss": 1.6538,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.567251461988304e-06,
+      "loss": 1.4373,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.625730994152047e-06,
+      "loss": 1.5558,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6842105263157896e-06,
+      "loss": 1.6474,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7426900584795324e-06,
+      "loss": 1.4925,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.801169590643275e-06,
+      "loss": 1.4051,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.859649122807018e-06,
+      "loss": 1.6904,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9181286549707605e-06,
+      "loss": 1.4152,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.976608187134503e-06,
+      "loss": 1.6578,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.035087719298246e-06,
+      "loss": 1.6255,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.093567251461989e-06,
+      "loss": 1.4421,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.152046783625731e-06,
+      "loss": 1.6922,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.210526315789474e-06,
+      "loss": 1.517,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.269005847953217e-06,
+      "loss": 1.4023,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3274853801169596e-06,
+      "loss": 1.6783,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.385964912280702e-06,
+      "loss": 1.4312,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.444444444444444e-06,
+      "loss": 1.6766,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.502923976608187e-06,
+      "loss": 1.409,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.56140350877193e-06,
+      "loss": 1.5556,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.619883040935672e-06,
+      "loss": 1.5815,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.678362573099415e-06,
+      "loss": 1.3822,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.736842105263158e-06,
+      "loss": 1.5838,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7953216374269005e-06,
+      "loss": 1.4154,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.853801169590643e-06,
+      "loss": 1.5955,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.912280701754386e-06,
+      "loss": 1.3953,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.970760233918129e-06,
+      "loss": 1.4495,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.029239766081871e-06,
+      "loss": 1.4098,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.087719298245615e-06,
+      "loss": 1.5533,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.146198830409357e-06,
+      "loss": 1.4449,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.2046783625731e-06,
+      "loss": 1.5225,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.263157894736842e-06,
+      "loss": 1.3709,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.321637426900586e-06,
+      "loss": 1.6519,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.380116959064328e-06,
+      "loss": 1.4963,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.438596491228071e-06,
+      "loss": 1.3287,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.497076023391813e-06,
+      "loss": 1.5227,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.555555555555557e-06,
+      "loss": 1.4223,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.6140350877192985e-06,
+      "loss": 1.577,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.672514619883041e-06,
+      "loss": 1.3564,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.730994152046784e-06,
+      "loss": 1.5745,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.789473684210527e-06,
+      "loss": 1.5788,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.847953216374269e-06,
+      "loss": 1.3753,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.906432748538012e-06,
+      "loss": 1.5284,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.964912280701755e-06,
+      "loss": 1.4129,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.023391812865498e-06,
+      "loss": 1.6592,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.08187134502924e-06,
+      "loss": 1.3833,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.140350877192983e-06,
+      "loss": 1.4965,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.198830409356725e-06,
+      "loss": 1.6155,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.2573099415204685e-06,
+      "loss": 1.1249,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.31578947368421e-06,
+      "loss": 1.5901,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.374269005847954e-06,
+      "loss": 1.4462,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.432748538011696e-06,
+      "loss": 1.5716,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.491228070175439e-06,
+      "loss": 1.3723,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.549707602339181e-06,
+      "loss": 1.4986,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.608187134502925e-06,
+      "loss": 1.3544,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.5177,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.72514619883041e-06,
+      "loss": 1.4976,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.783625730994152e-06,
+      "loss": 1.3462,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.842105263157896e-06,
+      "loss": 1.5759,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.9005847953216375e-06,
+      "loss": 1.3463,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.959064327485381e-06,
+      "loss": 1.5752,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.017543859649123e-06,
+      "loss": 1.3113,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.0760233918128665e-06,
+      "loss": 1.4802,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.134502923976608e-06,
+      "loss": 1.5377,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.192982456140352e-06,
+      "loss": 1.3576,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.251461988304094e-06,
+      "loss": 1.4832,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.309941520467837e-06,
+      "loss": 1.3812,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.368421052631579e-06,
+      "loss": 1.499,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.426900584795322e-06,
+      "loss": 1.4956,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.485380116959065e-06,
+      "loss": 1.3741,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.5438596491228074e-06,
+      "loss": 1.3764,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.60233918128655e-06,
+      "loss": 1.55,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.660818713450294e-06,
+      "loss": 1.363,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.719298245614036e-06,
+      "loss": 1.51,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.77777777777778e-06,
+      "loss": 1.349,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.836257309941521e-06,
+      "loss": 1.476,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.894736842105265e-06,
+      "loss": 1.5003,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.953216374269006e-06,
+      "loss": 1.3396,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.01169590643275e-06,
+      "loss": 1.4577,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.070175438596492e-06,
+      "loss": 1.3576,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.128654970760235e-06,
+      "loss": 1.5445,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.187134502923977e-06,
+      "loss": 1.4865,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.24561403508772e-06,
+      "loss": 1.525,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.304093567251463e-06,
+      "loss": 1.5168,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.362573099415205e-06,
+      "loss": 1.4549,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.421052631578948e-06,
+      "loss": 1.3329,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.47953216374269e-06,
+      "loss": 1.5272,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.538011695906434e-06,
+      "loss": 1.3632,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.596491228070176e-06,
+      "loss": 1.5299,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.654970760233919e-06,
+      "loss": 1.3342,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.713450292397661e-06,
+      "loss": 1.4542,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.771929824561405e-06,
+      "loss": 1.0088,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.830409356725146e-06,
+      "loss": 1.5769,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.888888888888888e-06,
+      "loss": 1.3191,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.947368421052632e-06,
+      "loss": 1.5463,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.005847953216374e-06,
+      "loss": 1.3175,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.064327485380117e-06,
+      "loss": 1.5167,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.12280701754386e-06,
+      "loss": 1.4113,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.181286549707603e-06,
+      "loss": 1.2871,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.239766081871345e-06,
+      "loss": 1.4952,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.298245614035088e-06,
+      "loss": 1.3924,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.35672514619883e-06,
+      "loss": 1.4787,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.415204678362574e-06,
+      "loss": 1.3674,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.473684210526315e-06,
+      "loss": 1.4609,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.532163742690059e-06,
+      "loss": 1.3949,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.590643274853801e-06,
+      "loss": 1.3751,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.649122807017545e-06,
+      "loss": 1.4624,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.707602339181286e-06,
+      "loss": 1.3828,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.76608187134503e-06,
+      "loss": 1.4779,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.824561403508772e-06,
+      "loss": 1.3539,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.883040935672515e-06,
+      "loss": 1.3468,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.941520467836257e-06,
+      "loss": 1.4809,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1e-05,
+      "loss": 1.3546,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999999186697928e-06,
+      "loss": 1.4563,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999996746791975e-06,
+      "loss": 1.5188,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999992680282933e-06,
+      "loss": 1.3184,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999986987172128e-06,
+      "loss": 1.4566,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999979667461412e-06,
+      "loss": 1.4401,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999970721153164e-06,
+      "loss": 1.364,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999960148250296e-06,
+      "loss": 1.4407,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999947948756245e-06,
+      "loss": 1.3821,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999934122674986e-06,
+      "loss": 1.3517,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99991867001101e-06,
+      "loss": 1.4693,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999901590769348e-06,
+      "loss": 1.3214,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999882884955554e-06,
+      "loss": 1.4465,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999862552575714e-06,
+      "loss": 1.3367,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999840593636446e-06,
+      "loss": 1.3679,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999817008144888e-06,
+      "loss": 1.5652,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999791796108716e-06,
+      "loss": 1.4259,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99976495753613e-06,
+      "loss": 1.4833,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999736492435867e-06,
+      "loss": 1.3446,
+      "step": 189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99970640081718e-06,
+      "loss": 1.4881,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999674682689862e-06,
+      "loss": 1.323,
+      "step": 191
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999641338064233e-06,
+      "loss": 1.3506,
+      "step": 192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999606366951135e-06,
+      "loss": 1.3299,
+      "step": 193
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999569769361952e-06,
+      "loss": 1.5416,
+      "step": 194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999531545308584e-06,
+      "loss": 1.3818,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999491694803471e-06,
+      "loss": 1.4898,
+      "step": 196
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999450217859572e-06,
+      "loss": 1.3905,
+      "step": 197
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999407114490384e-06,
+      "loss": 1.5496,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99936238470993e-06,
+      "loss": 1.3825,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99931602853276e-06,
+      "loss": 1.3338,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999268045973953e-06,
+      "loss": 1.5145,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999218437049123e-06,
+      "loss": 1.3934,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999167201774404e-06,
+      "loss": 1.4511,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999114340166467e-06,
+      "loss": 1.3156,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999059852242508e-06,
+      "loss": 1.405,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999003738020253e-06,
+      "loss": 1.4294,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998945997517957e-06,
+      "loss": 1.3047,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998886630754406e-06,
+      "loss": 1.528,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99882563774891e-06,
+      "loss": 1.3298,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998763018521314e-06,
+      "loss": 1.4505,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998698773091988e-06,
+      "loss": 1.5091,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998632901481833e-06,
+      "loss": 1.3456,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998565403712278e-06,
+      "loss": 1.3117,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99849627980528e-06,
+      "loss": 1.4102,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99842552978333e-06,
+      "loss": 1.4528,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998353153669443e-06,
+      "loss": 1.3164,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998279151487163e-06,
+      "loss": 1.4928,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998203523260566e-06,
+      "loss": 1.4082,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998126269014255e-06,
+      "loss": 1.46,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99804738877336e-06,
+      "loss": 1.3632,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997966882563549e-06,
+      "loss": 1.3278,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997884750411004e-06,
+      "loss": 1.3953,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997800992342448e-06,
+      "loss": 1.321,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997715608385133e-06,
+      "loss": 1.4743,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99762859856683e-06,
+      "loss": 1.3525,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997539962915849e-06,
+      "loss": 1.5634,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997449701461024e-06,
+      "loss": 1.3546,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997357814231718e-06,
+      "loss": 1.2639,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997264301257825e-06,
+      "loss": 1.5338,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997169162569766e-06,
+      "loss": 1.3439,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997072398198492e-06,
+      "loss": 1.4245,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996974008175482e-06,
+      "loss": 1.2716,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996873992532745e-06,
+      "loss": 1.4861,
+      "step": 233
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99677235130282e-06,
+      "loss": 1.3007,
+      "step": 234
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996669084518767e-06,
+      "loss": 1.4613,
+      "step": 235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996564192214187e-06,
+      "loss": 1.2907,
+      "step": 236
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996457674423202e-06,
+      "loss": 1.4759,
+      "step": 237
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996349531180462e-06,
+      "loss": 1.3361,
+      "step": 238
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996239762521152e-06,
+      "loss": 1.484,
+      "step": 239
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996128368480978e-06,
+      "loss": 1.3019,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996015349096182e-06,
+      "loss": 1.3274,
+      "step": 241
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995900704403531e-06,
+      "loss": 1.5364,
+      "step": 242
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99578443444032e-06,
+      "loss": 1.3337,
+      "step": 243
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995666539244375e-06,
+      "loss": 1.4622,
+      "step": 244
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99554701885405e-06,
+      "loss": 1.3189,
+      "step": 245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995425873308227e-06,
+      "loss": 1.4641,
+      "step": 246
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995303102646316e-06,
+      "loss": 1.4734,
+      "step": 247
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99517870690826e-06,
+      "loss": 1.3216,
+      "step": 248
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995052686134525e-06,
+      "loss": 1.4165,
+      "step": 249
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994925040366109e-06,
+      "loss": 1.4158,
+      "step": 250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994795769644537e-06,
+      "loss": 1.3312,
+      "step": 251
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994664874011864e-06,
+      "loss": 1.527,
+      "step": 252
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994532353510672e-06,
+      "loss": 1.3222,
+      "step": 253
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994398208184074e-06,
+      "loss": 1.4445,
+      "step": 254
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994262438075713e-06,
+      "loss": 1.3766,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.994125043229753e-06,
+      "loss": 1.3196,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993986023690894e-06,
+      "loss": 1.2452,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99384537950436e-06,
+      "loss": 1.5186,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993703110715907e-06,
+      "loss": 1.3551,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99355921737182e-06,
+      "loss": 1.4855,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993413699518906e-06,
+      "loss": 1.3218,
+      "step": 261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993266557204509e-06,
+      "loss": 1.4116,
+      "step": 262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993117790476494e-06,
+      "loss": 1.4344,
+      "step": 263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99296739938326e-06,
+      "loss": 1.3683,
+      "step": 264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992815383973731e-06,
+      "loss": 1.4367,
+      "step": 265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992661744297365e-06,
+      "loss": 1.3409,
+      "step": 266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992506480404137e-06,
+      "loss": 1.4591,
+      "step": 267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992349592344565e-06,
+      "loss": 1.3301,
+      "step": 268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992191080169682e-06,
+      "loss": 1.5458,
+      "step": 269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992030943931059e-06,
+      "loss": 1.2739,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99186918368079e-06,
+      "loss": 1.3192,
+      "step": 271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.9917057994715e-06,
+      "loss": 1.4119,
+      "step": 272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991540791356342e-06,
+      "loss": 1.3334,
+      "step": 273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991374159388993e-06,
+      "loss": 1.4788,
+      "step": 274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991205903623666e-06,
+      "loss": 1.3064,
+      "step": 275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991036024115096e-06,
+      "loss": 1.3342,
+      "step": 276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990864520918547e-06,
+      "loss": 1.3003,
+      "step": 277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990691394089815e-06,
+      "loss": 1.446,
+      "step": 278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990516643685222e-06,
+      "loss": 1.3673,
+      "step": 279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990340269761616e-06,
+      "loss": 1.473,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990162272376376e-06,
+      "loss": 1.5467,
+      "step": 281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989982651587406e-06,
+      "loss": 1.3509,
+      "step": 282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989801407453146e-06,
+      "loss": 1.443,
+      "step": 283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989618540032555e-06,
+      "loss": 1.3448,
+      "step": 284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989434049385122e-06,
+      "loss": 1.452,
+      "step": 285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989247935570868e-06,
+      "loss": 1.264,
+      "step": 286
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989060198650337e-06,
+      "loss": 1.2548,
+      "step": 287
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98887083868461e-06,
+      "loss": 1.3765,
+      "step": 288
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988679855735283e-06,
+      "loss": 1.3236,
+      "step": 289
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98848724986449e-06,
+      "loss": 1.4429,
+      "step": 290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988293021134888e-06,
+      "loss": 1.2887,
+      "step": 291
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988097169609666e-06,
+      "loss": 1.3111,
+      "step": 292
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987899695352535e-06,
+      "loss": 1.403,
+      "step": 293
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987700598427741e-06,
+      "loss": 1.3086,
+      "step": 294
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987499878900052e-06,
+      "loss": 1.438,
+      "step": 295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987297536834771e-06,
+      "loss": 1.3824,
+      "step": 296
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987093572297716e-06,
+      "loss": 1.4682,
+      "step": 297
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986887985355247e-06,
+      "loss": 1.2695,
+      "step": 298
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986680776074245e-06,
+      "loss": 1.3058,
+      "step": 299
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986471944522119e-06,
+      "loss": 1.3279,
+      "step": 300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986261490766801e-06,
+      "loss": 1.4574,
+      "step": 301
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986049414876766e-06,
+      "loss": 1.3014,
+      "step": 302
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985835716921e-06,
+      "loss": 1.3888,
+      "step": 303
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985620396969024e-06,
+      "loss": 1.3058,
+      "step": 304
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985403455090888e-06,
+      "loss": 1.4345,
+      "step": 305
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985184891357165e-06,
+      "loss": 1.2967,
+      "step": 306
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98496470583896e-06,
+      "loss": 1.3117,
+      "step": 307
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984742898607907e-06,
+      "loss": 1.4588,
+      "step": 308
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984519469736157e-06,
+      "loss": 1.2468,
+      "step": 309
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984294419296404e-06,
+      "loss": 1.4058,
+      "step": 310
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984067747361857e-06,
+      "loss": 1.3776,
+      "step": 311
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983839454006257e-06,
+      "loss": 1.1894,
+      "step": 312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983609539303875e-06,
+      "loss": 1.4141,
+      "step": 313
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983378003329507e-06,
+      "loss": 1.319,
+      "step": 314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983144846158472e-06,
+      "loss": 1.4934,
+      "step": 315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982910067866627e-06,
+      "loss": 1.4611,
+      "step": 316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982673668530347e-06,
+      "loss": 1.2973,
+      "step": 317
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982435648226537e-06,
+      "loss": 1.5327,
+      "step": 318
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982196007032631e-06,
+      "loss": 1.295,
+      "step": 319
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98195474502659e-06,
+      "loss": 1.3357,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.9817118622869e-06,
+      "loss": 1.2691,
+      "step": 321
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981467358892578e-06,
+      "loss": 1.3148,
+      "step": 322
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981221234923164e-06,
+      "loss": 1.3049,
+      "step": 323
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980973490458728e-06,
+      "loss": 1.387,
+      "step": 324
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980724125579866e-06,
+      "loss": 1.385,
+      "step": 325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980473140367703e-06,
+      "loss": 1.4798,
+      "step": 326
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980220534903889e-06,
+      "loss": 1.3603,
+      "step": 327
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.9799663092706e-06,
+      "loss": 1.2818,
+      "step": 328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979710463550543e-06,
+      "loss": 1.3954,
+      "step": 329
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97945299782695e-06,
+      "loss": 1.2844,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979193912183577e-06,
+      "loss": 1.4543,
+      "step": 331
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978933206704715e-06,
+      "loss": 1.3176,
+      "step": 332
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978670881475173e-06,
+      "loss": 1.4979,
+      "step": 333
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978406936580292e-06,
+      "loss": 1.2856,
+      "step": 334
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97814137210594e-06,
+      "loss": 1.2986,
+      "step": 335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977874188138508e-06,
+      "loss": 1.482,
+      "step": 336
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977605384764919e-06,
+      "loss": 1.3221,
+      "step": 337
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977334962072618e-06,
+      "loss": 1.5148,
+      "step": 338
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977062920149583e-06,
+      "loss": 1.3563,
+      "step": 339
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97678925908431e-06,
+      "loss": 1.3828,
+      "step": 340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976513978965829e-06,
+      "loss": 1.2791,
+      "step": 341
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976237079883697e-06,
+      "loss": 1.457,
+      "step": 342
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97595856192799e-06,
+      "loss": 1.3243,
+      "step": 343
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97567842518932e-06,
+      "loss": 1.3726,
+      "step": 344
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975396669758816e-06,
+      "loss": 1.3522,
+      "step": 345
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975113295728144e-06,
+      "loss": 1.4159,
+      "step": 346
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97482830318949e-06,
+      "loss": 1.3352,
+      "step": 347
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974541692235568e-06,
+      "loss": 1.1511,
+      "step": 348
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97425346295962e-06,
+      "loss": 1.3986,
+      "step": 349
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973963615455409e-06,
+      "loss": 1.2809,
+      "step": 350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973672149817232e-06,
+      "loss": 1.3885,
+      "step": 351
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973379066139908e-06,
+      "loss": 1.4055,
+      "step": 352
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973084364518781e-06,
+      "loss": 1.3145,
+      "step": 353
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972788045049726e-06,
+      "loss": 1.3827,
+      "step": 354
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972490107829142e-06,
+      "loss": 1.3591,
+      "step": 355
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972190552953954e-06,
+      "loss": 1.4003,
+      "step": 356
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971889380521612e-06,
+      "loss": 1.3721,
+      "step": 357
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971586590630094e-06,
+      "loss": 1.2939,
+      "step": 358
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971282183377905e-06,
+      "loss": 1.3863,
+      "step": 359
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970976158864074e-06,
+      "loss": 1.3124,
+      "step": 360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970668517188158e-06,
+      "loss": 1.3245,
+      "step": 361
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970359258450238e-06,
+      "loss": 1.4136,
+      "step": 362
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970048382750925e-06,
+      "loss": 1.3441,
+      "step": 363
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.96973589019135e-06,
+      "loss": 1.3662,
+      "step": 364
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969421780873175e-06,
+      "loss": 1.4309,
+      "step": 365
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969106054898586e-06,
+      "loss": 1.2941,
+      "step": 366
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968788712370296e-06,
+      "loss": 1.4155,
+      "step": 367
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968469753391544e-06,
+      "loss": 1.2962,
+      "step": 368
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.96814917806609e-06,
+      "loss": 1.4919,
+      "step": 369
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967826986498227e-06,
+      "loss": 1.3344,
+      "step": 370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96750317879277e-06,
+      "loss": 1.2827,
+      "step": 371
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96717775505506e-06,
+      "loss": 1.5068,
+      "step": 372
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966850715390965e-06,
+      "loss": 1.3774,
+      "step": 373
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966522059906878e-06,
+      "loss": 1.4171,
+      "step": 374
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966191788709716e-06,
+      "loss": 1.2807,
+      "step": 375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965859901906923e-06,
+      "loss": 1.4212,
+      "step": 376
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96552639960647e-06,
+      "loss": 1.2983,
+      "step": 377
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965191281916852e-06,
+      "loss": 1.3246,
+      "step": 378
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964854548947092e-06,
+      "loss": 1.4004,
+      "step": 379
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96451620080673e-06,
+      "loss": 1.3163,
+      "step": 380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964176237605845e-06,
+      "loss": 1.4074,
+      "step": 381
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963834659455027e-06,
+      "loss": 1.3054,
+      "step": 382
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963491466465405e-06,
+      "loss": 1.3324,
+      "step": 383
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963146658748623e-06,
+      "loss": 1.2948,
+      "step": 384
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962800236416854e-06,
+      "loss": 1.3612,
+      "step": 385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.9624521995828e-06,
+      "loss": 1.4448,
+      "step": 386
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96210254835968e-06,
+      "loss": 1.3698,
+      "step": 387
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961751282861245e-06,
+      "loss": 1.4932,
+      "step": 388
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96139840320177e-06,
+      "loss": 1.2965,
+      "step": 389
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961043909496051e-06,
+      "loss": 1.3441,
+      "step": 390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960687801859417e-06,
+      "loss": 1.3227,
+      "step": 391
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960330080407712e-06,
+      "loss": 1.2727,
+      "step": 392
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959970745257315e-06,
+      "loss": 1.2971,
+      "step": 393
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95960979652512e-06,
+      "loss": 1.3036,
+      "step": 394
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959247234328556e-06,
+      "loss": 1.3869,
+      "step": 395
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95888305878557e-06,
+      "loss": 1.3038,
+      "step": 396
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958517270014637e-06,
+      "loss": 1.4462,
+      "step": 397
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958149868134752e-06,
+      "loss": 1.3175,
+      "step": 398
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957780853265441e-06,
+      "loss": 1.2503,
+      "step": 399
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957410225526755e-06,
+      "loss": 1.4402,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957037985039262e-06,
+      "loss": 1.2912,
+      "step": 401
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956664131924062e-06,
+      "loss": 1.3909,
+      "step": 402
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95628866630278e-06,
+      "loss": 1.3611,
+      "step": 403
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955911588297558e-06,
+      "loss": 1.2888,
+      "step": 404
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955532898031069e-06,
+      "loss": 1.4032,
+      "step": 405
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955152595626509e-06,
+      "loss": 1.3285,
+      "step": 406
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954770681207597e-06,
+      "loss": 1.332,
+      "step": 407
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954387154898582e-06,
+      "loss": 1.391,
+      "step": 408
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954002016824226e-06,
+      "loss": 1.2683,
+      "step": 409
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95361526710983e-06,
+      "loss": 1.4518,
+      "step": 410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953226905881208e-06,
+      "loss": 1.3093,
+      "step": 411
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952836933264702e-06,
+      "loss": 1.3981,
+      "step": 412
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952445349387177e-06,
+      "loss": 1.1846,
+      "step": 413
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952052154376027e-06,
+      "loss": 1.3036,
+      "step": 414
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951657348359163e-06,
+      "loss": 1.3449,
+      "step": 415
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951260931465025e-06,
+      "loss": 1.344,
+      "step": 416
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950862903822577e-06,
+      "loss": 1.3437,
+      "step": 417
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950463265561303e-06,
+      "loss": 1.3324,
+      "step": 418
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950062016811216e-06,
+      "loss": 1.3463,
+      "step": 419
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949659157702849e-06,
+      "loss": 1.3866,
+      "step": 420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949254688367263e-06,
+      "loss": 1.4379,
+      "step": 421
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948848608936036e-06,
+      "loss": 1.3764,
+      "step": 422
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948440919541277e-06,
+      "loss": 1.4542,
+      "step": 423
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948031620315617e-06,
+      "loss": 1.2797,
+      "step": 424
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947620711392206e-06,
+      "loss": 1.2771,
+      "step": 425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.947208192904722e-06,
+      "loss": 1.3587,
+      "step": 426
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946794064987371e-06,
+      "loss": 1.2806,
+      "step": 427
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94637832777487e-06,
+      "loss": 1.3436,
+      "step": 428
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945960981402471e-06,
+      "loss": 1.3995,
+      "step": 429
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945542026005946e-06,
+      "loss": 1.3143,
+      "step": 430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945121461721588e-06,
+      "loss": 1.4221,
+      "step": 431
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944699288686217e-06,
+      "loss": 1.2687,
+      "step": 432
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944275507037174e-06,
+      "loss": 1.4448,
+      "step": 433
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943850116912322e-06,
+      "loss": 1.3448,
+      "step": 434
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943423118450051e-06,
+      "loss": 1.2532,
+      "step": 435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942994511789275e-06,
+      "loss": 1.4228,
+      "step": 436
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942564297069424e-06,
+      "loss": 1.2794,
+      "step": 437
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942132474430459e-06,
+      "loss": 1.3812,
+      "step": 438
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941699044012861e-06,
+      "loss": 1.2718,
+      "step": 439
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94126400595763e-06,
+      "loss": 1.4217,
+      "step": 440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940827360406297e-06,
+      "loss": 1.2965,
+      "step": 441
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940389107500913e-06,
+      "loss": 1.164,
+      "step": 442
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939949247384046e-06,
+      "loss": 1.428,
+      "step": 443
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939507780198795e-06,
+      "loss": 1.3007,
+      "step": 444
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939064706088777e-06,
+      "loss": 1.3705,
+      "step": 445
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938620025198134e-06,
+      "loss": 1.3057,
+      "step": 446
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938173737671531e-06,
+      "loss": 1.281,
+      "step": 447
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937725843654153e-06,
+      "loss": 1.199,
+      "step": 448
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937276343291708e-06,
+      "loss": 1.432,
+      "step": 449
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936825236730431e-06,
+      "loss": 1.3776,
+      "step": 450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936372524117074e-06,
+      "loss": 1.3829,
+      "step": 451
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935918205598917e-06,
+      "loss": 1.3653,
+      "step": 452
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935462281323756e-06,
+      "loss": 1.4354,
+      "step": 453
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935004751439912e-06,
+      "loss": 1.3049,
+      "step": 454
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934545616096232e-06,
+      "loss": 1.1922,
+      "step": 455
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934084875442082e-06,
+      "loss": 1.4805,
+      "step": 456
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933622529627348e-06,
+      "loss": 1.2798,
+      "step": 457
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933158578802441e-06,
+      "loss": 1.223,
+      "step": 458
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932693023118299e-06,
+      "loss": 1.446,
+      "step": 459
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93222586272637e-06,
+      "loss": 1.3189,
+      "step": 460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931757097778637e-06,
+      "loss": 1.3551,
+      "step": 461
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931286728427592e-06,
+      "loss": 1.2786,
+      "step": 462
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930814754826262e-06,
+      "loss": 1.4596,
+      "step": 463
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930341177128188e-06,
+      "loss": 1.2982,
+      "step": 464
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929865995487434e-06,
+      "loss": 1.3416,
+      "step": 465
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929389210058589e-06,
+      "loss": 1.3839,
+      "step": 466
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928910820996757e-06,
+      "loss": 1.3324,
+      "step": 467
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92843082845757e-06,
+      "loss": 1.2852,
+      "step": 468
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927949232597181e-06,
+      "loss": 1.4254,
+      "step": 469
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927466033572262e-06,
+      "loss": 1.261,
+      "step": 470
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926981231540007e-06,
+      "loss": 1.3022,
+      "step": 471
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926494826658133e-06,
+      "loss": 1.3526,
+      "step": 472
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926006819084878e-06,
+      "loss": 1.2753,
+      "step": 473
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925517208979e-06,
+      "loss": 1.42,
+      "step": 474
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92502599649978e-06,
+      "loss": 1.2716,
+      "step": 475
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92453318180702e-06,
+      "loss": 1.2752,
+      "step": 476
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924038765061042e-06,
+      "loss": 1.2489,
+      "step": 477
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92354274642269e-06,
+      "loss": 1.2717,
+      "step": 478
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92304512605333e-06,
+      "loss": 1.4087,
+      "step": 479
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.922545904114848e-06,
+      "loss": 1.3351,
+      "step": 480
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92204508076965e-06,
+      "loss": 1.3101,
+      "step": 481
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.921542656180667e-06,
+      "loss": 1.3347,
+      "step": 482
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.921038630511345e-06,
+      "loss": 1.3782,
+      "step": 483
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920533003925658e-06,
+      "loss": 1.2909,
+      "step": 484
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920025776588093e-06,
+      "loss": 1.3016,
+      "step": 485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919516948663666e-06,
+      "loss": 1.3902,
+      "step": 486
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919006520317903e-06,
+      "loss": 1.2797,
+      "step": 487
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.918494491716863e-06,
+      "loss": 1.4312,
+      "step": 488
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917980863027114e-06,
+      "loss": 1.2486,
+      "step": 489
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917465634415757e-06,
+      "loss": 1.1813,
+      "step": 490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.916948806050401e-06,
+      "loss": 1.4735,
+      "step": 491
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.916430378099183e-06,
+      "loss": 1.3834,
+      "step": 492
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915910350730758e-06,
+      "loss": 1.1981,
+      "step": 493
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915388724114301e-06,
+      "loss": 1.2939,
+      "step": 494
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91486549841951e-06,
+      "loss": 1.3742,
+      "step": 495
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.9143406738166e-06,
+      "loss": 1.3857,
+      "step": 496
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913814250476307e-06,
+      "loss": 1.4306,
+      "step": 497
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91328622856989e-06,
+      "loss": 1.2771,
+      "step": 498
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91275660826912e-06,
+      "loss": 1.4334,
+      "step": 499
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.9122253897463e-06,
+      "loss": 1.193,
+      "step": 500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.911692573174243e-06,
+      "loss": 1.3037,
+      "step": 501
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.911158158726286e-06,
+      "loss": 1.3636,
+      "step": 502
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910622146576285e-06,
+      "loss": 1.2507,
+      "step": 503
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910084536898615e-06,
+      "loss": 1.409,
+      "step": 504
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.909545329868173e-06,
+      "loss": 1.2777,
+      "step": 505
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.909004525660374e-06,
+      "loss": 1.2981,
+      "step": 506
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908462124451152e-06,
+      "loss": 1.3949,
+      "step": 507
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.907918126416963e-06,
+      "loss": 1.3099,
+      "step": 508
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90737253173478e-06,
+      "loss": 1.4083,
+      "step": 509
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.906825340582095e-06,
+      "loss": 1.2921,
+      "step": 510
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.906276553136924e-06,
+      "loss": 1.3728,
+      "step": 511
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905726169577797e-06,
+      "loss": 1.384,
+      "step": 512
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905174190083763e-06,
+      "loss": 1.3367,
+      "step": 513
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.904620614834397e-06,
+      "loss": 1.2648,
+      "step": 514
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.904065444009785e-06,
+      "loss": 1.418,
+      "step": 515
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.903508677790536e-06,
+      "loss": 1.2607,
+      "step": 516
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90295031635778e-06,
+      "loss": 1.3872,
+      "step": 517
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.902390359893162e-06,
+      "loss": 1.3265,
+      "step": 518
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.901828808578846e-06,
+      "loss": 1.3652,
+      "step": 519
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.901265662597519e-06,
+      "loss": 1.2978,
+      "step": 520
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.900700922132382e-06,
+      "loss": 1.264,
+      "step": 521
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90013458736716e-06,
+      "loss": 1.3994,
+      "step": 522
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899566658486091e-06,
+      "loss": 1.2789,
+      "step": 523
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.898997135673934e-06,
+      "loss": 1.3191,
+      "step": 524
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.898426019115966e-06,
+      "loss": 1.2903,
+      "step": 525
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.897853308997984e-06,
+      "loss": 1.5101,
+      "step": 526
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.897279005506306e-06,
+      "loss": 1.349,
+      "step": 527
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.896703108827758e-06,
+      "loss": 1.3092,
+      "step": 528
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.896125619149697e-06,
+      "loss": 1.3755,
+      "step": 529
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.895546536659988e-06,
+      "loss": 1.4013,
+      "step": 530
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.894965861547023e-06,
+      "loss": 1.2774,
+      "step": 531
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.894383593999706e-06,
+      "loss": 1.2935,
+      "step": 532
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.893799734207458e-06,
+      "loss": 1.4246,
+      "step": 533
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.893214282360226e-06,
+      "loss": 1.2625,
+      "step": 534
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.892627238648465e-06,
+      "loss": 1.1791,
+      "step": 535
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.892038603263154e-06,
+      "loss": 1.3706,
+      "step": 536
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.891448376395788e-06,
+      "loss": 1.2965,
+      "step": 537
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890856558238382e-06,
+      "loss": 1.3744,
+      "step": 538
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890263148983466e-06,
+      "loss": 1.3219,
+      "step": 539
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.889668148824086e-06,
+      "loss": 1.3795,
+      "step": 540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.889071557953812e-06,
+      "loss": 1.2708,
+      "step": 541
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.888473376566723e-06,
+      "loss": 1.3537,
+      "step": 542
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.887873604857424e-06,
+      "loss": 1.3932,
+      "step": 543
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88727224302103e-06,
+      "loss": 1.2861,
+      "step": 544
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.886669291253178e-06,
+      "loss": 1.4272,
+      "step": 545
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.886064749750022e-06,
+      "loss": 1.2659,
+      "step": 546
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88545861870823e-06,
+      "loss": 1.4116,
+      "step": 547
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88485089832499e-06,
+      "loss": 1.221,
+      "step": 548
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.884241588798004e-06,
+      "loss": 1.2664,
+      "step": 549
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.883630690325496e-06,
+      "loss": 1.327,
+      "step": 550
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.883018203106203e-06,
+      "loss": 1.2657,
+      "step": 551
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88240412733938e-06,
+      "loss": 1.2658,
+      "step": 552
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.881788463224798e-06,
+      "loss": 1.3611,
+      "step": 553
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.881171210962746e-06,
+      "loss": 1.2549,
+      "step": 554
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88055237075403e-06,
+      "loss": 1.3989,
+      "step": 555
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.879931942799968e-06,
+      "loss": 1.4195,
+      "step": 556
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.8793099273024e-06,
+      "loss": 1.3467,
+      "step": 557
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.878686324463683e-06,
+      "loss": 1.386,
+      "step": 558
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.878061134486685e-06,
+      "loss": 1.2394,
+      "step": 559
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.877434357574794e-06,
+      "loss": 1.3793,
+      "step": 560
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.876805993931915e-06,
+      "loss": 1.3992,
+      "step": 561
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.876176043762467e-06,
+      "loss": 1.3014,
+      "step": 562
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.875544507271384e-06,
+      "loss": 1.3719,
+      "step": 563
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87491138466412e-06,
+      "loss": 1.2854,
+      "step": 564
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.874276676146643e-06,
+      "loss": 1.1915,
+      "step": 565
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.873640381925437e-06,
+      "loss": 1.3546,
+      "step": 566
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.873002502207502e-06,
+      "loss": 1.2663,
+      "step": 567
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.872363037200351e-06,
+      "loss": 1.4638,
+      "step": 568
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.871721987112019e-06,
+      "loss": 1.2932,
+      "step": 569
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87107935215105e-06,
+      "loss": 1.4265,
+      "step": 570
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.870435132526508e-06,
+      "loss": 1.2724,
+      "step": 571
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.869789328447969e-06,
+      "loss": 1.2875,
+      "step": 572
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86914194012553e-06,
+      "loss": 1.3453,
+      "step": 573
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.868492967769796e-06,
+      "loss": 1.4195,
+      "step": 574
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.867842411591895e-06,
+      "loss": 1.2446,
+      "step": 575
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.867190271803466e-06,
+      "loss": 1.3703,
+      "step": 576
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86653654861666e-06,
+      "loss": 1.2869,
+      "step": 577
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.865881242244152e-06,
+      "loss": 1.1544,
+      "step": 578
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86522435289912e-06,
+      "loss": 1.3967,
+      "step": 579
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.864565880795269e-06,
+      "loss": 1.2766,
+      "step": 580
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.863905826146813e-06,
+      "loss": 1.3341,
+      "step": 581
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.863244189168482e-06,
+      "loss": 1.2711,
+      "step": 582
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86258097007552e-06,
+      "loss": 1.3224,
+      "step": 583
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.861916169083684e-06,
+      "loss": 1.3986,
+      "step": 584
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.861249786409248e-06,
+      "loss": 1.2596,
+      "step": 585
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.860581822269002e-06,
+      "loss": 1.3254,
+      "step": 586
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.859912276880249e-06,
+      "loss": 1.257,
+      "step": 587
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.859241150460803e-06,
+      "loss": 1.4058,
+      "step": 588
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.858568443228997e-06,
+      "loss": 1.3172,
+      "step": 589
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.857894155403677e-06,
+      "loss": 1.3861,
+      "step": 590
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.857218287204204e-06,
+      "loss": 1.1259,
+      "step": 591
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.856540838850449e-06,
+      "loss": 1.2412,
+      "step": 592
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.855861810562804e-06,
+      "loss": 1.3274,
+      "step": 593
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.855181202562168e-06,
+      "loss": 1.284,
+      "step": 594
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.854499015069955e-06,
+      "loss": 1.2493,
+      "step": 595
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.853815248308101e-06,
+      "loss": 1.3996,
+      "step": 596
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.853129902499046e-06,
+      "loss": 1.3644,
+      "step": 597
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.852442977865747e-06,
+      "loss": 1.2586,
+      "step": 598
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.851754474631673e-06,
+      "loss": 1.3836,
+      "step": 599
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.851064393020812e-06,
+      "loss": 1.143,
+      "step": 600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.850372733257662e-06,
+      "loss": 1.3261,
+      "step": 601
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.849679495567233e-06,
+      "loss": 1.352,
+      "step": 602
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.848984680175049e-06,
+      "loss": 1.3592,
+      "step": 603
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.84828828730715e-06,
+      "loss": 1.4217,
+      "step": 604
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.847590317190083e-06,
+      "loss": 1.2586,
+      "step": 605
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.846890770050915e-06,
+      "loss": 1.4317,
+      "step": 606
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.846189646117224e-06,
+      "loss": 1.1531,
+      "step": 607
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.8454869456171e-06,
+      "loss": 1.2964,
+      "step": 608
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.844782668779145e-06,
+      "loss": 1.3269,
+      "step": 609
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.844076815832476e-06,
+      "loss": 1.279,
+      "step": 610
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.843369387006721e-06,
+      "loss": 1.3934,
+      "step": 611
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.84266038253202e-06,
+      "loss": 1.2783,
+      "step": 612
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.841949802639031e-06,
+      "loss": 1.1772,
+      "step": 613
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.841237647558915e-06,
+      "loss": 1.3557,
+      "step": 614
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.840523917523354e-06,
+      "loss": 1.2819,
+      "step": 615
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.839808612764541e-06,
+      "loss": 1.2905,
+      "step": 616
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.839091733515176e-06,
+      "loss": 1.4001,
+      "step": 617
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.838373280008477e-06,
+      "loss": 1.3062,
+      "step": 618
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83765325247817e-06,
+      "loss": 1.2892,
+      "step": 619
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.836931651158496e-06,
+      "loss": 1.2493,
+      "step": 620
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.836208476284208e-06,
+      "loss": 1.2678,
+      "step": 621
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83548372809057e-06,
+      "loss": 1.3727,
+      "step": 622
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.834757406813353e-06,
+      "loss": 1.2479,
+      "step": 623
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.834029512688852e-06,
+      "loss": 1.4663,
+      "step": 624
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83330004595386e-06,
+      "loss": 1.309,
+      "step": 625
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83256900684569e-06,
+      "loss": 1.3668,
+      "step": 626
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.831836395602164e-06,
+      "loss": 1.13,
+      "step": 627
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.831102212461613e-06,
+      "loss": 1.2732,
+      "step": 628
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83036645766289e-06,
+      "loss": 1.3612,
+      "step": 629
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.829629131445342e-06,
+      "loss": 1.2576,
+      "step": 630
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.828890234048842e-06,
+      "loss": 1.4077,
+      "step": 631
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.828149765713768e-06,
+      "loss": 1.3594,
+      "step": 632
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.827407726681008e-06,
+      "loss": 1.2946,
+      "step": 633
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.826664117191967e-06,
+      "loss": 1.307,
+      "step": 634
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82591893748855e-06,
+      "loss": 1.3418,
+      "step": 635
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.825172187813185e-06,
+      "loss": 1.1851,
+      "step": 636
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.824423868408803e-06,
+      "loss": 1.2684,
+      "step": 637
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82367397951885e-06,
+      "loss": 1.3567,
+      "step": 638
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.822922521387277e-06,
+      "loss": 1.2332,
+      "step": 639
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82216949425855e-06,
+      "loss": 1.3737,
+      "step": 640
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.821414898377646e-06,
+      "loss": 1.2532,
+      "step": 641
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82065873399005e-06,
+      "loss": 1.257,
+      "step": 642
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.819901001341758e-06,
+      "loss": 1.3835,
+      "step": 643
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.819141700679275e-06,
+      "loss": 1.2565,
+      "step": 644
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.81838083224962e-06,
+      "loss": 1.4305,
+      "step": 645
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.817618396300317e-06,
+      "loss": 1.2615,
+      "step": 646
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.816854393079402e-06,
+      "loss": 1.3265,
+      "step": 647
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.816088822835423e-06,
+      "loss": 1.3015,
+      "step": 648
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.815321685817436e-06,
+      "loss": 1.2976,
+      "step": 649
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.814552982275004e-06,
+      "loss": 1.4154,
+      "step": 650
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.813782712458206e-06,
+      "loss": 1.2686,
+      "step": 651
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.813010876617626e-06,
+      "loss": 1.3212,
+      "step": 652
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.812237475004354e-06,
+      "loss": 1.2974,
+      "step": 653
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.811462507869998e-06,
+      "loss": 1.337,
+      "step": 654
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.810685975466672e-06,
+      "loss": 1.2271,
+      "step": 655
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.809907878046994e-06,
+      "loss": 1.225,
+      "step": 656
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.809128215864096e-06,
+      "loss": 1.4449,
+      "step": 657
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.808346989171621e-06,
+      "loss": 1.2681,
+      "step": 658
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.807564198223716e-06,
+      "loss": 1.2548,
+      "step": 659
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.806779843275041e-06,
+      "loss": 1.3039,
+      "step": 660
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.805993924580763e-06,
+      "loss": 1.3129,
+      "step": 661
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.805206442396555e-06,
+      "loss": 1.1391,
+      "step": 662
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.804417396978605e-06,
+      "loss": 1.3841,
+      "step": 663
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.803626788583603e-06,
+      "loss": 1.2593,
+      "step": 664
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.802834617468753e-06,
+      "loss": 1.3446,
+      "step": 665
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.802040883891762e-06,
+      "loss": 1.3307,
+      "step": 666
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.801245588110849e-06,
+      "loss": 1.4132,
+      "step": 667
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.80044873038474e-06,
+      "loss": 1.3367,
+      "step": 668
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79965031097267e-06,
+      "loss": 1.2573,
+      "step": 669
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.798850330134385e-06,
+      "loss": 1.3148,
+      "step": 670
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.798048788130128e-06,
+      "loss": 1.264,
+      "step": 671
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.797245685220662e-06,
+      "loss": 1.2195,
+      "step": 672
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.796441021667254e-06,
+      "loss": 1.31,
+      "step": 673
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.795634797731675e-06,
+      "loss": 1.2789,
+      "step": 674
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.794827013676206e-06,
+      "loss": 1.3521,
+      "step": 675
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79401766976364e-06,
+      "loss": 1.2724,
+      "step": 676
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.793206766257271e-06,
+      "loss": 1.3827,
+      "step": 677
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.7923943034209e-06,
+      "loss": 1.2088,
+      "step": 678
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.791580281518844e-06,
+      "loss": 1.2764,
+      "step": 679
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.790764700815917e-06,
+      "loss": 1.3242,
+      "step": 680
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.789947561577445e-06,
+      "loss": 1.3591,
+      "step": 681
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.789128864069262e-06,
+      "loss": 1.309,
+      "step": 682
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.788308608557707e-06,
+      "loss": 1.3308,
+      "step": 683
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.787486795309621e-06,
+      "loss": 1.2872,
+      "step": 684
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.786663424592366e-06,
+      "loss": 1.1908,
+      "step": 685
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.785838496673796e-06,
+      "loss": 1.2666,
+      "step": 686
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.78501201182228e-06,
+      "loss": 1.2478,
+      "step": 687
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.784183970306687e-06,
+      "loss": 1.3339,
+      "step": 688
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.783354372396398e-06,
+      "loss": 1.2663,
+      "step": 689
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.7825232183613e-06,
+      "loss": 1.3856,
+      "step": 690
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.781690508471782e-06,
+      "loss": 1.1934,
+      "step": 691
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.780856242998744e-06,
+      "loss": 1.2593,
+      "step": 692
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.78002042221359e-06,
+      "loss": 1.3508,
+      "step": 693
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.779183046388228e-06,
+      "loss": 1.2107,
+      "step": 694
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.778344115795075e-06,
+      "loss": 1.3576,
+      "step": 695
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.777503630707054e-06,
+      "loss": 1.323,
+      "step": 696
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77666159139759e-06,
+      "loss": 1.3297,
+      "step": 697
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.775817998140615e-06,
+      "loss": 1.3027,
+      "step": 698
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.774972851210572e-06,
+      "loss": 1.2468,
+      "step": 699
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.774126150882402e-06,
+      "loss": 1.2922,
+      "step": 700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.773277897431552e-06,
+      "loss": 1.2693,
+      "step": 701
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77242809113398e-06,
+      "loss": 1.3807,
+      "step": 702
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.771576732266147e-06,
+      "loss": 1.309,
+      "step": 703
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.770723821105012e-06,
+      "loss": 1.3285,
+      "step": 704
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.76986935792805e-06,
+      "loss": 1.2401,
+      "step": 705
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.769013343013234e-06,
+      "loss": 1.3453,
+      "step": 706
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.768155776639044e-06,
+      "loss": 1.1788,
+      "step": 707
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.76729665908446e-06,
+      "loss": 1.3086,
+      "step": 708
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.766435990628977e-06,
+      "loss": 1.3563,
+      "step": 709
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.765573771552586e-06,
+      "loss": 1.239,
+      "step": 710
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.764710002135784e-06,
+      "loss": 1.3748,
+      "step": 711
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.763844682659573e-06,
+      "loss": 1.3371,
+      "step": 712
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.762977813405461e-06,
+      "loss": 1.3705,
+      "step": 713
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.762109394655456e-06,
+      "loss": 1.2464,
+      "step": 714
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.761239426692077e-06,
+      "loss": 1.28,
+      "step": 715
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.76036790979834e-06,
+      "loss": 1.3444,
+      "step": 716
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.759494844257766e-06,
+      "loss": 1.2409,
+      "step": 717
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.758620230354386e-06,
+      "loss": 1.3034,
+      "step": 718
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.757744068372725e-06,
+      "loss": 1.2952,
+      "step": 719
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.756866358597819e-06,
+      "loss": 1.3166,
+      "step": 720
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.755987101315205e-06,
+      "loss": 1.2513,
+      "step": 721
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.755106296810925e-06,
+      "loss": 1.3146,
+      "step": 722
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.754223945371524e-06,
+      "loss": 1.2684,
+      "step": 723
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.753340047284045e-06,
+      "loss": 1.337,
+      "step": 724
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.752454602836042e-06,
+      "loss": 1.2471,
+      "step": 725
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.751567612315567e-06,
+      "loss": 1.3122,
+      "step": 726
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.750679076011175e-06,
+      "loss": 1.309,
+      "step": 727
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.74978899421193e-06,
+      "loss": 1.2846,
+      "step": 728
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.748897367207391e-06,
+      "loss": 1.3527,
+      "step": 729
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.748004195287622e-06,
+      "loss": 1.2539,
+      "step": 730
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.747109478743191e-06,
+      "loss": 1.3553,
+      "step": 731
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.74621321786517e-06,
+      "loss": 1.2481,
+      "step": 732
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.74531541294513e-06,
+      "loss": 1.2465,
+      "step": 733
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.744416064275145e-06,
+      "loss": 1.225,
+      "step": 734
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.743515172147793e-06,
+      "loss": 1.2519,
+      "step": 735
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.742612736856151e-06,
+      "loss": 1.285,
+      "step": 736
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.741708758693805e-06,
+      "loss": 1.4152,
+      "step": 737
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.74080323795483e-06,
+      "loss": 1.2949,
+      "step": 738
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.739896174933815e-06,
+      "loss": 1.3214,
+      "step": 739
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.738987569925851e-06,
+      "loss": 1.3046,
+      "step": 740
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.738077423226519e-06,
+      "loss": 1.2818,
+      "step": 741
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.73716573513191e-06,
+      "loss": 1.2322,
+      "step": 742
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.73625250593862e-06,
+      "loss": 1.2468,
+      "step": 743
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.735337735943735e-06,
+      "loss": 1.2446,
+      "step": 744
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.734421425444853e-06,
+      "loss": 1.3246,
+      "step": 745
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.733503574740067e-06,
+      "loss": 1.2876,
+      "step": 746
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.732584184127973e-06,
+      "loss": 1.3962,
+      "step": 747
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.73166325390767e-06,
+      "loss": 1.257,
+      "step": 748
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.730740784378755e-06,
+      "loss": 1.3583,
+      "step": 749
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.729816775841324e-06,
+      "loss": 1.3599,
+      "step": 750
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.728891228595976e-06,
+      "loss": 1.2575,
+      "step": 751
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.727964142943815e-06,
+      "loss": 1.3567,
+      "step": 752
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.72703551918644e-06,
+      "loss": 1.2798,
+      "step": 753
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.72610535762595e-06,
+      "loss": 1.3445,
+      "step": 754
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.725173658564948e-06,
+      "loss": 1.2326,
+      "step": 755
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.724240422306531e-06,
+      "loss": 1.173,
+      "step": 756
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.723305649154305e-06,
+      "loss": 1.3457,
+      "step": 757
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.722369339412369e-06,
+      "loss": 1.3223,
+      "step": 758
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.721431493385322e-06,
+      "loss": 1.3811,
+      "step": 759
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.72049211137827e-06,
+      "loss": 1.2733,
+      "step": 760
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.71955119369681e-06,
+      "loss": 1.2911,
+      "step": 761
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.718608740647041e-06,
+      "loss": 1.2241,
+      "step": 762
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.717664752535566e-06,
+      "loss": 1.0814,
+      "step": 763
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.716719229669482e-06,
+      "loss": 1.3076,
+      "step": 764
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.715772172356388e-06,
+      "loss": 1.3141,
+      "step": 765
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.71482358090438e-06,
+      "loss": 1.2353,
+      "step": 766
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.713873455622058e-06,
+      "loss": 1.4087,
+      "step": 767
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.712921796818512e-06,
+      "loss": 1.2946,
+      "step": 768
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.71196860480334e-06,
+      "loss": 1.1718,
+      "step": 769
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.711013879886638e-06,
+      "loss": 1.2801,
+      "step": 770
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.710057622378992e-06,
+      "loss": 1.3064,
+      "step": 771
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.709099832591495e-06,
+      "loss": 1.3854,
+      "step": 772
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.708140510835737e-06,
+      "loss": 1.3628,
+      "step": 773
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.707179657423806e-06,
+      "loss": 1.3291,
+      "step": 774
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.706217272668283e-06,
+      "loss": 1.3273,
+      "step": 775
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.705253356882258e-06,
+      "loss": 1.2516,
+      "step": 776
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.704287910379308e-06,
+      "loss": 1.3639,
+      "step": 777
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.703320933473515e-06,
+      "loss": 1.3092,
+      "step": 778
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.702352426479458e-06,
+      "loss": 1.1072,
+      "step": 779
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.701382389712208e-06,
+      "loss": 1.3446,
+      "step": 780
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.700410823487342e-06,
+      "loss": 1.2429,
+      "step": 781
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.699437728120933e-06,
+      "loss": 1.391,
+      "step": 782
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.698463103929542e-06,
+      "loss": 1.3037,
+      "step": 783
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.697486951230241e-06,
+      "loss": 1.3697,
+      "step": 784
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.69650927034059e-06,
+      "loss": 1.008,
+      "step": 785
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.695530061578649e-06,
+      "loss": 1.3623,
+      "step": 786
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.694549325262973e-06,
+      "loss": 1.2844,
+      "step": 787
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.693567061712623e-06,
+      "loss": 1.357,
+      "step": 788
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.69258327124714e-06,
+      "loss": 1.332,
+      "step": 789
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.69159795418658e-06,
+      "loss": 1.3154,
+      "step": 790
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.690611110851485e-06,
+      "loss": 1.2734,
+      "step": 791
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.689622741562891e-06,
+      "loss": 1.2997,
+      "step": 792
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.688632846642341e-06,
+      "loss": 1.3451,
+      "step": 793
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.687641426411865e-06,
+      "loss": 1.31,
+      "step": 794
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.686648481193994e-06,
+      "loss": 1.3273,
+      "step": 795
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.685654011311751e-06,
+      "loss": 1.267,
+      "step": 796
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.684658017088663e-06,
+      "loss": 1.3223,
+      "step": 797
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.683660498848745e-06,
+      "loss": 1.1346,
+      "step": 798
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.682661456916509e-06,
+      "loss": 1.2711,
+      "step": 799
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.681660891616967e-06,
+      "loss": 1.3567,
+      "step": 800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.68065880327562e-06,
+      "loss": 1.2522,
+      "step": 801
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.679655192218473e-06,
+      "loss": 1.377,
+      "step": 802
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.678650058772017e-06,
+      "loss": 1.2499,
+      "step": 803
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.677643403263246e-06,
+      "loss": 1.3367,
+      "step": 804
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.676635226019645e-06,
+      "loss": 1.3315,
+      "step": 805
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.675625527369192e-06,
+      "loss": 1.2461,
+      "step": 806
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.674614307640368e-06,
+      "loss": 1.4218,
+      "step": 807
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.673601567162143e-06,
+      "loss": 1.1654,
+      "step": 808
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.67258730626398e-06,
+      "loss": 1.3561,
+      "step": 809
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.671571525275842e-06,
+      "loss": 1.2184,
+      "step": 810
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.67055422452818e-06,
+      "loss": 1.349,
+      "step": 811
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.669535404351947e-06,
+      "loss": 1.3222,
+      "step": 812
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.668515065078583e-06,
+      "loss": 1.3521,
+      "step": 813
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.667493207040029e-06,
+      "loss": 1.1243,
+      "step": 814
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.666469830568714e-06,
+      "loss": 1.2678,
+      "step": 815
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.665444935997566e-06,
+      "loss": 1.2808,
+      "step": 816
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.664418523660004e-06,
+      "loss": 1.2288,
+      "step": 817
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.663390593889938e-06,
+      "loss": 1.3898,
+      "step": 818
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.66236114702178e-06,
+      "loss": 1.2709,
+      "step": 819
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.661330183390426e-06,
+      "loss": 1.3919,
+      "step": 820
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.660297703331273e-06,
+      "loss": 1.3716,
+      "step": 821
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.65926370718021e-06,
+      "loss": 1.2391,
+      "step": 822
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.658228195273612e-06,
+      "loss": 1.3426,
+      "step": 823
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.657191167948356e-06,
+      "loss": 1.2669,
+      "step": 824
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.656152625541807e-06,
+      "loss": 1.3166,
+      "step": 825
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.655112568391826e-06,
+      "loss": 1.2738,
+      "step": 826
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.654070996836766e-06,
+      "loss": 1.188,
+      "step": 827
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.653027911215469e-06,
+      "loss": 1.2648,
+      "step": 828
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.651983311867274e-06,
+      "loss": 1.2671,
+      "step": 829
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.650937199132014e-06,
+      "loss": 1.3178,
+      "step": 830
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.649889573350006e-06,
+      "loss": 1.3137,
+      "step": 831
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.648840434862066e-06,
+      "loss": 1.2981,
+      "step": 832
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.647789784009502e-06,
+      "loss": 1.3134,
+      "step": 833
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.646737621134112e-06,
+      "loss": 1.2655,
+      "step": 834
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.645683946578189e-06,
+      "loss": 1.2918,
+      "step": 835
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.644628760684509e-06,
+      "loss": 1.3203,
+      "step": 836
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.643572063796352e-06,
+      "loss": 1.2441,
+      "step": 837
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.64251385625748e-06,
+      "loss": 1.3292,
+      "step": 838
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.641454138412153e-06,
+      "loss": 1.2515,
+      "step": 839
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.640392910605116e-06,
+      "loss": 1.3195,
+      "step": 840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.639330173181612e-06,
+      "loss": 1.253,
+      "step": 841
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.638265926487367e-06,
+      "loss": 1.3723,
+      "step": 842
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.637200170868607e-06,
+      "loss": 1.1267,
+      "step": 843
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.636132906672042e-06,
+      "loss": 1.3012,
+      "step": 844
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.635064134244876e-06,
+      "loss": 1.2998,
+      "step": 845
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.633993853934803e-06,
+      "loss": 1.3546,
+      "step": 846
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.632922066090007e-06,
+      "loss": 1.2163,
+      "step": 847
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.631848771059165e-06,
+      "loss": 1.3169,
+      "step": 848
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.63077396919144e-06,
+      "loss": 1.229,
+      "step": 849
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.629697660836486e-06,
+      "loss": 1.2671,
+      "step": 850
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.628619846344453e-06,
+      "loss": 1.2551,
+      "step": 851
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.627540526065973e-06,
+      "loss": 1.3348,
+      "step": 852
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.626459700352172e-06,
+      "loss": 1.313,
+      "step": 853
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.625377369554666e-06,
+      "loss": 1.2888,
+      "step": 854
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.62429353402556e-06,
+      "loss": 1.2402,
+      "step": 855
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.623208194117446e-06,
+      "loss": 1.1743,
+      "step": 856
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.622121350183412e-06,
+      "loss": 1.3167,
+      "step": 857
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.621033002577025e-06,
+      "loss": 1.2747,
+      "step": 858
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.61994315165235e-06,
+      "loss": 1.3084,
+      "step": 859
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.618851797763942e-06,
+      "loss": 1.2608,
+      "step": 860
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.617758941266835e-06,
+      "loss": 1.3267,
+      "step": 861
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.616664582516562e-06,
+      "loss": 1.2837,
+      "step": 862
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.615568721869139e-06,
+      "loss": 1.1045,
+      "step": 863
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.614471359681072e-06,
+      "loss": 1.3613,
+      "step": 864
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.613372496309357e-06,
+      "loss": 1.2897,
+      "step": 865
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.612272132111476e-06,
+      "loss": 1.2788,
+      "step": 866
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.611170267445401e-06,
+      "loss": 1.3265,
+      "step": 867
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.610066902669593e-06,
+      "loss": 1.2912,
+      "step": 868
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.608962038142996e-06,
+      "loss": 1.2435,
+      "step": 869
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.60785567422505e-06,
+      "loss": 1.2265,
+      "step": 870
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.606747811275674e-06,
+      "loss": 1.2388,
+      "step": 871
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.60563844965528e-06,
+      "loss": 1.2965,
+      "step": 872
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.60452758972477e-06,
+      "loss": 1.2803,
+      "step": 873
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.603415231845523e-06,
+      "loss": 1.3094,
+      "step": 874
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.60230137637942e-06,
+      "loss": 1.2281,
+      "step": 875
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.601186023688815e-06,
+      "loss": 1.1718,
+      "step": 876
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.600069174136559e-06,
+      "loss": 1.4221,
+      "step": 877
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.598950828085985e-06,
+      "loss": 1.2561,
+      "step": 878
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.597830985900913e-06,
+      "loss": 1.0991,
+      "step": 879
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.596709647945655e-06,
+      "loss": 1.3249,
+      "step": 880
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.595586814585002e-06,
+      "loss": 1.3266,
+      "step": 881
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.594462486184237e-06,
+      "loss": 1.296,
+      "step": 882
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.593336663109128e-06,
+      "loss": 1.2609,
+      "step": 883
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.592209345725924e-06,
+      "loss": 1.3127,
+      "step": 884
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.591080534401371e-06,
+      "loss": 1.2625,
+      "step": 885
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.58995022950269e-06,
+      "loss": 1.1817,
+      "step": 886
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.588818431397598e-06,
+      "loss": 1.3046,
+      "step": 887
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.587685140454287e-06,
+      "loss": 1.297,
+      "step": 888
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.586550357041444e-06,
+      "loss": 1.3808,
+      "step": 889
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.585414081528235e-06,
+      "loss": 1.2922,
+      "step": 890
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.584276314284316e-06,
+      "loss": 1.2468,
+      "step": 891
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.583137055679824e-06,
+      "loss": 1.1624,
+      "step": 892
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.581996306085386e-06,
+      "loss": 1.3437,
+      "step": 893
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.580854065872112e-06,
+      "loss": 1.2552,
+      "step": 894
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.579710335411594e-06,
+      "loss": 1.2634,
+      "step": 895
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.578565115075913e-06,
+      "loss": 1.2566,
+      "step": 896
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.577418405237633e-06,
+      "loss": 1.3892,
+      "step": 897
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.576270206269802e-06,
+      "loss": 1.3251,
+      "step": 898
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.575120518545954e-06,
+      "loss": 1.151,
+      "step": 899
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.573969342440107e-06,
+      "loss": 1.3301,
+      "step": 900
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.572816678326759e-06,
+      "loss": 1.2535,
+      "step": 901
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.571662526580898e-06,
+      "loss": 1.322,
+      "step": 902
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.570506887577994e-06,
+      "loss": 1.2482,
+      "step": 903
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.569349761694e-06,
+      "loss": 1.2983,
+      "step": 904
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.568191149305352e-06,
+      "loss": 1.1707,
+      "step": 905
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.567031050788975e-06,
+      "loss": 1.3086,
+      "step": 906
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.565869466522265e-06,
+      "loss": 1.3774,
+      "step": 907
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.564706396883118e-06,
+      "loss": 1.2511,
+      "step": 908
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.563541842249903e-06,
+      "loss": 1.2873,
+      "step": 909
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.562375803001471e-06,
+      "loss": 1.2715,
+      "step": 910
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.561208279517159e-06,
+      "loss": 1.3054,
+      "step": 911
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.560039272176787e-06,
+      "loss": 1.3298,
+      "step": 912
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.55886878136066e-06,
+      "loss": 1.317,
+      "step": 913
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.557696807449562e-06,
+      "loss": 1.2727,
+      "step": 914
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.556523350824759e-06,
+      "loss": 1.3145,
+      "step": 915
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.555348411868001e-06,
+      "loss": 1.3666,
+      "step": 916
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.55417199096152e-06,
+      "loss": 1.2656,
+      "step": 917
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.552994088488033e-06,
+      "loss": 1.2883,
+      "step": 918
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.551814704830734e-06,
+      "loss": 1.2849,
+      "step": 919
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.550633840373301e-06,
+      "loss": 1.2736,
+      "step": 920
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.549451495499895e-06,
+      "loss": 1.2435,
+      "step": 921
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.548267670595155e-06,
+      "loss": 1.2951,
+      "step": 922
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.547082366044207e-06,
+      "loss": 1.2585,
+      "step": 923
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.545895582232653e-06,
+      "loss": 1.2583,
+      "step": 924
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.544707319546577e-06,
+      "loss": 1.329,
+      "step": 925
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.543517578372552e-06,
+      "loss": 1.2455,
+      "step": 926
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.542326359097619e-06,
+      "loss": 1.2855,
+      "step": 927
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.54113366210931e-06,
+      "loss": 1.2697,
+      "step": 928
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.539939487795632e-06,
+      "loss": 1.3319,
+      "step": 929
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.538743836545077e-06,
+      "loss": 1.3415,
+      "step": 930
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.537546708746611e-06,
+      "loss": 1.2723,
+      "step": 931
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.536348104789692e-06,
+      "loss": 1.2623,
+      "step": 932
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.535148025064243e-06,
+      "loss": 1.2583,
+      "step": 933
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.53394646996068e-06,
+      "loss": 1.1581,
+      "step": 934
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.532743439869892e-06,
+      "loss": 1.2985,
+      "step": 935
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.531538935183252e-06,
+      "loss": 1.2809,
+      "step": 936
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.530332956292605e-06,
+      "loss": 1.2078,
+      "step": 937
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.529125503590286e-06,
+      "loss": 1.2334,
+      "step": 938
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.527916577469104e-06,
+      "loss": 1.2708,
+      "step": 939
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.526706178322348e-06,
+      "loss": 1.2605,
+      "step": 940
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.525494306543782e-06,
+      "loss": 1.1041,
+      "step": 941
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.524280962527658e-06,
+      "loss": 1.2657,
+      "step": 942
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.5230661466687e-06,
+      "loss": 1.3127,
+      "step": 943
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.521849859362111e-06,
+      "loss": 1.3308,
+      "step": 944
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.520632101003579e-06,
+      "loss": 1.3324,
+      "step": 945
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.519412871989264e-06,
+      "loss": 1.2279,
+      "step": 946
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.518192172715806e-06,
+      "loss": 1.3032,
+      "step": 947
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.516970003580325e-06,
+      "loss": 1.3602,
+      "step": 948
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.515746364980415e-06,
+      "loss": 1.2106,
+      "step": 949
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.514521257314158e-06,
+      "loss": 1.1964,
+      "step": 950
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.5132946809801e-06,
+      "loss": 1.3396,
+      "step": 951
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.512066636377273e-06,
+      "loss": 1.2596,
+      "step": 952
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.51083712390519e-06,
+      "loss": 1.2921,
+      "step": 953
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.509606143963832e-06,
+      "loss": 1.2856,
+      "step": 954
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.508373696953664e-06,
+      "loss": 1.2374,
+      "step": 955
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.507139783275628e-06,
+      "loss": 1.2993,
+      "step": 956
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.50590440333114e-06,
+      "loss": 1.0969,
+      "step": 957
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.504667557522097e-06,
+      "loss": 1.3093,
+      "step": 958
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.503429246250867e-06,
+      "loss": 1.3427,
+      "step": 959
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.502189469920305e-06,
+      "loss": 1.2317,
+      "step": 960
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.50094822893373e-06,
+      "loss": 1.3757,
+      "step": 961
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.499705523694944e-06,
+      "loss": 1.268,
+      "step": 962
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.498461354608228e-06,
+      "loss": 1.1695,
+      "step": 963
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.497215722078334e-06,
+      "loss": 1.3724,
+      "step": 964
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.495968626510492e-06,
+      "loss": 1.251,
+      "step": 965
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.494720068310409e-06,
+      "loss": 1.3403,
+      "step": 966
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.493470047884268e-06,
+      "loss": 1.2918,
+      "step": 967
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.492218565638724e-06,
+      "loss": 1.2716,
+      "step": 968
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.490965621980912e-06,
+      "loss": 1.2772,
+      "step": 969
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.48971121731844e-06,
+      "loss": 1.1577,
+      "step": 970
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.488455352059394e-06,
+      "loss": 1.3515,
+      "step": 971
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.487198026612333e-06,
+      "loss": 1.3037,
+      "step": 972
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.485939241386286e-06,
+      "loss": 1.2452,
+      "step": 973
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.484678996790768e-06,
+      "loss": 1.3233,
+      "step": 974
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.483417293235759e-06,
+      "loss": 1.2991,
+      "step": 975
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.48215413113172e-06,
+      "loss": 1.2302,
+      "step": 976
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.480889510889583e-06,
+      "loss": 1.2032,
+      "step": 977
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.479623432920755e-06,
+      "loss": 1.3105,
+      "step": 978
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.478355897637119e-06,
+      "loss": 1.3807,
+      "step": 979
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.477086905451027e-06,
+      "loss": 1.2453,
+      "step": 980
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.475816456775313e-06,
+      "loss": 1.3065,
+      "step": 981
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.474544552023277e-06,
+      "loss": 1.3297,
+      "step": 982
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.473271191608699e-06,
+      "loss": 1.2421,
+      "step": 983
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.471996375945827e-06,
+      "loss": 1.3076,
+      "step": 984
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.470720105449387e-06,
+      "loss": 1.2581,
+      "step": 985
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.469442380534576e-06,
+      "loss": 1.0456,
+      "step": 986
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.468163201617063e-06,
+      "loss": 1.2703,
+      "step": 987
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.466882569112992e-06,
+      "loss": 1.2371,
+      "step": 988
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.465600483438982e-06,
+      "loss": 1.3368,
+      "step": 989
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.46431694501212e-06,
+      "loss": 1.3356,
+      "step": 990
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.463031954249967e-06,
+      "loss": 1.3454,
+      "step": 991
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.461745511570559e-06,
+      "loss": 1.272,
+      "step": 992
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.460457617392403e-06,
+      "loss": 1.226,
+      "step": 993
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.459168272134476e-06,
+      "loss": 1.2898,
+      "step": 994
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.457877476216228e-06,
+      "loss": 1.2703,
+      "step": 995
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.456585230057587e-06,
+      "loss": 1.2606,
+      "step": 996
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.455291534078942e-06,
+      "loss": 1.2583,
+      "step": 997
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.453996388701162e-06,
+      "loss": 1.2506,
+      "step": 998
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.452699794345583e-06,
+      "loss": 1.1785,
+      "step": 999
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.451401751434015e-06,
+      "loss": 1.3129,
+      "step": 1000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.45010226038874e-06,
+      "loss": 1.2543,
+      "step": 1001
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.448801321632506e-06,
+      "loss": 1.2627,
+      "step": 1002
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.44749893558854e-06,
+      "loss": 1.2322,
+      "step": 1003
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.446195102680531e-06,
+      "loss": 1.2832,
+      "step": 1004
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.444889823332645e-06,
+      "loss": 1.3236,
+      "step": 1005
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.443583097969517e-06,
+      "loss": 1.1056,
+      "step": 1006
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.442274927016252e-06,
+      "loss": 1.3433,
+      "step": 1007
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.440965310898425e-06,
+      "loss": 1.2532,
+      "step": 1008
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.439654250042081e-06,
+      "loss": 1.3241,
+      "step": 1009
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.438341744873735e-06,
+      "loss": 1.2835,
+      "step": 1010
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.437027795820373e-06,
+      "loss": 1.3067,
+      "step": 1011
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.435712403309449e-06,
+      "loss": 1.2074,
+      "step": 1012
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.434395567768889e-06,
+      "loss": 1.2895,
+      "step": 1013
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.433077289627087e-06,
+      "loss": 1.3311,
+      "step": 1014
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.431757569312903e-06,
+      "loss": 1.2778,
+      "step": 1015
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.430436407255674e-06,
+      "loss": 1.3618,
+      "step": 1016
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.429113803885199e-06,
+      "loss": 1.3581,
+      "step": 1017
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.42778975963175e-06,
+      "loss": 1.2379,
+      "step": 1018
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.426464274926066e-06,
+      "loss": 1.3021,
+      "step": 1019
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.425137350199353e-06,
+      "loss": 1.311,
+      "step": 1020
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.423808985883289e-06,
+      "loss": 1.3119,
+      "step": 1021
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.422479182410018e-06,
+      "loss": 1.0397,
+      "step": 1022
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.421147940212152e-06,
+      "loss": 1.2899,
+      "step": 1023
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.419815259722774e-06,
+      "loss": 1.2239,
+      "step": 1024
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.418481141375431e-06,
+      "loss": 1.2664,
+      "step": 1025
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.417145585604139e-06,
+      "loss": 1.2887,
+      "step": 1026
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.415808592843383e-06,
+      "loss": 1.3927,
+      "step": 1027
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.414470163528114e-06,
+      "loss": 1.1449,
+      "step": 1028
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.413130298093752e-06,
+      "loss": 1.2257,
+      "step": 1029
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.411788996976182e-06,
+      "loss": 1.2706,
+      "step": 1030
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.410446260611758e-06,
+      "loss": 1.2772,
+      "step": 1031
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.4091020894373e-06,
+      "loss": 1.3175,
+      "step": 1032
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.407756483890097e-06,
+      "loss": 1.2399,
+      "step": 1033
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.406409444407896e-06,
+      "loss": 1.3611,
+      "step": 1034
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.405060971428924e-06,
+      "loss": 1.154,
+      "step": 1035
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.403711065391863e-06,
+      "loss": 1.2665,
+      "step": 1036
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.402359726735867e-06,
+      "loss": 1.302,
+      "step": 1037
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.401006955900555e-06,
+      "loss": 1.301,
+      "step": 1038
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.399652753326014e-06,
+      "loss": 1.3711,
+      "step": 1039
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.398297119452789e-06,
+      "loss": 1.2766,
+      "step": 1040
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.396940054721899e-06,
+      "loss": 1.1835,
+      "step": 1041
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.395581559574825e-06,
+      "loss": 1.2569,
+      "step": 1042
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.394221634453513e-06,
+      "loss": 1.2264,
+      "step": 1043
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.392860279800377e-06,
+      "loss": 1.2306,
+      "step": 1044
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.391497496058291e-06,
+      "loss": 1.3067,
+      "step": 1045
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.3901332836706e-06,
+      "loss": 1.2441,
+      "step": 1046
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.388767643081109e-06,
+      "loss": 1.3017,
+      "step": 1047
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.387400574734089e-06,
+      "loss": 1.1135,
+      "step": 1048
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.386032079074277e-06,
+      "loss": 1.317,
+      "step": 1049
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.384662156546874e-06,
+      "loss": 1.2705,
+      "step": 1050
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.383290807597541e-06,
+      "loss": 1.3026,
+      "step": 1051
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.381918032672409e-06,
+      "loss": 1.2922,
+      "step": 1052
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.380543832218069e-06,
+      "loss": 1.3294,
+      "step": 1053
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.379168206681578e-06,
+      "loss": 1.1973,
+      "step": 1054
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.377791156510456e-06,
+      "loss": 1.2685,
+      "step": 1055
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.376412682152685e-06,
+      "loss": 1.259,
+      "step": 1056
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.37503278405671e-06,
+      "loss": 1.1984,
+      "step": 1057
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.373651462671445e-06,
+      "loss": 1.2106,
+      "step": 1058
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.372268718446259e-06,
+      "loss": 1.2289,
+      "step": 1059
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.370884551830988e-06,
+      "loss": 1.3345,
+      "step": 1060
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.36949896327593e-06,
+      "loss": 1.2668,
+      "step": 1061
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.368111953231849e-06,
+      "loss": 1.2564,
+      "step": 1062
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.366723522149965e-06,
+      "loss": 1.2821,
+      "step": 1063
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.365333670481964e-06,
+      "loss": 1.1487,
+      "step": 1064
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.363942398679992e-06,
+      "loss": 1.2458,
+      "step": 1065
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.362549707196664e-06,
+      "loss": 1.3685,
+      "step": 1066
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.361155596485046e-06,
+      "loss": 1.2822,
+      "step": 1067
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.359760066998675e-06,
+      "loss": 1.3491,
+      "step": 1068
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.358363119191544e-06,
+      "loss": 1.2757,
+      "step": 1069
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.35696475351811e-06,
+      "loss": 1.2273,
+      "step": 1070
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.355564970433288e-06,
+      "loss": 1.2842,
+      "step": 1071
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.35416377039246e-06,
+      "loss": 1.2779,
+      "step": 1072
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.352761153851466e-06,
+      "loss": 1.3119,
+      "step": 1073
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.351357121266603e-06,
+      "loss": 1.2771,
+      "step": 1074
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.349951673094633e-06,
+      "loss": 1.3065,
+      "step": 1075
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.348544809792779e-06,
+      "loss": 1.2651,
+      "step": 1076
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.347136531818722e-06,
+      "loss": 1.209,
+      "step": 1077
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.345726839630605e-06,
+      "loss": 1.2824,
+      "step": 1078
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.344315733687029e-06,
+      "loss": 1.2755,
+      "step": 1079
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.342903214447056e-06,
+      "loss": 1.2761,
+      "step": 1080
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.341489282370209e-06,
+      "loss": 1.234,
+      "step": 1081
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.340073937916469e-06,
+      "loss": 1.2885,
+      "step": 1082
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.338657181546277e-06,
+      "loss": 1.3075,
+      "step": 1083
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.337239013720534e-06,
+      "loss": 1.2027,
+      "step": 1084
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.3358194349006e-06,
+      "loss": 1.2044,
+      "step": 1085
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.334398445548291e-06,
+      "loss": 1.2725,
+      "step": 1086
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.332976046125887e-06,
+      "loss": 1.3283,
+      "step": 1087
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.331552237096124e-06,
+      "loss": 1.2771,
+      "step": 1088
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.330127018922195e-06,
+      "loss": 1.2888,
+      "step": 1089
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.328700392067754e-06,
+      "loss": 1.273,
+      "step": 1090
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.327272356996913e-06,
+      "loss": 1.312,
+      "step": 1091
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.325842914174238e-06,
+      "loss": 1.2381,
+      "step": 1092
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.324412064064764e-06,
+      "loss": 1.1659,
+      "step": 1093
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.32297980713397e-06,
+      "loss": 1.2739,
+      "step": 1094
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.321546143847802e-06,
+      "loss": 1.2017,
+      "step": 1095
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.32011107467266e-06,
+      "loss": 1.2883,
+      "step": 1096
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.318674600075399e-06,
+      "loss": 1.2384,
+      "step": 1097
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.317236720523338e-06,
+      "loss": 1.3066,
+      "step": 1098
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.315797436484248e-06,
+      "loss": 1.3286,
+      "step": 1099
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.314356748426357e-06,
+      "loss": 1.1064,
+      "step": 1100
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.312914656818353e-06,
+      "loss": 1.2501,
+      "step": 1101
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.311471162129375e-06,
+      "loss": 1.2571,
+      "step": 1102
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.310026264829027e-06,
+      "loss": 1.2584,
+      "step": 1103
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.308579965387357e-06,
+      "loss": 1.2497,
+      "step": 1104
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.307132264274883e-06,
+      "loss": 1.228,
+      "step": 1105
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.305683161962569e-06,
+      "loss": 1.1009,
+      "step": 1106
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.30423265892184e-06,
+      "loss": 1.2733,
+      "step": 1107
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.302780755624572e-06,
+      "loss": 1.2145,
+      "step": 1108
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.301327452543103e-06,
+      "loss": 1.2807,
+      "step": 1109
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.29987275015022e-06,
+      "loss": 1.2201,
+      "step": 1110
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.298416648919169e-06,
+      "loss": 1.2632,
+      "step": 1111
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.29695914932365e-06,
+      "loss": 1.2476,
+      "step": 1112
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.295500251837818e-06,
+      "loss": 1.267,
+      "step": 1113
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.294039956936283e-06,
+      "loss": 1.3972,
+      "step": 1114
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.292578265094109e-06,
+      "loss": 1.2445,
+      "step": 1115
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.291115176786814e-06,
+      "loss": 1.2275,
+      "step": 1116
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.289650692490373e-06,
+      "loss": 1.2755,
+      "step": 1117
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.28818481268121e-06,
+      "loss": 1.2284,
+      "step": 1118
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.286717537836211e-06,
+      "loss": 1.1051,
+      "step": 1119
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.285248868432708e-06,
+      "loss": 1.2432,
+      "step": 1120
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.283778804948492e-06,
+      "loss": 1.2632,
+      "step": 1121
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.282307347861802e-06,
+      "loss": 1.3327,
+      "step": 1122
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.280834497651334e-06,
+      "loss": 1.2254,
+      "step": 1123
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.279360254796239e-06,
+      "loss": 1.2908,
+      "step": 1124
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.277884619776116e-06,
+      "loss": 1.2773,
+      "step": 1125
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.276407593071023e-06,
+      "loss": 1.2091,
+      "step": 1126
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.274929175161468e-06,
+      "loss": 1.2829,
+      "step": 1127
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.27344936652841e-06,
+      "loss": 1.3081,
+      "step": 1128
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.271968167653256e-06,
+      "loss": 1.156,
+      "step": 1129
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.27048557901788e-06,
+      "loss": 1.3366,
+      "step": 1130
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.269001601104593e-06,
+      "loss": 1.2167,
+      "step": 1131
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.267516234396167e-06,
+      "loss": 1.3286,
+      "step": 1132
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.266029479375822e-06,
+      "loss": 1.2495,
+      "step": 1133
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.264541336527228e-06,
+      "loss": 1.3132,
+      "step": 1134
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.26305180633451e-06,
+      "loss": 1.089,
+      "step": 1135
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.261560889282248e-06,
+      "loss": 1.2483,
+      "step": 1136
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.260068585855461e-06,
+      "loss": 1.2561,
+      "step": 1137
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.25857489653963e-06,
+      "loss": 1.2429,
+      "step": 1138
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.257079821820684e-06,
+      "loss": 1.2235,
+      "step": 1139
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.255583362184998e-06,
+      "loss": 1.2325,
+      "step": 1140
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.254085518119407e-06,
+      "loss": 1.284,
+      "step": 1141
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.252586290111188e-06,
+      "loss": 1.2429,
+      "step": 1142
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.251085678648072e-06,
+      "loss": 1.2444,
+      "step": 1143
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.249583684218238e-06,
+      "loss": 1.2941,
+      "step": 1144
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.248080307310315e-06,
+      "loss": 1.3084,
+      "step": 1145
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.246575548413386e-06,
+      "loss": 1.2785,
+      "step": 1146
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.245069408016978e-06,
+      "loss": 1.2532,
+      "step": 1147
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.24356188661107e-06,
+      "loss": 1.0588,
+      "step": 1148
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.242052984686092e-06,
+      "loss": 1.2332,
+      "step": 1149
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.24054270273292e-06,
+      "loss": 1.3133,
+      "step": 1150
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.239031041242878e-06,
+      "loss": 1.2454,
+      "step": 1151
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.237518000707745e-06,
+      "loss": 1.2994,
+      "step": 1152
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.236003581619743e-06,
+      "loss": 1.2476,
+      "step": 1153
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.234487784471545e-06,
+      "loss": 1.3648,
+      "step": 1154
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.232970609756267e-06,
+      "loss": 1.1061,
+      "step": 1155
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.231452057967482e-06,
+      "loss": 1.2883,
+      "step": 1156
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.229932129599206e-06,
+      "loss": 1.3851,
+      "step": 1157
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.2284108251459e-06,
+      "loss": 1.088,
+      "step": 1158
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.226888145102484e-06,
+      "loss": 1.3235,
+      "step": 1159
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.225364089964307e-06,
+      "loss": 1.3008,
+      "step": 1160
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.223838660227183e-06,
+      "loss": 1.1944,
+      "step": 1161
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.222311856387366e-06,
+      "loss": 1.3327,
+      "step": 1162
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.220783678941551e-06,
+      "loss": 1.209,
+      "step": 1163
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.219254128386894e-06,
+      "loss": 1.0292,
+      "step": 1164
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.217723205220983e-06,
+      "loss": 1.3055,
+      "step": 1165
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.216190909941862e-06,
+      "loss": 1.2397,
+      "step": 1166
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.214657243048021e-06,
+      "loss": 1.2659,
+      "step": 1167
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.213122205038389e-06,
+      "loss": 1.3212,
+      "step": 1168
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.21158579641235e-06,
+      "loss": 1.2761,
+      "step": 1169
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.210048017669727e-06,
+      "loss": 1.3519,
+      "step": 1170
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.208508869310793e-06,
+      "loss": 1.0632,
+      "step": 1171
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.206968351836262e-06,
+      "loss": 1.2657,
+      "step": 1172
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.205426465747303e-06,
+      "loss": 1.3134,
+      "step": 1173
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.203883211545517e-06,
+      "loss": 1.2928,
+      "step": 1174
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.20233858973296e-06,
+      "loss": 1.2624,
+      "step": 1175
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.200792600812128e-06,
+      "loss": 1.2467,
+      "step": 1176
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.199245245285964e-06,
+      "loss": 1.0355,
+      "step": 1177
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.197696523657856e-06,
+      "loss": 1.2709,
+      "step": 1178
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.196146436431635e-06,
+      "loss": 1.2534,
+      "step": 1179
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.194594984111574e-06,
+      "loss": 1.2279,
+      "step": 1180
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.193042167202398e-06,
+      "loss": 1.2622,
+      "step": 1181
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.191487986209266e-06,
+      "loss": 1.2343,
+      "step": 1182
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.189932441637787e-06,
+      "loss": 1.247,
+      "step": 1183
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.188375533994014e-06,
+      "loss": 1.0795,
+      "step": 1184
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.186817263784438e-06,
+      "loss": 1.2928,
+      "step": 1185
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.185257631516e-06,
+      "loss": 1.2622,
+      "step": 1186
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.183696637696078e-06,
+      "loss": 1.2896,
+      "step": 1187
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.182134282832498e-06,
+      "loss": 1.2362,
+      "step": 1188
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.180570567433525e-06,
+      "loss": 1.2586,
+      "step": 1189
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.179005492007871e-06,
+      "loss": 1.2593,
+      "step": 1190
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.177439057064684e-06,
+      "loss": 1.3244,
+      "step": 1191
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.175871263113558e-06,
+      "loss": 1.3582,
+      "step": 1192
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.174302110664533e-06,
+      "loss": 1.0597,
+      "step": 1193
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.172731600228081e-06,
+      "loss": 1.2809,
+      "step": 1194
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.171159732315129e-06,
+      "loss": 1.2244,
+      "step": 1195
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.169586507437033e-06,
+      "loss": 1.2547,
+      "step": 1196
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.168011926105598e-06,
+      "loss": 1.2653,
+      "step": 1197
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.166435988833066e-06,
+      "loss": 1.2569,
+      "step": 1198
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.164858696132126e-06,
+      "loss": 1.2806,
+      "step": 1199
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.163280048515902e-06,
+      "loss": 1.1052,
+      "step": 1200
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.161700046497959e-06,
+      "loss": 1.2704,
+      "step": 1201
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.160118690592308e-06,
+      "loss": 1.2923,
+      "step": 1202
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.158535981313395e-06,
+      "loss": 1.2597,
+      "step": 1203
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.156951919176107e-06,
+      "loss": 1.2648,
+      "step": 1204
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.155366504695777e-06,
+      "loss": 1.3026,
+      "step": 1205
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.15377973838817e-06,
+      "loss": 1.3211,
+      "step": 1206
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.152191620769492e-06,
+      "loss": 1.2217,
+      "step": 1207
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.150602152356394e-06,
+      "loss": 1.3054,
+      "step": 1208
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.149011333665965e-06,
+      "loss": 1.237,
+      "step": 1209
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.147419165215728e-06,
+      "loss": 1.2458,
+      "step": 1210
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.145825647523649e-06,
+      "loss": 1.2854,
+      "step": 1211
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.144230781108132e-06,
+      "loss": 1.227,
+      "step": 1212
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.142634566488023e-06,
+      "loss": 1.1099,
+      "step": 1213
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.141037004182602e-06,
+      "loss": 1.2353,
+      "step": 1214
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.13943809471159e-06,
+      "loss": 1.2515,
+      "step": 1215
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.137837838595144e-06,
+      "loss": 1.2773,
+      "step": 1216
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.136236236353862e-06,
+      "loss": 1.2293,
+      "step": 1217
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.134633288508779e-06,
+      "loss": 1.2892,
+      "step": 1218
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.133028995581366e-06,
+      "loss": 1.2289,
+      "step": 1219
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.131423358093533e-06,
+      "loss": 1.0853,
+      "step": 1220
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.129816376567628e-06,
+      "loss": 1.2495,
+      "step": 1221
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.128208051526436e-06,
+      "loss": 1.3128,
+      "step": 1222
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.126598383493179e-06,
+      "loss": 1.2909,
+      "step": 1223
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.124987372991512e-06,
+      "loss": 1.2463,
+      "step": 1224
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.123375020545534e-06,
+      "loss": 1.2769,
+      "step": 1225
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.121761326679777e-06,
+      "loss": 1.1333,
+      "step": 1226
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.120146291919206e-06,
+      "loss": 1.3806,
+      "step": 1227
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.118529916789228e-06,
+      "loss": 1.3112,
+      "step": 1228
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.116912201815681e-06,
+      "loss": 1.1592,
+      "step": 1229
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.115293147524847e-06,
+      "loss": 1.2153,
+      "step": 1230
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.113672754443432e-06,
+      "loss": 1.1997,
+      "step": 1231
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.112051023098586e-06,
+      "loss": 1.3085,
+      "step": 1232
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.110427954017891e-06,
+      "loss": 1.2319,
+      "step": 1233
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.108803547729368e-06,
+      "loss": 1.3502,
+      "step": 1234
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.107177804761468e-06,
+      "loss": 1.2672,
+      "step": 1235
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.105550725643078e-06,
+      "loss": 1.2391,
+      "step": 1236
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.103922310903524e-06,
+      "loss": 1.2968,
+      "step": 1237
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.10229256107256e-06,
+      "loss": 1.2617,
+      "step": 1238
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.100661476680379e-06,
+      "loss": 1.247,
+      "step": 1239
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.099029058257606e-06,
+      "loss": 1.247,
+      "step": 1240
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.097395306335301e-06,
+      "loss": 1.2653,
+      "step": 1241
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.09576022144496e-06,
+      "loss": 1.0589,
+      "step": 1242
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.094123804118507e-06,
+      "loss": 1.2108,
+      "step": 1243
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.092486054888304e-06,
+      "loss": 1.2767,
+      "step": 1244
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.090846974287142e-06,
+      "loss": 1.2477,
+      "step": 1245
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.089206562848254e-06,
+      "loss": 1.3482,
+      "step": 1246
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.087564821105295e-06,
+      "loss": 1.2832,
+      "step": 1247
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.085921749592359e-06,
+      "loss": 1.2834,
+      "step": 1248
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.084277348843973e-06,
+      "loss": 1.0367,
+      "step": 1249
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.082631619395092e-06,
+      "loss": 1.2616,
+      "step": 1250
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.08098456178111e-06,
+      "loss": 1.3003,
+      "step": 1251
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.079336176537846e-06,
+      "loss": 1.2597,
+      "step": 1252
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.077686464201554e-06,
+      "loss": 1.2384,
+      "step": 1253
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.07603542530892e-06,
+      "loss": 1.3076,
+      "step": 1254
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.074383060397063e-06,
+      "loss": 1.1436,
+      "step": 1255
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.072729370003532e-06,
+      "loss": 1.2161,
+      "step": 1256
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.071074354666304e-06,
+      "loss": 1.194,
+      "step": 1257
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.069418014923793e-06,
+      "loss": 1.254,
+      "step": 1258
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.067760351314838e-06,
+      "loss": 1.2522,
+      "step": 1259
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.066101364378714e-06,
+      "loss": 1.181,
+      "step": 1260
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.064441054655122e-06,
+      "loss": 1.3178,
+      "step": 1261
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.062779422684198e-06,
+      "loss": 1.3093,
+      "step": 1262
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.061116469006504e-06,
+      "loss": 1.2323,
+      "step": 1263
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.05945219416303e-06,
+      "loss": 1.233,
+      "step": 1264
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.057786598695206e-06,
+      "loss": 1.1066,
+      "step": 1265
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.056119683144879e-06,
+      "loss": 1.2238,
+      "step": 1266
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.054451448054336e-06,
+      "loss": 1.2669,
+      "step": 1267
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.052781893966284e-06,
+      "loss": 1.2929,
+      "step": 1268
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.051111021423868e-06,
+      "loss": 1.3279,
+      "step": 1269
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.049438830970655e-06,
+      "loss": 1.2192,
+      "step": 1270
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.047765323150644e-06,
+      "loss": 1.1332,
+      "step": 1271
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.046090498508264e-06,
+      "loss": 1.2198,
+      "step": 1272
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.044414357588367e-06,
+      "loss": 1.2815,
+      "step": 1273
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.042736900936236e-06,
+      "loss": 1.2425,
+      "step": 1274
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.041058129097586e-06,
+      "loss": 1.2491,
+      "step": 1275
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.039378042618556e-06,
+      "loss": 1.2259,
+      "step": 1276
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.037696642045713e-06,
+      "loss": 1.2699,
+      "step": 1277
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.036013927926049e-06,
+      "loss": 1.1672,
+      "step": 1278
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.034329900806987e-06,
+      "loss": 1.2587,
+      "step": 1279
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.03264456123638e-06,
+      "loss": 1.3115,
+      "step": 1280
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.030957909762497e-06,
+      "loss": 1.2682,
+      "step": 1281
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.029269946934048e-06,
+      "loss": 1.2855,
+      "step": 1282
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.027580673300157e-06,
+      "loss": 1.3266,
+      "step": 1283
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.025890089410383e-06,
+      "loss": 1.0936,
+      "step": 1284
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.024198195814707e-06,
+      "loss": 1.2883,
+      "step": 1285
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.022504993063538e-06,
+      "loss": 1.2325,
+      "step": 1286
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.020810481707709e-06,
+      "loss": 1.2923,
+      "step": 1287
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.01911466229848e-06,
+      "loss": 1.2343,
+      "step": 1288
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.017417535387538e-06,
+      "loss": 1.2622,
+      "step": 1289
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.015719101526993e-06,
+      "loss": 1.2195,
+      "step": 1290
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.01401936126938e-06,
+      "loss": 1.0625,
+      "step": 1291
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.01231831516766e-06,
+      "loss": 1.2605,
+      "step": 1292
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.01061596377522e-06,
+      "loss": 1.2738,
+      "step": 1293
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.008912307645869e-06,
+      "loss": 1.2804,
+      "step": 1294
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.007207347333842e-06,
+      "loss": 1.3186,
+      "step": 1295
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.005501083393799e-06,
+      "loss": 1.2073,
+      "step": 1296
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.003793516380822e-06,
+      "loss": 1.2276,
+      "step": 1297
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.002084646850418e-06,
+      "loss": 1.3306,
+      "step": 1298
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.00037447535852e-06,
+      "loss": 1.2436,
+      "step": 1299
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.99866300246148e-06,
+      "loss": 1.0663,
+      "step": 1300
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.996950228716077e-06,
+      "loss": 1.3479,
+      "step": 1301
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.995236154679512e-06,
+      "loss": 1.2533,
+      "step": 1302
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.99352078090941e-06,
+      "loss": 1.3136,
+      "step": 1303
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.991804107963816e-06,
+      "loss": 1.2268,
+      "step": 1304
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.990086136401199e-06,
+      "loss": 1.2785,
+      "step": 1305
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.988366866780452e-06,
+      "loss": 1.2589,
+      "step": 1306
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.986646299660889e-06,
+      "loss": 1.1068,
+      "step": 1307
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.984924435602246e-06,
+      "loss": 1.2898,
+      "step": 1308
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.983201275164684e-06,
+      "loss": 1.2346,
+      "step": 1309
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.981476818908778e-06,
+      "loss": 1.2571,
+      "step": 1310
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.979751067395534e-06,
+      "loss": 1.2496,
+      "step": 1311
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.978024021186373e-06,
+      "loss": 1.3014,
+      "step": 1312
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.97629568084314e-06,
+      "loss": 1.241,
+      "step": 1313
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.9745660469281e-06,
+      "loss": 1.236,
+      "step": 1314
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.972835120003936e-06,
+      "loss": 1.2395,
+      "step": 1315
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.971102900633756e-06,
+      "loss": 1.2736,
+      "step": 1316
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.969369389381091e-06,
+      "loss": 1.2115,
+      "step": 1317
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.967634586809884e-06,
+      "loss": 1.2334,
+      "step": 1318
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.965898493484503e-06,
+      "loss": 1.2575,
+      "step": 1319
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.964161109969738e-06,
+      "loss": 1.0861,
+      "step": 1320
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.962422436830794e-06,
+      "loss": 1.2601,
+      "step": 1321
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.960682474633295e-06,
+      "loss": 1.2129,
+      "step": 1322
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.958941223943292e-06,
+      "loss": 1.2886,
+      "step": 1323
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.957198685327247e-06,
+      "loss": 1.2494,
+      "step": 1324
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.955454859352044e-06,
+      "loss": 1.274,
+      "step": 1325
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.953709746584987e-06,
+      "loss": 1.2401,
+      "step": 1326
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.951963347593797e-06,
+      "loss": 1.0575,
+      "step": 1327
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.950215662946615e-06,
+      "loss": 1.3333,
+      "step": 1328
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.948466693211999e-06,
+      "loss": 1.2806,
+      "step": 1329
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.946716438958922e-06,
+      "loss": 1.3434,
+      "step": 1330
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.944964900756784e-06,
+      "loss": 1.2453,
+      "step": 1331
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.943212079175392e-06,
+      "loss": 1.2679,
+      "step": 1332
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.941457974784977e-06,
+      "loss": 1.2812,
+      "step": 1333
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.939702588156188e-06,
+      "loss": 1.2285,
+      "step": 1334
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.937945919860086e-06,
+      "loss": 1.2853,
+      "step": 1335
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.93618797046815e-06,
+      "loss": 0.8914,
+      "step": 1336
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.934428740552282e-06,
+      "loss": 1.3055,
+      "step": 1337
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.932668230684794e-06,
+      "loss": 1.2125,
+      "step": 1338
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.930906441438416e-06,
+      "loss": 1.2736,
+      "step": 1339
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.929143373386296e-06,
+      "loss": 1.2098,
+      "step": 1340
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.927379027101994e-06,
+      "loss": 1.3307,
+      "step": 1341
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.925613403159494e-06,
+      "loss": 1.2805,
+      "step": 1342
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.923846502133186e-06,
+      "loss": 1.0778,
+      "step": 1343
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.92207832459788e-06,
+      "loss": 1.2537,
+      "step": 1344
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.920308871128802e-06,
+      "loss": 1.3151,
+      "step": 1345
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.918538142301591e-06,
+      "loss": 1.2394,
+      "step": 1346
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.916766138692303e-06,
+      "loss": 1.203,
+      "step": 1347
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.914992860877408e-06,
+      "loss": 1.3109,
+      "step": 1348
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.91321830943379e-06,
+      "loss": 1.0943,
+      "step": 1349
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.911442484938747e-06,
+      "loss": 1.2468,
+      "step": 1350
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.90966538796999e-06,
+      "loss": 1.2814,
+      "step": 1351
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.90788701910565e-06,
+      "loss": 1.2279,
+      "step": 1352
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.906107378924262e-06,
+      "loss": 1.2454,
+      "step": 1353
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.904326468004787e-06,
+      "loss": 1.259,
+      "step": 1354
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.902544286926585e-06,
+      "loss": 1.2809,
+      "step": 1355
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.900760836269442e-06,
+      "loss": 1.1576,
+      "step": 1356
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.898976116613548e-06,
+      "loss": 1.2902,
+      "step": 1357
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.897190128539512e-06,
+      "loss": 1.193,
+      "step": 1358
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.895402872628352e-06,
+      "loss": 1.2486,
+      "step": 1359
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.893614349461502e-06,
+      "loss": 1.3224,
+      "step": 1360
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.891824559620801e-06,
+      "loss": 1.2892,
+      "step": 1361
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.89003350368851e-06,
+      "loss": 1.054,
+      "step": 1362
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.88824118224729e-06,
+      "loss": 1.2134,
+      "step": 1363
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.886447595880228e-06,
+      "loss": 1.2712,
+      "step": 1364
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.884652745170812e-06,
+      "loss": 1.246,
+      "step": 1365
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.882856630702944e-06,
+      "loss": 1.2476,
+      "step": 1366
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.881059253060937e-06,
+      "loss": 1.2589,
+      "step": 1367
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.879260612829517e-06,
+      "loss": 1.3371,
+      "step": 1368
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.877460710593817e-06,
+      "loss": 1.2762,
+      "step": 1369
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.875659546939384e-06,
+      "loss": 1.1897,
+      "step": 1370
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.873857122452174e-06,
+      "loss": 1.2983,
+      "step": 1371
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.872053437718556e-06,
+      "loss": 1.0349,
+      "step": 1372
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.8702484933253e-06,
+      "loss": 1.2335,
+      "step": 1373
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.868442289859595e-06,
+      "loss": 1.3201,
+      "step": 1374
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.86663482790904e-06,
+      "loss": 1.3884,
+      "step": 1375
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.864826108061636e-06,
+      "loss": 1.2251,
+      "step": 1376
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.863016130905795e-06,
+      "loss": 1.2145,
+      "step": 1377
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.861204897030346e-06,
+      "loss": 1.1481,
+      "step": 1378
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.859392407024519e-06,
+      "loss": 1.2877,
+      "step": 1379
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.857578661477954e-06,
+      "loss": 1.2935,
+      "step": 1380
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.855763660980701e-06,
+      "loss": 1.3004,
+      "step": 1381
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.853947406123215e-06,
+      "loss": 1.1942,
+      "step": 1382
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.852129897496367e-06,
+      "loss": 1.2401,
+      "step": 1383
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.850311135691425e-06,
+      "loss": 1.2863,
+      "step": 1384
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.848491121300072e-06,
+      "loss": 1.0802,
+      "step": 1385
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.846669854914395e-06,
+      "loss": 1.2283,
+      "step": 1386
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.844847337126895e-06,
+      "loss": 1.266,
+      "step": 1387
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.843023568530469e-06,
+      "loss": 1.231,
+      "step": 1388
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.84119854971843e-06,
+      "loss": 1.325,
+      "step": 1389
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.839372281284494e-06,
+      "loss": 1.2343,
+      "step": 1390
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.837544763822785e-06,
+      "loss": 1.2534,
+      "step": 1391
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.835715997927832e-06,
+      "loss": 1.2436,
+      "step": 1392
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.833885984194569e-06,
+      "loss": 1.2379,
+      "step": 1393
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.83205472321834e-06,
+      "loss": 1.2604,
+      "step": 1394
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.83022221559489e-06,
+      "loss": 1.251,
+      "step": 1395
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.828388461920375e-06,
+      "loss": 1.2892,
+      "step": 1396
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.826553462791353e-06,
+      "loss": 1.2377,
+      "step": 1397
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.824717218804782e-06,
+      "loss": 1.0786,
+      "step": 1398
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.822879730558035e-06,
+      "loss": 1.2833,
+      "step": 1399
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.821040998648887e-06,
+      "loss": 1.2642,
+      "step": 1400
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.819201023675512e-06,
+      "loss": 1.2329,
+      "step": 1401
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.817359806236492e-06,
+      "loss": 1.3007,
+      "step": 1402
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.815517346930818e-06,
+      "loss": 1.3741,
+      "step": 1403
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.813673646357873e-06,
+      "loss": 1.2149,
+      "step": 1404
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.81182870511746e-06,
+      "loss": 1.2063,
+      "step": 1405
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.80998252380977e-06,
+      "loss": 1.3164,
+      "step": 1406
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.808135103035407e-06,
+      "loss": 1.139,
+      "step": 1407
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.806286443395375e-06,
+      "loss": 1.2935,
+      "step": 1408
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.804436545491082e-06,
+      "loss": 1.2093,
+      "step": 1409
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.802585409924338e-06,
+      "loss": 1.2993,
+      "step": 1410
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.800733037297353e-06,
+      "loss": 1.2674,
+      "step": 1411
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.798879428212748e-06,
+      "loss": 1.2803,
+      "step": 1412
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.797024583273536e-06,
+      "loss": 1.2279,
+      "step": 1413
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.79516850308314e-06,
+      "loss": 1.0618,
+      "step": 1414
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.79331118824538e-06,
+      "loss": 1.2654,
+      "step": 1415
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.791452639364478e-06,
+      "loss": 1.2334,
+      "step": 1416
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.789592857045059e-06,
+      "loss": 1.2559,
+      "step": 1417
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.787731841892154e-06,
+      "loss": 1.2766,
+      "step": 1418
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.785869594511182e-06,
+      "loss": 1.3081,
+      "step": 1419
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.784006115507976e-06,
+      "loss": 1.2024,
+      "step": 1420
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.782141405488762e-06,
+      "loss": 1.3292,
+      "step": 1421
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.780275465060173e-06,
+      "loss": 1.2499,
+      "step": 1422
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.778408294829232e-06,
+      "loss": 1.2638,
+      "step": 1423
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.776539895403375e-06,
+      "loss": 1.2562,
+      "step": 1424
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.774670267390426e-06,
+      "loss": 1.2977,
+      "step": 1425
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.772799411398618e-06,
+      "loss": 1.2006,
+      "step": 1426
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.770927328036575e-06,
+      "loss": 1.1544,
+      "step": 1427
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.769054017913329e-06,
+      "loss": 1.3146,
+      "step": 1428
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.767179481638303e-06,
+      "loss": 1.2038,
+      "step": 1429
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.765303719821328e-06,
+      "loss": 1.2058,
+      "step": 1430
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.763426733072624e-06,
+      "loss": 1.2495,
+      "step": 1431
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.761548522002814e-06,
+      "loss": 1.2731,
+      "step": 1432
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.75966908722292e-06,
+      "loss": 1.2271,
+      "step": 1433
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.757788429344362e-06,
+      "loss": 1.0517,
+      "step": 1434
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.755906548978957e-06,
+      "loss": 1.3167,
+      "step": 1435
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.754023446738918e-06,
+      "loss": 1.2225,
+      "step": 1436
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.75213912323686e-06,
+      "loss": 1.2441,
+      "step": 1437
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.750253579085792e-06,
+      "loss": 1.337,
+      "step": 1438
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.748366814899119e-06,
+      "loss": 1.2089,
+      "step": 1439
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.746478831290648e-06,
+      "loss": 1.2488,
+      "step": 1440
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.744589628874576e-06,
+      "loss": 1.2505,
+      "step": 1441
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.742699208265502e-06,
+      "loss": 1.2203,
+      "step": 1442
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.740807570078419e-06,
+      "loss": 1.0874,
+      "step": 1443
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.738914714928712e-06,
+      "loss": 1.2423,
+      "step": 1444
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.737020643432174e-06,
+      "loss": 1.2219,
+      "step": 1445
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.735125356204982e-06,
+      "loss": 1.2858,
+      "step": 1446
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.73322885386371e-06,
+      "loss": 1.2371,
+      "step": 1447
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.731331137025333e-06,
+      "loss": 1.3179,
+      "step": 1448
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.729432206307218e-06,
+      "loss": 1.2549,
+      "step": 1449
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.727532062327124e-06,
+      "loss": 1.0428,
+      "step": 1450
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.72563070570321e-06,
+      "loss": 1.3534,
+      "step": 1451
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.723728137054026e-06,
+      "loss": 1.3227,
+      "step": 1452
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.721824356998518e-06,
+      "loss": 1.238,
+      "step": 1453
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.719919366156022e-06,
+      "loss": 1.2105,
+      "step": 1454
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.718013165146275e-06,
+      "loss": 1.2109,
+      "step": 1455
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.716105754589404e-06,
+      "loss": 1.1568,
+      "step": 1456
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.714197135105925e-06,
+      "loss": 1.2121,
+      "step": 1457
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.712287307316756e-06,
+      "loss": 1.3004,
+      "step": 1458
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.710376271843203e-06,
+      "loss": 1.2664,
+      "step": 1459
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.708464029306965e-06,
+      "loss": 1.2345,
+      "step": 1460
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.706550580330133e-06,
+      "loss": 1.2201,
+      "step": 1461
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.704635925535194e-06,
+      "loss": 1.2537,
+      "step": 1462
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.702720065545024e-06,
+      "loss": 0.9174,
+      "step": 1463
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.700803000982892e-06,
+      "loss": 1.2639,
+      "step": 1464
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.69888473247246e-06,
+      "loss": 1.1924,
+      "step": 1465
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.69696526063778e-06,
+      "loss": 1.2735,
+      "step": 1466
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.695044586103297e-06,
+      "loss": 1.2014,
+      "step": 1467
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.693122709493845e-06,
+      "loss": 1.3172,
+      "step": 1468
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.69119963143465e-06,
+      "loss": 1.2468,
+      "step": 1469
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.689275352551332e-06,
+      "loss": 1.2853,
+      "step": 1470
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.687349873469897e-06,
+      "loss": 1.2431,
+      "step": 1471
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.685423194816744e-06,
+      "loss": 1.2187,
+      "step": 1472
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.683495317218664e-06,
+      "loss": 1.303,
+      "step": 1473
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.68156624130283e-06,
+      "loss": 1.271,
+      "step": 1474
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.679635967696815e-06,
+      "loss": 1.2064,
+      "step": 1475
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.677704497028579e-06,
+      "loss": 1.2417,
+      "step": 1476
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.675771829926465e-06,
+      "loss": 1.2272,
+      "step": 1477
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.673837967019209e-06,
+      "loss": 1.2612,
+      "step": 1478
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.671902908935942e-06,
+      "loss": 1.0605,
+      "step": 1479
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.669966656306176e-06,
+      "loss": 1.2429,
+      "step": 1480
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.668029209759817e-06,
+      "loss": 1.3021,
+      "step": 1481
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.666090569927151e-06,
+      "loss": 1.2256,
+      "step": 1482
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.664150737438863e-06,
+      "loss": 1.2998,
+      "step": 1483
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.662209712926018e-06,
+      "loss": 1.2313,
+      "step": 1484
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.660267497020074e-06,
+      "loss": 1.1257,
+      "step": 1485
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.658324090352873e-06,
+      "loss": 1.2556,
+      "step": 1486
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.656379493556645e-06,
+      "loss": 1.1918,
+      "step": 1487
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.65443370726401e-06,
+      "loss": 1.2602,
+      "step": 1488
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.652486732107972e-06,
+      "loss": 1.2212,
+      "step": 1489
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.65053856872192e-06,
+      "loss": 1.278,
+      "step": 1490
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.648589217739635e-06,
+      "loss": 1.2744,
+      "step": 1491
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.64663867979528e-06,
+      "loss": 0.9953,
+      "step": 1492
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.644686955523408e-06,
+      "loss": 1.2098,
+      "step": 1493
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.642734045558952e-06,
+      "loss": 1.289,
+      "step": 1494
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.640779950537237e-06,
+      "loss": 1.246,
+      "step": 1495
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.63882467109397e-06,
+      "loss": 1.2149,
+      "step": 1496
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.636868207865244e-06,
+      "loss": 1.2085,
+      "step": 1497
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.634910561487538e-06,
+      "loss": 1.2244,
+      "step": 1498
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.632951732597715e-06,
+      "loss": 1.2421,
+      "step": 1499
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.63099172183302e-06,
+      "loss": 1.2409,
+      "step": 1500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.629030529831088e-06,
+      "loss": 1.2365,
+      "step": 1501
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.627068157229937e-06,
+      "loss": 1.2317,
+      "step": 1502
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.625104604667965e-06,
+      "loss": 1.2591,
+      "step": 1503
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.623139872783957e-06,
+      "loss": 1.2228,
+      "step": 1504
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.621173962217083e-06,
+      "loss": 1.0922,
+      "step": 1505
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.619206873606892e-06,
+      "loss": 1.2302,
+      "step": 1506
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.61723860759332e-06,
+      "loss": 1.2478,
+      "step": 1507
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.615269164816684e-06,
+      "loss": 1.3533,
+      "step": 1508
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.613298545917689e-06,
+      "loss": 1.2225,
+      "step": 1509
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.611326751537414e-06,
+      "loss": 1.2306,
+      "step": 1510
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.609353782317325e-06,
+      "loss": 1.2556,
+      "step": 1511
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.607379638899272e-06,
+      "loss": 1.266,
+      "step": 1512
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.605404321925485e-06,
+      "loss": 1.2237,
+      "step": 1513
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.603427832038574e-06,
+      "loss": 1.1634,
+      "step": 1514
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.601450169881533e-06,
+      "loss": 1.2065,
+      "step": 1515
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.599471336097736e-06,
+      "loss": 1.1757,
+      "step": 1516
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.597491331330941e-06,
+      "loss": 1.2511,
+      "step": 1517
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.595510156225281e-06,
+      "loss": 1.2172,
+      "step": 1518
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.59352781142528e-06,
+      "loss": 1.3439,
+      "step": 1519
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.591544297575828e-06,
+      "loss": 1.1786,
+      "step": 1520
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.58955961532221e-06,
+      "loss": 1.0765,
+      "step": 1521
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.587573765310082e-06,
+      "loss": 1.2511,
+      "step": 1522
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.585586748185481e-06,
+      "loss": 1.3031,
+      "step": 1523
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.583598564594827e-06,
+      "loss": 1.2442,
+      "step": 1524
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.581609215184916e-06,
+      "loss": 1.2177,
+      "step": 1525
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.579618700602927e-06,
+      "loss": 1.254,
+      "step": 1526
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.577627021496413e-06,
+      "loss": 0.9054,
+      "step": 1527
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.57563417851331e-06,
+      "loss": 1.2538,
+      "step": 1528
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.573640172301933e-06,
+      "loss": 1.2899,
+      "step": 1529
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.571645003510971e-06,
+      "loss": 1.2339,
+      "step": 1530
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.569648672789496e-06,
+      "loss": 1.2095,
+      "step": 1531
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.567651180786955e-06,
+      "loss": 1.2479,
+      "step": 1532
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.565652528153175e-06,
+      "loss": 1.2382,
+      "step": 1533
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.563652715538356e-06,
+      "loss": 1.0892,
+      "step": 1534
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.561651743593083e-06,
+      "loss": 1.22,
+      "step": 1535
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.55964961296831e-06,
+      "loss": 1.2062,
+      "step": 1536
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.557646324315375e-06,
+      "loss": 1.2781,
+      "step": 1537
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.555641878285987e-06,
+      "loss": 1.2194,
+      "step": 1538
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.553636275532236e-06,
+      "loss": 1.2337,
+      "step": 1539
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.551629516706587e-06,
+      "loss": 1.2788,
+      "step": 1540
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.549621602461877e-06,
+      "loss": 1.1098,
+      "step": 1541
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.547612533451325e-06,
+      "loss": 1.2179,
+      "step": 1542
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.545602310328522e-06,
+      "loss": 1.2997,
+      "step": 1543
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.543590933747437e-06,
+      "loss": 1.1985,
+      "step": 1544
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.541578404362409e-06,
+      "loss": 1.2497,
+      "step": 1545
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.539564722828161e-06,
+      "loss": 1.3329,
+      "step": 1546
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.537549889799781e-06,
+      "loss": 1.2545,
+      "step": 1547
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.535533905932739e-06,
+      "loss": 1.1824,
+      "step": 1548
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.533516771882876e-06,
+      "loss": 1.2169,
+      "step": 1549
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.531498488306404e-06,
+      "loss": 1.0167,
+      "step": 1550
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.529479055859918e-06,
+      "loss": 1.2985,
+      "step": 1551
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.527458475200379e-06,
+      "loss": 1.2417,
+      "step": 1552
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.525436746985125e-06,
+      "loss": 1.2015,
+      "step": 1553
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.523413871871864e-06,
+      "loss": 1.2438,
+      "step": 1554
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.521389850518682e-06,
+      "loss": 1.2726,
+      "step": 1555
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.519364683584033e-06,
+      "loss": 1.2024,
+      "step": 1556
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.51733837172675e-06,
+      "loss": 1.0118,
+      "step": 1557
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.51531091560603e-06,
+      "loss": 1.3116,
+      "step": 1558
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.513282315881449e-06,
+      "loss": 1.2631,
+      "step": 1559
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.511252573212952e-06,
+      "loss": 1.2904,
+      "step": 1560
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.509221688260855e-06,
+      "loss": 1.3039,
+      "step": 1561
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.50718966168585e-06,
+      "loss": 1.2194,
+      "step": 1562
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.505156494148997e-06,
+      "loss": 1.0874,
+      "step": 1563
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.503122186311727e-06,
+      "loss": 1.223,
+      "step": 1564
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.501086738835843e-06,
+      "loss": 1.212,
+      "step": 1565
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.499050152383519e-06,
+      "loss": 1.178,
+      "step": 1566
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.497012427617297e-06,
+      "loss": 1.2173,
+      "step": 1567
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.494973565200093e-06,
+      "loss": 1.2579,
+      "step": 1568
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.492933565795191e-06,
+      "loss": 1.2586,
+      "step": 1569
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.490892430066247e-06,
+      "loss": 0.8891,
+      "step": 1570
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.488850158677283e-06,
+      "loss": 1.2199,
+      "step": 1571
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.486806752292691e-06,
+      "loss": 1.2829,
+      "step": 1572
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.484762211577238e-06,
+      "loss": 1.2171,
+      "step": 1573
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.482716537196052e-06,
+      "loss": 1.2917,
+      "step": 1574
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.480669729814635e-06,
+      "loss": 1.2522,
+      "step": 1575
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.478621790098853e-06,
+      "loss": 1.1998,
+      "step": 1576
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.47657271871495e-06,
+      "loss": 1.2072,
+      "step": 1577
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.474522516329524e-06,
+      "loss": 1.3905,
+      "step": 1578
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.472471183609556e-06,
+      "loss": 0.9779,
+      "step": 1579
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.470418721222382e-06,
+      "loss": 1.2444,
+      "step": 1580
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.468365129835712e-06,
+      "loss": 1.2028,
+      "step": 1581
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.466310410117622e-06,
+      "loss": 1.2075,
+      "step": 1582
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.464254562736556e-06,
+      "loss": 1.2857,
+      "step": 1583
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.462197588361323e-06,
+      "loss": 1.2717,
+      "step": 1584
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.4601394876611e-06,
+      "loss": 1.2679,
+      "step": 1585
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.458080261305432e-06,
+      "loss": 0.9855,
+      "step": 1586
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.456019909964224e-06,
+      "loss": 1.2409,
+      "step": 1587
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.453958434307756e-06,
+      "loss": 1.224,
+      "step": 1588
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.451895835006664e-06,
+      "loss": 1.2416,
+      "step": 1589
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.449832112731958e-06,
+      "loss": 1.2561,
+      "step": 1590
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.447767268155009e-06,
+      "loss": 1.2203,
+      "step": 1591
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.445701301947553e-06,
+      "loss": 1.2337,
+      "step": 1592
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.443634214781693e-06,
+      "loss": 1.2301,
+      "step": 1593
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.441566007329897e-06,
+      "loss": 1.3009,
+      "step": 1594
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.439496680264993e-06,
+      "loss": 1.2427,
+      "step": 1595
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.437426234260176e-06,
+      "loss": 1.2616,
+      "step": 1596
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.435354669989007e-06,
+      "loss": 1.2426,
+      "step": 1597
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.43328198812541e-06,
+      "loss": 1.2363,
+      "step": 1598
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.43120818934367e-06,
+      "loss": 1.1311,
+      "step": 1599
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.429133274318436e-06,
+      "loss": 1.1926,
+      "step": 1600
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.427057243724722e-06,
+      "loss": 1.2016,
+      "step": 1601
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.424980098237904e-06,
+      "loss": 1.217,
+      "step": 1602
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.422901838533722e-06,
+      "loss": 1.2198,
+      "step": 1603
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.420822465288274e-06,
+      "loss": 1.207,
+      "step": 1604
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.418741979178027e-06,
+      "loss": 1.0326,
+      "step": 1605
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.416660380879804e-06,
+      "loss": 1.262,
+      "step": 1606
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.414577671070793e-06,
+      "loss": 1.2752,
+      "step": 1607
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.412493850428545e-06,
+      "loss": 1.2581,
+      "step": 1608
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.410408919630966e-06,
+      "loss": 1.239,
+      "step": 1609
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.408322879356332e-06,
+      "loss": 1.1814,
+      "step": 1610
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.40623573028327e-06,
+      "loss": 1.2264,
+      "step": 1611
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.40414747309078e-06,
+      "loss": 1.0731,
+      "step": 1612
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.402058108458209e-06,
+      "loss": 1.3635,
+      "step": 1613
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.399967637065274e-06,
+      "loss": 1.1452,
+      "step": 1614
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.39787605959205e-06,
+      "loss": 1.1179,
+      "step": 1615
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.395783376718967e-06,
+      "loss": 1.2386,
+      "step": 1616
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.393689589126821e-06,
+      "loss": 1.2046,
+      "step": 1617
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.391594697496764e-06,
+      "loss": 1.2267,
+      "step": 1618
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.38949870251031e-06,
+      "loss": 1.2232,
+      "step": 1619
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.387401604849326e-06,
+      "loss": 1.2303,
+      "step": 1620
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.385303405196043e-06,
+      "loss": 1.1888,
+      "step": 1621
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.383204104233051e-06,
+      "loss": 1.3272,
+      "step": 1622
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.381103702643295e-06,
+      "loss": 1.1706,
+      "step": 1623
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.379002201110077e-06,
+      "loss": 1.2012,
+      "step": 1624
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.376899600317063e-06,
+      "loss": 1.222,
+      "step": 1625
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.374795900948271e-06,
+      "loss": 1.2222,
+      "step": 1626
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.372691103688079e-06,
+      "loss": 1.2267,
+      "step": 1627
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.37058520922122e-06,
+      "loss": 1.1978,
+      "step": 1628
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.368478218232787e-06,
+      "loss": 1.1867,
+      "step": 1629
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.366370131408229e-06,
+      "loss": 1.2147,
+      "step": 1630
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.364260949433345e-06,
+      "loss": 1.2294,
+      "step": 1631
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.3621506729943e-06,
+      "loss": 1.2195,
+      "step": 1632
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.360039302777614e-06,
+      "loss": 1.212,
+      "step": 1633
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.357926839470152e-06,
+      "loss": 0.9802,
+      "step": 1634
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.35581328375915e-06,
+      "loss": 1.2795,
+      "step": 1635
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.353698636332186e-06,
+      "loss": 1.2595,
+      "step": 1636
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.351582897877203e-06,
+      "loss": 1.2295,
+      "step": 1637
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.349466069082492e-06,
+      "loss": 1.2939,
+      "step": 1638
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.347348150636702e-06,
+      "loss": 1.2484,
+      "step": 1639
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.345229143228837e-06,
+      "loss": 1.2356,
+      "step": 1640
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.343109047548253e-06,
+      "loss": 1.1091,
+      "step": 1641
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.340987864284662e-06,
+      "loss": 1.2643,
+      "step": 1642
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.33886559412813e-06,
+      "loss": 1.248,
+      "step": 1643
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.336742237769073e-06,
+      "loss": 1.2271,
+      "step": 1644
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.334617795898266e-06,
+      "loss": 1.2173,
+      "step": 1645
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.332492269206834e-06,
+      "loss": 1.2173,
+      "step": 1646
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.330365658386252e-06,
+      "loss": 1.2158,
+      "step": 1647
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.328237964128353e-06,
+      "loss": 1.3372,
+      "step": 1648
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.32610918712532e-06,
+      "loss": 1.2194,
+      "step": 1649
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.323979328069689e-06,
+      "loss": 1.0334,
+      "step": 1650
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.321848387654347e-06,
+      "loss": 1.2492,
+      "step": 1651
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.319716366572535e-06,
+      "loss": 1.2548,
+      "step": 1652
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.31758326551784e-06,
+      "loss": 1.2891,
+      "step": 1653
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.315449085184209e-06,
+      "loss": 1.2104,
+      "step": 1654
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.313313826265932e-06,
+      "loss": 1.2874,
+      "step": 1655
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.311177489457653e-06,
+      "loss": 1.2181,
+      "step": 1656
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.309040075454366e-06,
+      "loss": 1.0701,
+      "step": 1657
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.306901584951422e-06,
+      "loss": 1.2924,
+      "step": 1658
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.30476201864451e-06,
+      "loss": 1.2221,
+      "step": 1659
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.302621377229679e-06,
+      "loss": 1.1964,
+      "step": 1660
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.300479661403323e-06,
+      "loss": 1.2187,
+      "step": 1661
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.29833687186219e-06,
+      "loss": 1.2779,
+      "step": 1662
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.296193009303369e-06,
+      "loss": 1.1996,
+      "step": 1663
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.294048074424303e-06,
+      "loss": 1.0649,
+      "step": 1664
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.291902067922791e-06,
+      "loss": 1.2157,
+      "step": 1665
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.289754990496965e-06,
+      "loss": 1.2091,
+      "step": 1666
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.28760684284532e-06,
+      "loss": 1.2948,
+      "step": 1667
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.28545762566669e-06,
+      "loss": 1.2538,
+      "step": 1668
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.283307339660261e-06,
+      "loss": 1.2716,
+      "step": 1669
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.281155985525568e-06,
+      "loss": 1.2587,
+      "step": 1670
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.27900356396249e-06,
+      "loss": 1.2239,
+      "step": 1671
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.27685007567125e-06,
+      "loss": 1.2089,
+      "step": 1672
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.27469552135243e-06,
+      "loss": 1.2106,
+      "step": 1673
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.272539901706946e-06,
+      "loss": 1.2271,
+      "step": 1674
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.27038321743607e-06,
+      "loss": 1.3006,
+      "step": 1675
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.268225469241413e-06,
+      "loss": 1.2496,
+      "step": 1676
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.266066657824936e-06,
+      "loss": 1.0698,
+      "step": 1677
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.263906783888947e-06,
+      "loss": 1.2539,
+      "step": 1678
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.261745848136096e-06,
+      "loss": 1.2327,
+      "step": 1679
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.259583851269384e-06,
+      "loss": 1.2423,
+      "step": 1680
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.257420793992149e-06,
+      "loss": 1.2524,
+      "step": 1681
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.255256677008081e-06,
+      "loss": 1.314,
+      "step": 1682
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.25309150102121e-06,
+      "loss": 1.2385,
+      "step": 1683
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.25092526673592e-06,
+      "loss": 1.3297,
+      "step": 1684
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.248757974856923e-06,
+      "loss": 1.2367,
+      "step": 1685
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.246589626089287e-06,
+      "loss": 0.9321,
+      "step": 1686
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.244420221138424e-06,
+      "loss": 1.1974,
+      "step": 1687
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.242249760710084e-06,
+      "loss": 1.261,
+      "step": 1688
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.240078245510364e-06,
+      "loss": 1.2107,
+      "step": 1689
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.237905676245702e-06,
+      "loss": 1.1974,
+      "step": 1690
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.235732053622883e-06,
+      "loss": 1.285,
+      "step": 1691
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.233557378349029e-06,
+      "loss": 1.2935,
+      "step": 1692
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.231381651131606e-06,
+      "loss": 1.0804,
+      "step": 1693
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.229204872678427e-06,
+      "loss": 1.272,
+      "step": 1694
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.227027043697642e-06,
+      "loss": 1.3144,
+      "step": 1695
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.224848164897743e-06,
+      "loss": 1.1915,
+      "step": 1696
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.222668236987563e-06,
+      "loss": 1.2829,
+      "step": 1697
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.220487260676283e-06,
+      "loss": 1.2687,
+      "step": 1698
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.218305236673415e-06,
+      "loss": 1.1722,
+      "step": 1699
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.216122165688822e-06,
+      "loss": 1.2583,
+      "step": 1700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.213938048432697e-06,
+      "loss": 1.2811,
+      "step": 1701
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.211752885615583e-06,
+      "loss": 1.2123,
+      "step": 1702
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.209566677948358e-06,
+      "loss": 1.2311,
+      "step": 1703
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.20737942614224e-06,
+      "loss": 1.2493,
+      "step": 1704
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.205191130908785e-06,
+      "loss": 1.2364,
+      "step": 1705
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.203001792959897e-06,
+      "loss": 1.0281,
+      "step": 1706
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.200811413007808e-06,
+      "loss": 1.2462,
+      "step": 1707
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.198619991765098e-06,
+      "loss": 1.3321,
+      "step": 1708
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.19642752994468e-06,
+      "loss": 1.2476,
+      "step": 1709
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.194234028259806e-06,
+      "loss": 1.2832,
+      "step": 1710
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.192039487424073e-06,
+      "loss": 1.2329,
+      "step": 1711
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.189843908151404e-06,
+      "loss": 1.1692,
+      "step": 1712
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.187647291156074e-06,
+      "loss": 1.2402,
+      "step": 1713
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.185449637152681e-06,
+      "loss": 1.2508,
+      "step": 1714
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.183250946856173e-06,
+      "loss": 1.2496,
+      "step": 1715
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.181051220981826e-06,
+      "loss": 1.2847,
+      "step": 1716
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.17885046024526e-06,
+      "loss": 1.1504,
+      "step": 1717
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.176648665362426e-06,
+      "loss": 1.303,
+      "step": 1718
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.174445837049614e-06,
+      "loss": 1.2932,
+      "step": 1719
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.172241976023451e-06,
+      "loss": 1.2028,
+      "step": 1720
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.170037083000898e-06,
+      "loss": 1.2056,
+      "step": 1721
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.167831158699254e-06,
+      "loss": 1.102,
+      "step": 1722
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.16562420383615e-06,
+      "loss": 1.2675,
+      "step": 1723
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.163416219129556e-06,
+      "loss": 1.2627,
+      "step": 1724
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.161207205297774e-06,
+      "loss": 1.2578,
+      "step": 1725
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.158997163059444e-06,
+      "loss": 1.3005,
+      "step": 1726
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.156786093133537e-06,
+      "loss": 1.1992,
+      "step": 1727
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.15457399623936e-06,
+      "loss": 1.1546,
+      "step": 1728
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.152360873096558e-06,
+      "loss": 1.2563,
+      "step": 1729
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.1501467244251e-06,
+      "loss": 1.3052,
+      "step": 1730
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.147931550945301e-06,
+      "loss": 1.235,
+      "step": 1731
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.145715353377797e-06,
+      "loss": 1.2428,
+      "step": 1732
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.143498132443568e-06,
+      "loss": 1.2901,
+      "step": 1733
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.14127988886392e-06,
+      "loss": 1.2471,
+      "step": 1734
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.139060623360494e-06,
+      "loss": 1.1248,
+      "step": 1735
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.136840336655263e-06,
+      "loss": 1.2589,
+      "step": 1736
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.134619029470535e-06,
+      "loss": 1.2825,
+      "step": 1737
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.132396702528941e-06,
+      "loss": 1.2452,
+      "step": 1738
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.130173356553459e-06,
+      "loss": 1.1943,
+      "step": 1739
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.127948992267383e-06,
+      "loss": 1.2354,
+      "step": 1740
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.12572361039435e-06,
+      "loss": 1.1244,
+      "step": 1741
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.123497211658317e-06,
+      "loss": 1.2258,
+      "step": 1742
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.121269796783585e-06,
+      "loss": 1.2217,
+      "step": 1743
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.119041366494771e-06,
+      "loss": 1.2049,
+      "step": 1744
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.116811921516837e-06,
+      "loss": 1.2703,
+      "step": 1745
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.114581462575063e-06,
+      "loss": 1.2663,
+      "step": 1746
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.112349990395065e-06,
+      "loss": 1.1943,
+      "step": 1747
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.11011750570279e-06,
+      "loss": 1.0512,
+      "step": 1748
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.107884009224508e-06,
+      "loss": 1.2675,
+      "step": 1749
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.105649501686823e-06,
+      "loss": 1.2243,
+      "step": 1750
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.103413983816666e-06,
+      "loss": 1.2459,
+      "step": 1751
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.101177456341301e-06,
+      "loss": 1.2311,
+      "step": 1752
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.098939919988315e-06,
+      "loss": 1.3024,
+      "step": 1753
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.096701375485622e-06,
+      "loss": 1.2261,
+      "step": 1754
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.094461823561473e-06,
+      "loss": 1.2098,
+      "step": 1755
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.092221264944438e-06,
+      "loss": 1.2479,
+      "step": 1756
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.089979700363415e-06,
+      "loss": 1.0453,
+      "step": 1757
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.087737130547637e-06,
+      "loss": 1.1846,
+      "step": 1758
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.085493556226653e-06,
+      "loss": 1.1772,
+      "step": 1759
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.083248978130348e-06,
+      "loss": 1.2297,
+      "step": 1760
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.081003396988929e-06,
+      "loss": 1.2254,
+      "step": 1761
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.07875681353293e-06,
+      "loss": 1.2438,
+      "step": 1762
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.076509228493215e-06,
+      "loss": 1.2122,
+      "step": 1763
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.074260642600963e-06,
+      "loss": 1.0697,
+      "step": 1764
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.072011056587691e-06,
+      "loss": 1.263,
+      "step": 1765
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.069760471185237e-06,
+      "loss": 1.18,
+      "step": 1766
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.06750888712576e-06,
+      "loss": 1.194,
+      "step": 1767
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.065256305141749e-06,
+      "loss": 1.1981,
+      "step": 1768
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.063002725966014e-06,
+      "loss": 1.2517,
+      "step": 1769
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.060748150331696e-06,
+      "loss": 1.1678,
+      "step": 1770
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.058492578972248e-06,
+      "loss": 1.0783,
+      "step": 1771
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.056236012621461e-06,
+      "loss": 1.1835,
+      "step": 1772
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.05397845201344e-06,
+      "loss": 1.2167,
+      "step": 1773
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.051719897882617e-06,
+      "loss": 1.2327,
+      "step": 1774
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.049460350963747e-06,
+      "loss": 1.2936,
+      "step": 1775
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.047199811991906e-06,
+      "loss": 1.2613,
+      "step": 1776
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.044938281702496e-06,
+      "loss": 1.1816,
+      "step": 1777
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.04267576083124e-06,
+      "loss": 1.2379,
+      "step": 1778
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.040412250114184e-06,
+      "loss": 1.2115,
+      "step": 1779
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.038147750287692e-06,
+      "loss": 1.2578,
+      "step": 1780
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.035882262088456e-06,
+      "loss": 1.2758,
+      "step": 1781
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.033615786253486e-06,
+      "loss": 1.2948,
+      "step": 1782
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.031348323520113e-06,
+      "loss": 1.2178,
+      "step": 1783
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.029079874625989e-06,
+      "loss": 1.035,
+      "step": 1784
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.02681044030909e-06,
+      "loss": 1.2421,
+      "step": 1785
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.024540021307709e-06,
+      "loss": 1.2576,
+      "step": 1786
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.02226861836046e-06,
+      "loss": 1.2593,
+      "step": 1787
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.01999623220628e-06,
+      "loss": 1.3122,
+      "step": 1788
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.01772286358442e-06,
+      "loss": 1.2475,
+      "step": 1789
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.015448513234457e-06,
+      "loss": 1.2081,
+      "step": 1790
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.013173181896283e-06,
+      "loss": 1.2084,
+      "step": 1791
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.010896870310111e-06,
+      "loss": 1.1778,
+      "step": 1792
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.008619579216473e-06,
+      "loss": 1.2205,
+      "step": 1793
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.006341309356221e-06,
+      "loss": 1.2035,
+      "step": 1794
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.004062061470519e-06,
+      "loss": 1.2507,
+      "step": 1795
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.001781836300858e-06,
+      "loss": 1.2464,
+      "step": 1796
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.999500634589041e-06,
+      "loss": 1.2223,
+      "step": 1797
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.99721845707719e-06,
+      "loss": 1.2326,
+      "step": 1798
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.994935304507748e-06,
+      "loss": 1.2552,
+      "step": 1799
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.992651177623466e-06,
+      "loss": 1.0174,
+      "step": 1800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.990366077167425e-06,
+      "loss": 1.2842,
+      "step": 1801
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.98808000388301e-06,
+      "loss": 1.2675,
+      "step": 1802
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.985792958513932e-06,
+      "loss": 1.294,
+      "step": 1803
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.983504941804213e-06,
+      "loss": 1.2295,
+      "step": 1804
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.981215954498193e-06,
+      "loss": 1.2038,
+      "step": 1805
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.978925997340525e-06,
+      "loss": 1.14,
+      "step": 1806
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.976635071076185e-06,
+      "loss": 1.2797,
+      "step": 1807
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.974343176450455e-06,
+      "loss": 1.2713,
+      "step": 1808
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.972050314208934e-06,
+      "loss": 1.2542,
+      "step": 1809
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.969756485097543e-06,
+      "loss": 1.1622,
+      "step": 1810
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.96746168986251e-06,
+      "loss": 1.2251,
+      "step": 1811
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.96516592925038e-06,
+      "loss": 1.2256,
+      "step": 1812
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.96286920400801e-06,
+      "loss": 0.9844,
+      "step": 1813
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.960571514882576e-06,
+      "loss": 1.2232,
+      "step": 1814
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.958272862621562e-06,
+      "loss": 1.3009,
+      "step": 1815
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.955973247972765e-06,
+      "loss": 1.23,
+      "step": 1816
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.953672671684305e-06,
+      "loss": 1.2002,
+      "step": 1817
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.951371134504599e-06,
+      "loss": 1.1857,
+      "step": 1818
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.949068637182389e-06,
+      "loss": 0.9887,
+      "step": 1819
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.946765180466725e-06,
+      "loss": 1.2534,
+      "step": 1820
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.944460765106969e-06,
+      "loss": 1.1757,
+      "step": 1821
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.942155391852797e-06,
+      "loss": 1.2282,
+      "step": 1822
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.939849061454192e-06,
+      "loss": 1.2834,
+      "step": 1823
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.937541774661455e-06,
+      "loss": 1.2784,
+      "step": 1824
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.93523353222519e-06,
+      "loss": 1.1905,
+      "step": 1825
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.93292433489632e-06,
+      "loss": 1.2234,
+      "step": 1826
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.930614183426074e-06,
+      "loss": 1.1834,
+      "step": 1827
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.92830307856599e-06,
+      "loss": 1.2588,
+      "step": 1828
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.925991021067924e-06,
+      "loss": 1.0211,
+      "step": 1829
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.92367801168403e-06,
+      "loss": 1.2265,
+      "step": 1830
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.921364051166785e-06,
+      "loss": 1.1933,
+      "step": 1831
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.919049140268962e-06,
+      "loss": 1.1364,
+      "step": 1832
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.916733279743653e-06,
+      "loss": 1.1561,
+      "step": 1833
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.914416470344255e-06,
+      "loss": 1.2359,
+      "step": 1834
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.912098712824474e-06,
+      "loss": 1.0627,
+      "step": 1835
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.909780007938327e-06,
+      "loss": 1.2631,
+      "step": 1836
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.907460356440133e-06,
+      "loss": 1.2504,
+      "step": 1837
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.905139759084529e-06,
+      "loss": 1.2464,
+      "step": 1838
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.902818216626446e-06,
+      "loss": 1.302,
+      "step": 1839
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.900495729821136e-06,
+      "loss": 1.2127,
+      "step": 1840
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.89817229942415e-06,
+      "loss": 1.2432,
+      "step": 1841
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.89584792619135e-06,
+      "loss": 1.033,
+      "step": 1842
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.893522610878898e-06,
+      "loss": 1.2819,
+      "step": 1843
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.891196354243276e-06,
+      "loss": 1.1456,
+      "step": 1844
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.888869157041257e-06,
+      "loss": 1.219,
+      "step": 1845
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.886541020029927e-06,
+      "loss": 1.213,
+      "step": 1846
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.884211943966683e-06,
+      "loss": 1.2139,
+      "step": 1847
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.881881929609215e-06,
+      "loss": 1.0382,
+      "step": 1848
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.879550977715528e-06,
+      "loss": 1.2723,
+      "step": 1849
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.87721908904393e-06,
+      "loss": 1.2574,
+      "step": 1850
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.874886264353035e-06,
+      "loss": 1.1945,
+      "step": 1851
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.872552504401754e-06,
+      "loss": 1.1939,
+      "step": 1852
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.87021780994931e-06,
+      "loss": 1.2352,
+      "step": 1853
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.86788218175523e-06,
+      "loss": 1.2998,
+      "step": 1854
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.865545620579342e-06,
+      "loss": 0.962,
+      "step": 1855
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.863208127181777e-06,
+      "loss": 1.2564,
+      "step": 1856
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.860869702322969e-06,
+      "loss": 1.2401,
+      "step": 1857
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.858530346763657e-06,
+      "loss": 1.3397,
+      "step": 1858
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.856190061264882e-06,
+      "loss": 1.2118,
+      "step": 1859
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.85384884658799e-06,
+      "loss": 1.2543,
+      "step": 1860
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.851506703494624e-06,
+      "loss": 1.2831,
+      "step": 1861
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.849163632746735e-06,
+      "loss": 1.2412,
+      "step": 1862
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.846819635106569e-06,
+      "loss": 1.2304,
+      "step": 1863
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.844474711336678e-06,
+      "loss": 1.0954,
+      "step": 1864
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.842128862199918e-06,
+      "loss": 1.217,
+      "step": 1865
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.839782088459438e-06,
+      "loss": 1.2654,
+      "step": 1866
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.837434390878697e-06,
+      "loss": 1.2927,
+      "step": 1867
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.835085770221446e-06,
+      "loss": 1.2249,
+      "step": 1868
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.832736227251742e-06,
+      "loss": 1.332,
+      "step": 1869
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.830385762733939e-06,
+      "loss": 1.3279,
+      "step": 1870
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.828034377432694e-06,
+      "loss": 0.9317,
+      "step": 1871
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.82568207211296e-06,
+      "loss": 1.3223,
+      "step": 1872
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.823328847539993e-06,
+      "loss": 1.2413,
+      "step": 1873
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.820974704479343e-06,
+      "loss": 1.2654,
+      "step": 1874
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.818619643696863e-06,
+      "loss": 1.2324,
+      "step": 1875
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.816263665958705e-06,
+      "loss": 1.2896,
+      "step": 1876
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.813906772031315e-06,
+      "loss": 1.2568,
+      "step": 1877
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.811548962681441e-06,
+      "loss": 1.0509,
+      "step": 1878
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.80919023867613e-06,
+      "loss": 1.3152,
+      "step": 1879
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.806830600782717e-06,
+      "loss": 1.2162,
+      "step": 1880
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.80447004976885e-06,
+      "loss": 1.2897,
+      "step": 1881
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.802108586402459e-06,
+      "loss": 1.1857,
+      "step": 1882
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.79974621145178e-06,
+      "loss": 1.2942,
+      "step": 1883
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.797382925685342e-06,
+      "loss": 1.007,
+      "step": 1884
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.795018729871973e-06,
+      "loss": 1.3372,
+      "step": 1885
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.792653624780791e-06,
+      "loss": 1.2022,
+      "step": 1886
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.790287611181217e-06,
+      "loss": 1.2086,
+      "step": 1887
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.787920689842965e-06,
+      "loss": 1.2214,
+      "step": 1888
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.785552861536043e-06,
+      "loss": 1.2735,
+      "step": 1889
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.783184127030752e-06,
+      "loss": 1.2111,
+      "step": 1890
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.780814487097696e-06,
+      "loss": 1.1588,
+      "step": 1891
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.778443942507764e-06,
+      "loss": 1.1901,
+      "step": 1892
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.776072494032146e-06,
+      "loss": 1.2919,
+      "step": 1893
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.773700142442322e-06,
+      "loss": 1.2204,
+      "step": 1894
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.77132688851007e-06,
+      "loss": 1.1937,
+      "step": 1895
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.768952733007454e-06,
+      "loss": 1.2089,
+      "step": 1896
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.766577676706839e-06,
+      "loss": 1.2576,
+      "step": 1897
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.764201720380882e-06,
+      "loss": 1.2031,
+      "step": 1898
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.76182486480253e-06,
+      "loss": 1.2641,
+      "step": 1899
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.759447110745022e-06,
+      "loss": 1.2236,
+      "step": 1900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.757068458981894e-06,
+      "loss": 1.2385,
+      "step": 1901
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.754688910286967e-06,
+      "loss": 1.2535,
+      "step": 1902
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.75230846543436e-06,
+      "loss": 1.1747,
+      "step": 1903
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.749927125198482e-06,
+      "loss": 1.2015,
+      "step": 1904
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.747544890354031e-06,
+      "loss": 1.2434,
+      "step": 1905
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.745161761675999e-06,
+      "loss": 1.265,
+      "step": 1906
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.742777739939667e-06,
+      "loss": 0.9687,
+      "step": 1907
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.740392825920605e-06,
+      "loss": 1.2637,
+      "step": 1908
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.738007020394678e-06,
+      "loss": 1.1743,
+      "step": 1909
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.735620324138037e-06,
+      "loss": 1.3294,
+      "step": 1910
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.733232737927123e-06,
+      "loss": 1.2072,
+      "step": 1911
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.730844262538671e-06,
+      "loss": 1.2621,
+      "step": 1912
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.728454898749697e-06,
+      "loss": 1.099,
+      "step": 1913
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.726064647337515e-06,
+      "loss": 1.2356,
+      "step": 1914
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.723673509079718e-06,
+      "loss": 1.234,
+      "step": 1915
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.721281484754201e-06,
+      "loss": 1.2706,
+      "step": 1916
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.718888575139134e-06,
+      "loss": 1.2396,
+      "step": 1917
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.716494781012982e-06,
+      "loss": 1.2423,
+      "step": 1918
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.714100103154494e-06,
+      "loss": 1.2178,
+      "step": 1919
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.711704542342711e-06,
+      "loss": 1.0276,
+      "step": 1920
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.709308099356957e-06,
+      "loss": 1.2325,
+      "step": 1921
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.706910774976849e-06,
+      "loss": 1.2205,
+      "step": 1922
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.70451256998228e-06,
+      "loss": 1.196,
+      "step": 1923
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.702113485153443e-06,
+      "loss": 1.2581,
+      "step": 1924
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.699713521270804e-06,
+      "loss": 1.2155,
+      "step": 1925
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.697312679115126e-06,
+      "loss": 1.0248,
+      "step": 1926
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.694910959467447e-06,
+      "loss": 1.2441,
+      "step": 1927
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.692508363109104e-06,
+      "loss": 1.2136,
+      "step": 1928
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.690104890821705e-06,
+      "loss": 1.2209,
+      "step": 1929
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.687700543387158e-06,
+      "loss": 1.296,
+      "step": 1930
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.685295321587636e-06,
+      "loss": 1.2683,
+      "step": 1931
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.682889226205616e-06,
+      "loss": 1.2144,
+      "step": 1932
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.680482258023848e-06,
+      "loss": 1.3112,
+      "step": 1933
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.678074417825372e-06,
+      "loss": 1.2057,
+      "step": 1934
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.675665706393502e-06,
+      "loss": 1.3214,
+      "step": 1935
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.67325612451185e-06,
+      "loss": 1.0357,
+      "step": 1936
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.670845672964296e-06,
+      "loss": 1.2527,
+      "step": 1937
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.668434352535015e-06,
+      "loss": 1.217,
+      "step": 1938
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.666022164008458e-06,
+      "loss": 1.2152,
+      "step": 1939
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.663609108169359e-06,
+      "loss": 1.1657,
+      "step": 1940
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.661195185802737e-06,
+      "loss": 1.2416,
+      "step": 1941
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.65878039769389e-06,
+      "loss": 1.0763,
+      "step": 1942
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.656364744628402e-06,
+      "loss": 1.2198,
+      "step": 1943
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.65394822739213e-06,
+      "loss": 1.2012,
+      "step": 1944
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.651530846771224e-06,
+      "loss": 1.2255,
+      "step": 1945
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.6491126035521e-06,
+      "loss": 1.2871,
+      "step": 1946
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.646693498521472e-06,
+      "loss": 1.2704,
+      "step": 1947
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.644273532466319e-06,
+      "loss": 1.2596,
+      "step": 1948
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.641852706173907e-06,
+      "loss": 1.0862,
+      "step": 1949
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.639431020431783e-06,
+      "loss": 1.2783,
+      "step": 1950
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.637008476027771e-06,
+      "loss": 1.1992,
+      "step": 1951
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.634585073749977e-06,
+      "loss": 1.2439,
+      "step": 1952
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.63216081438678e-06,
+      "loss": 1.1868,
+      "step": 1953
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.629735698726846e-06,
+      "loss": 1.2174,
+      "step": 1954
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.627309727559114e-06,
+      "loss": 1.0283,
+      "step": 1955
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.624882901672801e-06,
+      "loss": 1.2734,
+      "step": 1956
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.622455221857408e-06,
+      "loss": 1.2344,
+      "step": 1957
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.6200266889027086e-06,
+      "loss": 1.1919,
+      "step": 1958
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.617597303598754e-06,
+      "loss": 1.1654,
+      "step": 1959
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.6151670667358736e-06,
+      "loss": 1.2523,
+      "step": 1960
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.612735979104677e-06,
+      "loss": 1.2232,
+      "step": 1961
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.6103040414960415e-06,
+      "loss": 1.1841,
+      "step": 1962
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.6078712547011355e-06,
+      "loss": 1.2428,
+      "step": 1963
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.605437619511388e-06,
+      "loss": 1.3469,
+      "step": 1964
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.6030031367185154e-06,
+      "loss": 1.0642,
+      "step": 1965
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.600567807114503e-06,
+      "loss": 1.2309,
+      "step": 1966
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5981316314916165e-06,
+      "loss": 1.2131,
+      "step": 1967
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.595694610642392e-06,
+      "loss": 1.2408,
+      "step": 1968
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5932567453596475e-06,
+      "loss": 1.2097,
+      "step": 1969
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.590818036436466e-06,
+      "loss": 1.22,
+      "step": 1970
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.588378484666214e-06,
+      "loss": 0.8925,
+      "step": 1971
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.585938090842526e-06,
+      "loss": 1.2902,
+      "step": 1972
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5834968557593155e-06,
+      "loss": 1.2245,
+      "step": 1973
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5810547802107655e-06,
+      "loss": 1.2437,
+      "step": 1974
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.578611864991336e-06,
+      "loss": 1.2021,
+      "step": 1975
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.576168110895754e-06,
+      "loss": 1.2733,
+      "step": 1976
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.573723518719028e-06,
+      "loss": 1.2717,
+      "step": 1977
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.571278089256433e-06,
+      "loss": 1.1204,
+      "step": 1978
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5688318233035175e-06,
+      "loss": 1.2139,
+      "step": 1979
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.566384721656103e-06,
+      "loss": 1.2423,
+      "step": 1980
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.563936785110285e-06,
+      "loss": 1.204,
+      "step": 1981
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.561488014462426e-06,
+      "loss": 1.1881,
+      "step": 1982
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.559038410509161e-06,
+      "loss": 1.1661,
+      "step": 1983
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.556587974047399e-06,
+      "loss": 1.2741,
+      "step": 1984
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5541367058743174e-06,
+      "loss": 1.0148,
+      "step": 1985
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.551684606787366e-06,
+      "loss": 1.2509,
+      "step": 1986
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.549231677584262e-06,
+      "loss": 1.2125,
+      "step": 1987
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.546777919062995e-06,
+      "loss": 1.2268,
+      "step": 1988
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.544323332021826e-06,
+      "loss": 1.2444,
+      "step": 1989
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.541867917259278e-06,
+      "loss": 1.2166,
+      "step": 1990
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.539411675574153e-06,
+      "loss": 1.1319,
+      "step": 1991
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.536954607765519e-06,
+      "loss": 1.1882,
+      "step": 1992
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.534496714632704e-06,
+      "loss": 1.2832,
+      "step": 1993
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.532037996975319e-06,
+      "loss": 1.2394,
+      "step": 1994
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.529578455593232e-06,
+      "loss": 1.2014,
+      "step": 1995
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.527118091286587e-06,
+      "loss": 1.2627,
+      "step": 1996
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5246569048557875e-06,
+      "loss": 1.2616,
+      "step": 1997
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.522194897101511e-06,
+      "loss": 0.963,
+      "step": 1998
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.519732068824699e-06,
+      "loss": 1.3683,
+      "step": 1999
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.517268420826562e-06,
+      "loss": 1.017,
+      "step": 2000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.514803953908575e-06,
+      "loss": 1.2248,
+      "step": 2001
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5123386688724805e-06,
+      "loss": 1.2458,
+      "step": 2002
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.509872566520286e-06,
+      "loss": 1.2386,
+      "step": 2003
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.507405647654271e-06,
+      "loss": 1.2169,
+      "step": 2004
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.50493791307697e-06,
+      "loss": 1.2026,
+      "step": 2005
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.50246936359119e-06,
+      "loss": 1.2303,
+      "step": 2006
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 1.0245,
+      "step": 2007
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.49752982310674e-06,
+      "loss": 1.2993,
+      "step": 2008
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.4950588337150064e-06,
+      "loss": 1.1927,
+      "step": 2009
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.492587032628664e-06,
+      "loss": 1.2337,
+      "step": 2010
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.490114420651842e-06,
+      "loss": 1.1899,
+      "step": 2011
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.487640998588932e-06,
+      "loss": 1.2104,
+      "step": 2012
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.485166767244589e-06,
+      "loss": 1.1826,
+      "step": 2013
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.482691727423735e-06,
+      "loss": 0.9995,
+      "step": 2014
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.480215879931547e-06,
+      "loss": 1.1712,
+      "step": 2015
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.477739225573475e-06,
+      "loss": 1.2472,
+      "step": 2016
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4752617651552225e-06,
+      "loss": 1.1961,
+      "step": 2017
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.472783499482761e-06,
+      "loss": 1.1987,
+      "step": 2018
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.47030442936232e-06,
+      "loss": 1.1786,
+      "step": 2019
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.467824555600396e-06,
+      "loss": 1.0088,
+      "step": 2020
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.465343879003741e-06,
+      "loss": 1.235,
+      "step": 2021
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.46286240037937e-06,
+      "loss": 1.2592,
+      "step": 2022
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.460380120534563e-06,
+      "loss": 1.2631,
+      "step": 2023
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.457897040276853e-06,
+      "loss": 1.1684,
+      "step": 2024
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4554131604140425e-06,
+      "loss": 1.2459,
+      "step": 2025
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.452928481754185e-06,
+      "loss": 1.2213,
+      "step": 2026
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.450443005105601e-06,
+      "loss": 1.0426,
+      "step": 2027
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.447956731276867e-06,
+      "loss": 1.1888,
+      "step": 2028
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.445469661076818e-06,
+      "loss": 1.1761,
+      "step": 2029
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.442981795314552e-06,
+      "loss": 1.2027,
+      "step": 2030
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.440493134799425e-06,
+      "loss": 1.2734,
+      "step": 2031
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.438003680341046e-06,
+      "loss": 1.2391,
+      "step": 2032
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.435513432749289e-06,
+      "loss": 0.9874,
+      "step": 2033
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4330223928342814e-06,
+      "loss": 1.3339,
+      "step": 2034
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4305305614064145e-06,
+      "loss": 1.2207,
+      "step": 2035
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.428037939276327e-06,
+      "loss": 1.073,
+      "step": 2036
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.425544527254926e-06,
+      "loss": 1.2113,
+      "step": 2037
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4230503261533694e-06,
+      "loss": 1.1683,
+      "step": 2038
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4205553367830704e-06,
+      "loss": 1.2642,
+      "step": 2039
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.418059559955703e-06,
+      "loss": 1.2158,
+      "step": 2040
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.415562996483193e-06,
+      "loss": 1.2189,
+      "step": 2041
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4130656471777285e-06,
+      "loss": 1.1859,
+      "step": 2042
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4105675128517456e-06,
+      "loss": 1.0103,
+      "step": 2043
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.408068594317941e-06,
+      "loss": 1.273,
+      "step": 2044
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.405568892389265e-06,
+      "loss": 1.2888,
+      "step": 2045
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.403068407878923e-06,
+      "loss": 1.2011,
+      "step": 2046
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.400567141600374e-06,
+      "loss": 1.1857,
+      "step": 2047
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.39806509436733e-06,
+      "loss": 1.2063,
+      "step": 2048
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3955622669937646e-06,
+      "loss": 1.0239,
+      "step": 2049
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.393058660293895e-06,
+      "loss": 1.2121,
+      "step": 2050
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.390554275082198e-06,
+      "loss": 1.2219,
+      "step": 2051
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.388049112173401e-06,
+      "loss": 1.2335,
+      "step": 2052
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.385543172382489e-06,
+      "loss": 1.169,
+      "step": 2053
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.383036456524692e-06,
+      "loss": 1.2702,
+      "step": 2054
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.380528965415501e-06,
+      "loss": 1.2568,
+      "step": 2055
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3780206998706535e-06,
+      "loss": 1.0009,
+      "step": 2056
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3755116607061395e-06,
+      "loss": 1.2038,
+      "step": 2057
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.373001848738203e-06,
+      "loss": 1.2549,
+      "step": 2058
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.370491264783337e-06,
+      "loss": 1.2098,
+      "step": 2059
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.367979909658287e-06,
+      "loss": 1.1778,
+      "step": 2060
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.365467784180051e-06,
+      "loss": 1.1974,
+      "step": 2061
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.362954889165874e-06,
+      "loss": 1.0667,
+      "step": 2062
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.360441225433252e-06,
+      "loss": 1.1897,
+      "step": 2063
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.357926793799936e-06,
+      "loss": 1.224,
+      "step": 2064
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.355411595083918e-06,
+      "loss": 1.1777,
+      "step": 2065
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.352895630103448e-06,
+      "loss": 1.2489,
+      "step": 2066
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.35037889967702e-06,
+      "loss": 1.2427,
+      "step": 2067
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.347861404623382e-06,
+      "loss": 1.2056,
+      "step": 2068
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.345343145761522e-06,
+      "loss": 1.2776,
+      "step": 2069
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.342824123910688e-06,
+      "loss": 1.2131,
+      "step": 2070
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.340304339890366e-06,
+      "loss": 1.2311,
+      "step": 2071
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3377837945202965e-06,
+      "loss": 1.0473,
+      "step": 2072
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.3352624886204645e-06,
+      "loss": 1.1647,
+      "step": 2073
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.3327404230111045e-06,
+      "loss": 1.2102,
+      "step": 2074
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.330217598512696e-06,
+      "loss": 1.2608,
+      "step": 2075
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.327694015945966e-06,
+      "loss": 1.1709,
+      "step": 2076
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.325169676131887e-06,
+      "loss": 1.2807,
+      "step": 2077
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.322644579891683e-06,
+      "loss": 0.8728,
+      "step": 2078
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.320118728046818e-06,
+      "loss": 1.218,
+      "step": 2079
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.317592121419005e-06,
+      "loss": 1.2043,
+      "step": 2080
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.315064760830201e-06,
+      "loss": 1.1737,
+      "step": 2081
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.312536647102611e-06,
+      "loss": 1.2824,
+      "step": 2082
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.310007781058681e-06,
+      "loss": 1.1889,
+      "step": 2083
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.307478163521104e-06,
+      "loss": 1.2194,
+      "step": 2084
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.304947795312818e-06,
+      "loss": 1.1598,
+      "step": 2085
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.302416677257005e-06,
+      "loss": 1.1894,
+      "step": 2086
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.299884810177088e-06,
+      "loss": 1.2056,
+      "step": 2087
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.297352194896738e-06,
+      "loss": 1.2422,
+      "step": 2088
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.294818832239869e-06,
+      "loss": 1.1721,
+      "step": 2089
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.292284723030634e-06,
+      "loss": 1.2764,
+      "step": 2090
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.289749868093432e-06,
+      "loss": 1.2184,
+      "step": 2091
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2872142682529045e-06,
+      "loss": 1.0426,
+      "step": 2092
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.284677924333935e-06,
+      "loss": 1.1225,
+      "step": 2093
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2821408371616485e-06,
+      "loss": 1.251,
+      "step": 2094
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.279603007561415e-06,
+      "loss": 1.2283,
+      "step": 2095
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.277064436358838e-06,
+      "loss": 1.1855,
+      "step": 2096
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.274525124379773e-06,
+      "loss": 1.2394,
+      "step": 2097
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.271985072450307e-06,
+      "loss": 0.9772,
+      "step": 2098
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.269444281396776e-06,
+      "loss": 1.239,
+      "step": 2099
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.266902752045747e-06,
+      "loss": 1.2844,
+      "step": 2100
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.264360485224037e-06,
+      "loss": 1.2394,
+      "step": 2101
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.261817481758695e-06,
+      "loss": 1.2453,
+      "step": 2102
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.259273742477017e-06,
+      "loss": 1.2274,
+      "step": 2103
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.256729268206528e-06,
+      "loss": 1.2616,
+      "step": 2104
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.254184059775007e-06,
+      "loss": 1.2415,
+      "step": 2105
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.251638118010456e-06,
+      "loss": 1.2902,
+      "step": 2106
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.249091443741126e-06,
+      "loss": 1.0154,
+      "step": 2107
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.246544037795504e-06,
+      "loss": 1.2274,
+      "step": 2108
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.243995901002312e-06,
+      "loss": 1.2115,
+      "step": 2109
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.241447034190513e-06,
+      "loss": 1.1876,
+      "step": 2110
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.238897438189307e-06,
+      "loss": 1.1985,
+      "step": 2111
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2363471138281295e-06,
+      "loss": 1.1919,
+      "step": 2112
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.233796061936655e-06,
+      "loss": 1.2525,
+      "step": 2113
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.231244283344793e-06,
+      "loss": 0.8645,
+      "step": 2114
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2286917788826926e-06,
+      "loss": 1.2238,
+      "step": 2115
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.226138549380734e-06,
+      "loss": 1.2636,
+      "step": 2116
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.223584595669536e-06,
+      "loss": 1.3029,
+      "step": 2117
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2210299185799544e-06,
+      "loss": 1.1782,
+      "step": 2118
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2184745189430785e-06,
+      "loss": 1.1637,
+      "step": 2119
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.215918397590232e-06,
+      "loss": 1.185,
+      "step": 2120
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.213361555352975e-06,
+      "loss": 0.8892,
+      "step": 2121
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.210803993063103e-06,
+      "loss": 1.1928,
+      "step": 2122
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2082457115526435e-06,
+      "loss": 1.2543,
+      "step": 2123
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.205686711653856e-06,
+      "loss": 1.2378,
+      "step": 2124
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.203126994199241e-06,
+      "loss": 1.1903,
+      "step": 2125
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.200566560021525e-06,
+      "loss": 1.2029,
+      "step": 2126
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.19800540995367e-06,
+      "loss": 0.9811,
+      "step": 2127
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.1954435448288736e-06,
+      "loss": 1.2259,
+      "step": 2128
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.192880965480565e-06,
+      "loss": 1.2648,
+      "step": 2129
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1903176727424e-06,
+      "loss": 1.2586,
+      "step": 2130
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.187753667448274e-06,
+      "loss": 1.1678,
+      "step": 2131
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.185188950432311e-06,
+      "loss": 1.2685,
+      "step": 2132
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.182623522528866e-06,
+      "loss": 1.2444,
+      "step": 2133
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.18005738457253e-06,
+      "loss": 0.9164,
+      "step": 2134
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.177490537398117e-06,
+      "loss": 1.225,
+      "step": 2135
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.174922981840676e-06,
+      "loss": 1.2387,
+      "step": 2136
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1723547187354884e-06,
+      "loss": 1.2411,
+      "step": 2137
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1697857489180625e-06,
+      "loss": 1.223,
+      "step": 2138
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.167216073224136e-06,
+      "loss": 1.2191,
+      "step": 2139
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.164645692489683e-06,
+      "loss": 1.2279,
+      "step": 2140
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.162074607550896e-06,
+      "loss": 1.2294,
+      "step": 2141
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.159502819244206e-06,
+      "loss": 1.1768,
+      "step": 2142
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.156930328406268e-06,
+      "loss": 1.0746,
+      "step": 2143
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.154357135873968e-06,
+      "loss": 1.1811,
+      "step": 2144
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.151783242484418e-06,
+      "loss": 1.2531,
+      "step": 2145
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.149208649074959e-06,
+      "loss": 1.2944,
+      "step": 2146
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.146633356483162e-06,
+      "loss": 1.1576,
+      "step": 2147
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.144057365546819e-06,
+      "loss": 1.2553,
+      "step": 2148
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.141480677103959e-06,
+      "loss": 1.1682,
+      "step": 2149
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.138903291992828e-06,
+      "loss": 1.084,
+      "step": 2150
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.136325211051905e-06,
+      "loss": 1.2564,
+      "step": 2151
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.133746435119894e-06,
+      "loss": 1.2924,
+      "step": 2152
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.131166965035723e-06,
+      "loss": 1.229,
+      "step": 2153
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.128586801638547e-06,
+      "loss": 1.1996,
+      "step": 2154
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.12600594576775e-06,
+      "loss": 1.224,
+      "step": 2155
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.123424398262935e-06,
+      "loss": 0.9786,
+      "step": 2156
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.120842159963934e-06,
+      "loss": 1.2491,
+      "step": 2157
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.118259231710802e-06,
+      "loss": 1.1937,
+      "step": 2158
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1156756143438236e-06,
+      "loss": 1.1956,
+      "step": 2159
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.113091308703498e-06,
+      "loss": 1.2478,
+      "step": 2160
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.110506315630557e-06,
+      "loss": 1.2855,
+      "step": 2161
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.10792063596595e-06,
+      "loss": 1.2301,
+      "step": 2162
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1053342705508564e-06,
+      "loss": 1.1794,
+      "step": 2163
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.10274722022667e-06,
+      "loss": 1.1916,
+      "step": 2164
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.100159485835017e-06,
+      "loss": 1.2036,
+      "step": 2165
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.097571068217736e-06,
+      "loss": 1.2179,
+      "step": 2166
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.0949819682168985e-06,
+      "loss": 1.2279,
+      "step": 2167
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.092392186674788e-06,
+      "loss": 1.223,
+      "step": 2168
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.089801724433918e-06,
+      "loss": 1.0846,
+      "step": 2169
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.087210582337018e-06,
+      "loss": 1.1783,
+      "step": 2170
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.084618761227041e-06,
+      "loss": 1.1897,
+      "step": 2171
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.0820262619471594e-06,
+      "loss": 1.2314,
+      "step": 2172
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.07943308534077e-06,
+      "loss": 1.2383,
+      "step": 2173
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.076839232251483e-06,
+      "loss": 1.3074,
+      "step": 2174
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.074244703523137e-06,
+      "loss": 1.2521,
+      "step": 2175
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.071649499999783e-06,
+      "loss": 1.2212,
+      "step": 2176
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.069053622525697e-06,
+      "loss": 1.2747,
+      "step": 2177
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.066457071945371e-06,
+      "loss": 1.2832,
+      "step": 2178
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.063859849103516e-06,
+      "loss": 1.0788,
+      "step": 2179
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.061261954845065e-06,
+      "loss": 1.2817,
+      "step": 2180
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.058663390015165e-06,
+      "loss": 1.2134,
+      "step": 2181
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.056064155459184e-06,
+      "loss": 1.2333,
+      "step": 2182
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.053464252022708e-06,
+      "loss": 1.2251,
+      "step": 2183
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.050863680551539e-06,
+      "loss": 1.188,
+      "step": 2184
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.0482624418916965e-06,
+      "loss": 0.9642,
+      "step": 2185
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.045660536889419e-06,
+      "loss": 1.1804,
+      "step": 2186
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.043057966391158e-06,
+      "loss": 1.201,
+      "step": 2187
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.040454731243586e-06,
+      "loss": 1.2396,
+      "step": 2188
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.037850832293591e-06,
+      "loss": 1.2357,
+      "step": 2189
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.035246270388272e-06,
+      "loss": 1.2746,
+      "step": 2190
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.0326410463749496e-06,
+      "loss": 1.2036,
+      "step": 2191
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.0300351611011565e-06,
+      "loss": 1.0027,
+      "step": 2192
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.027428615414642e-06,
+      "loss": 1.1462,
+      "step": 2193
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.0248214101633685e-06,
+      "loss": 1.26,
+      "step": 2194
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.022213546195516e-06,
+      "loss": 1.2055,
+      "step": 2195
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.019605024359475e-06,
+      "loss": 1.1854,
+      "step": 2196
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.016995845503853e-06,
+      "loss": 1.2003,
+      "step": 2197
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.014386010477473e-06,
+      "loss": 1.2249,
+      "step": 2198
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.011775520129363e-06,
+      "loss": 0.9367,
+      "step": 2199
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.009164375308775e-06,
+      "loss": 1.2424,
+      "step": 2200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.006552576865164e-06,
+      "loss": 1.1789,
+      "step": 2201
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.003940125648208e-06,
+      "loss": 1.2382,
+      "step": 2202
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.001327022507787e-06,
+      "loss": 1.1791,
+      "step": 2203
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.998713268294e-06,
+      "loss": 1.248,
+      "step": 2204
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.996098863857155e-06,
+      "loss": 0.8478,
+      "step": 2205
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.993483810047774e-06,
+      "loss": 1.1753,
+      "step": 2206
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9908681077165855e-06,
+      "loss": 1.2489,
+      "step": 2207
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.988251757714535e-06,
+      "loss": 1.1808,
+      "step": 2208
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.985634760892774e-06,
+      "loss": 1.2865,
+      "step": 2209
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.983017118102666e-06,
+      "loss": 1.2216,
+      "step": 2210
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.980398830195785e-06,
+      "loss": 1.1581,
+      "step": 2211
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.977779898023915e-06,
+      "loss": 1.2187,
+      "step": 2212
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.975160322439048e-06,
+      "loss": 1.1794,
+      "step": 2213
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.972540104293388e-06,
+      "loss": 0.9589,
+      "step": 2214
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.969919244439346e-06,
+      "loss": 1.2039,
+      "step": 2215
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.967297743729542e-06,
+      "loss": 1.1817,
+      "step": 2216
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9646756030168036e-06,
+      "loss": 1.1819,
+      "step": 2217
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9620528231541705e-06,
+      "loss": 1.2198,
+      "step": 2218
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.959429404994886e-06,
+      "loss": 1.2274,
+      "step": 2219
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.956805349392404e-06,
+      "loss": 1.2052,
+      "step": 2220
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9541806572003825e-06,
+      "loss": 1.0528,
+      "step": 2221
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.951555329272689e-06,
+      "loss": 1.2279,
+      "step": 2222
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.948929366463397e-06,
+      "loss": 1.2591,
+      "step": 2223
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.946302769626789e-06,
+      "loss": 1.2771,
+      "step": 2224
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.94367553961735e-06,
+      "loss": 1.1787,
+      "step": 2225
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.941047677289773e-06,
+      "loss": 1.2209,
+      "step": 2226
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.938419183498957e-06,
+      "loss": 1.2405,
+      "step": 2227
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9357900591000034e-06,
+      "loss": 0.953,
+      "step": 2228
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9331603049482245e-06,
+      "loss": 1.2395,
+      "step": 2229
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.930529921899132e-06,
+      "loss": 1.2303,
+      "step": 2230
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9278989108084445e-06,
+      "loss": 1.2119,
+      "step": 2231
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9252672725320835e-06,
+      "loss": 1.2167,
+      "step": 2232
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.92263500792618e-06,
+      "loss": 1.254,
+      "step": 2233
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.920002117847059e-06,
+      "loss": 1.0955,
+      "step": 2234
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9173686031512595e-06,
+      "loss": 1.2184,
+      "step": 2235
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.914734464695514e-06,
+      "loss": 1.1856,
+      "step": 2236
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.912099703336767e-06,
+      "loss": 1.2442,
+      "step": 2237
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9094643199321576e-06,
+      "loss": 1.1943,
+      "step": 2238
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.906828315339035e-06,
+      "loss": 1.1987,
+      "step": 2239
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.904191690414941e-06,
+      "loss": 1.21,
+      "step": 2240
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9015544460176296e-06,
+      "loss": 1.1586,
+      "step": 2241
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.898916583005047e-06,
+      "loss": 1.2456,
+      "step": 2242
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.89627810223535e-06,
+      "loss": 1.2349,
+      "step": 2243
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.893639004566888e-06,
+      "loss": 1.1889,
+      "step": 2244
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.890999290858213e-06,
+      "loss": 1.2432,
+      "step": 2245
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.888358961968082e-06,
+      "loss": 1.2509,
+      "step": 2246
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.885718018755448e-06,
+      "loss": 1.1738,
+      "step": 2247
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.883076462079466e-06,
+      "loss": 1.2368,
+      "step": 2248
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.880434292799486e-06,
+      "loss": 1.2269,
+      "step": 2249
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.877791511775064e-06,
+      "loss": 0.9987,
+      "step": 2250
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.87514811986595e-06,
+      "loss": 1.2325,
+      "step": 2251
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.872504117932095e-06,
+      "loss": 1.2238,
+      "step": 2252
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8698595068336485e-06,
+      "loss": 1.2575,
+      "step": 2253
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.867214287430957e-06,
+      "loss": 1.2698,
+      "step": 2254
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.864568460584564e-06,
+      "loss": 1.2432,
+      "step": 2255
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.861922027155214e-06,
+      "loss": 1.211,
+      "step": 2256
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.859274988003848e-06,
+      "loss": 1.0505,
+      "step": 2257
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8566273439916e-06,
+      "loss": 1.2287,
+      "step": 2258
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8539790959798045e-06,
+      "loss": 1.2585,
+      "step": 2259
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.851330244829993e-06,
+      "loss": 1.2144,
+      "step": 2260
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.848680791403892e-06,
+      "loss": 1.2331,
+      "step": 2261
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8460307365634225e-06,
+      "loss": 1.2164,
+      "step": 2262
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.843380081170704e-06,
+      "loss": 0.8996,
+      "step": 2263
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.840728826088048e-06,
+      "loss": 1.232,
+      "step": 2264
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.838076972177965e-06,
+      "loss": 1.2137,
+      "step": 2265
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.835424520303159e-06,
+      "loss": 1.2295,
+      "step": 2266
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.832771471326524e-06,
+      "loss": 1.2284,
+      "step": 2267
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.830117826111155e-06,
+      "loss": 1.2535,
+      "step": 2268
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.827463585520337e-06,
+      "loss": 1.1763,
+      "step": 2269
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.824808750417552e-06,
+      "loss": 1.1792,
+      "step": 2270
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.822153321666469e-06,
+      "loss": 1.2492,
+      "step": 2271
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8194973001309585e-06,
+      "loss": 1.213,
+      "step": 2272
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.816840686675076e-06,
+      "loss": 1.1887,
+      "step": 2273
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.814183482163077e-06,
+      "loss": 1.2073,
+      "step": 2274
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8115256874594015e-06,
+      "loss": 1.2411,
+      "step": 2275
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.80886730342869e-06,
+      "loss": 0.9427,
+      "step": 2276
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.806208330935766e-06,
+      "loss": 1.2417,
+      "step": 2277
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8035487708456504e-06,
+      "loss": 1.2282,
+      "step": 2278
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.800888624023552e-06,
+      "loss": 1.2625,
+      "step": 2279
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.798227891334875e-06,
+      "loss": 1.2502,
+      "step": 2280
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.795566573645209e-06,
+      "loss": 1.1747,
+      "step": 2281
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.792904671820337e-06,
+      "loss": 1.189,
+      "step": 2282
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.790242186726231e-06,
+      "loss": 1.232,
+      "step": 2283
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.7875791192290515e-06,
+      "loss": 1.2525,
+      "step": 2284
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.784915470195152e-06,
+      "loss": 1.2193,
+      "step": 2285
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.782251240491071e-06,
+      "loss": 1.1681,
+      "step": 2286
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.7795864309835395e-06,
+      "loss": 1.1708,
+      "step": 2287
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.776921042539474e-06,
+      "loss": 1.238,
+      "step": 2288
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.774255076025982e-06,
+      "loss": 1.1605,
+      "step": 2289
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.771588532310358e-06,
+      "loss": 1.1922,
+      "step": 2290
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.768921412260083e-06,
+      "loss": 1.1829,
+      "step": 2291
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.766253716742828e-06,
+      "loss": 1.067,
+      "step": 2292
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.763585446626449e-06,
+      "loss": 1.1962,
+      "step": 2293
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.760916602778992e-06,
+      "loss": 1.2186,
+      "step": 2294
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.758247186068684e-06,
+      "loss": 1.2132,
+      "step": 2295
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.755577197363945e-06,
+      "loss": 1.2446,
+      "step": 2296
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.752906637533376e-06,
+      "loss": 1.1761,
+      "step": 2297
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.750235507445766e-06,
+      "loss": 1.2163,
+      "step": 2298
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.74756380797009e-06,
+      "loss": 0.9901,
+      "step": 2299
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.744891539975507e-06,
+      "loss": 1.2031,
+      "step": 2300
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.742218704331362e-06,
+      "loss": 1.2646,
+      "step": 2301
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.739545301907184e-06,
+      "loss": 1.266,
+      "step": 2302
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.736871333572686e-06,
+      "loss": 1.2061,
+      "step": 2303
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.734196800197763e-06,
+      "loss": 1.2054,
+      "step": 2304
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.7315217026525025e-06,
+      "loss": 1.2185,
+      "step": 2305
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.728846041807164e-06,
+      "loss": 1.0848,
+      "step": 2306
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.7261698185322e-06,
+      "loss": 1.1677,
+      "step": 2307
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.723493033698238e-06,
+      "loss": 1.2108,
+      "step": 2308
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.720815688176093e-06,
+      "loss": 1.2531,
+      "step": 2309
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.718137782836762e-06,
+      "loss": 1.2046,
+      "step": 2310
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.715459318551424e-06,
+      "loss": 1.2344,
+      "step": 2311
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.712780296191437e-06,
+      "loss": 1.0994,
+      "step": 2312
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.710100716628345e-06,
+      "loss": 1.2011,
+      "step": 2313
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.7074205807338685e-06,
+      "loss": 1.3963,
+      "step": 2314
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.704739889379914e-06,
+      "loss": 0.9442,
+      "step": 2315
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.702058643438566e-06,
+      "loss": 1.1664,
+      "step": 2316
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.699376843782088e-06,
+      "loss": 1.2939,
+      "step": 2317
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.696694491282927e-06,
+      "loss": 1.2919,
+      "step": 2318
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6940115868137065e-06,
+      "loss": 1.2022,
+      "step": 2319
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.691328131247232e-06,
+      "loss": 1.1799,
+      "step": 2320
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.688644125456489e-06,
+      "loss": 1.0342,
+      "step": 2321
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.685959570314638e-06,
+      "loss": 1.2325,
+      "step": 2322
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.683274466695021e-06,
+      "loss": 1.1883,
+      "step": 2323
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.680588815471159e-06,
+      "loss": 1.2665,
+      "step": 2324
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.677902617516752e-06,
+      "loss": 1.2028,
+      "step": 2325
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.675215873705673e-06,
+      "loss": 1.2205,
+      "step": 2326
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.672528584911977e-06,
+      "loss": 1.1901,
+      "step": 2327
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.669840752009895e-06,
+      "loss": 0.974,
+      "step": 2328
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.667152375873836e-06,
+      "loss": 1.1795,
+      "step": 2329
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6644634573783825e-06,
+      "loss": 1.2076,
+      "step": 2330
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6617739973982985e-06,
+      "loss": 1.2159,
+      "step": 2331
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6590839968085195e-06,
+      "loss": 1.2788,
+      "step": 2332
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.656393456484159e-06,
+      "loss": 1.2366,
+      "step": 2333
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.653702377300507e-06,
+      "loss": 1.1992,
+      "step": 2334
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.651010760133024e-06,
+      "loss": 1.0886,
+      "step": 2335
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.648318605857355e-06,
+      "loss": 1.1844,
+      "step": 2336
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6456259153493075e-06,
+      "loss": 1.2494,
+      "step": 2337
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6429326894848745e-06,
+      "loss": 1.1658,
+      "step": 2338
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.640238929140215e-06,
+      "loss": 1.2121,
+      "step": 2339
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.63754463519167e-06,
+      "loss": 1.1784,
+      "step": 2340
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.634849808515745e-06,
+      "loss": 0.9986,
+      "step": 2341
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6321544499891245e-06,
+      "loss": 1.1963,
+      "step": 2342
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.629458560488664e-06,
+      "loss": 1.1834,
+      "step": 2343
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.626762140891396e-06,
+      "loss": 1.1877,
+      "step": 2344
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.624065192074519e-06,
+      "loss": 1.2625,
+      "step": 2345
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.621367714915407e-06,
+      "loss": 1.2046,
+      "step": 2346
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.618669710291607e-06,
+      "loss": 1.2466,
+      "step": 2347
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.615971179080832e-06,
+      "loss": 1.24,
+      "step": 2348
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.613272122160975e-06,
+      "loss": 1.3628,
+      "step": 2349
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.610572540410094e-06,
+      "loss": 1.2043,
+      "step": 2350
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.607872434706418e-06,
+      "loss": 1.0035,
+      "step": 2351
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.605171805928349e-06,
+      "loss": 1.1917,
+      "step": 2352
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.602470654954456e-06,
+      "loss": 1.1718,
+      "step": 2353
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.599768982663483e-06,
+      "loss": 1.2569,
+      "step": 2354
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.597066789934336e-06,
+      "loss": 1.2505,
+      "step": 2355
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.594364077646098e-06,
+      "loss": 1.2104,
+      "step": 2356
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.591660846678016e-06,
+      "loss": 1.1341,
+      "step": 2357
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.588957097909509e-06,
+      "loss": 1.1758,
+      "step": 2358
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.58625283222016e-06,
+      "loss": 1.1882,
+      "step": 2359
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.583548050489725e-06,
+      "loss": 1.1836,
+      "step": 2360
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.580842753598126e-06,
+      "loss": 1.1987,
+      "step": 2361
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.57813694242545e-06,
+      "loss": 1.2152,
+      "step": 2362
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.575430617851957e-06,
+      "loss": 1.1529,
+      "step": 2363
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.572723780758069e-06,
+      "loss": 1.1541,
+      "step": 2364
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.570016432024377e-06,
+      "loss": 1.1781,
+      "step": 2365
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.567308572531637e-06,
+      "loss": 1.1914,
+      "step": 2366
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5646002031607726e-06,
+      "loss": 1.1981,
+      "step": 2367
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.561891324792873e-06,
+      "loss": 1.1353,
+      "step": 2368
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5591819383091934e-06,
+      "loss": 1.2083,
+      "step": 2369
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5564720445911515e-06,
+      "loss": 1.0598,
+      "step": 2370
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.553761644520336e-06,
+      "loss": 1.1793,
+      "step": 2371
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.55105073897849e-06,
+      "loss": 1.2219,
+      "step": 2372
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.548339328847536e-06,
+      "loss": 1.2589,
+      "step": 2373
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.545627415009546e-06,
+      "loss": 1.2004,
+      "step": 2374
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5429149983467635e-06,
+      "loss": 1.2083,
+      "step": 2375
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.540202079741594e-06,
+      "loss": 1.1831,
+      "step": 2376
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5374886600766095e-06,
+      "loss": 0.8928,
+      "step": 2377
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.534774740234537e-06,
+      "loss": 1.2549,
+      "step": 2378
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5320603210982745e-06,
+      "loss": 1.2312,
+      "step": 2379
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.529345403550877e-06,
+      "loss": 1.2403,
+      "step": 2380
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.526629988475567e-06,
+      "loss": 1.1831,
+      "step": 2381
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.523914076755722e-06,
+      "loss": 1.2196,
+      "step": 2382
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.521197669274885e-06,
+      "loss": 0.9967,
+      "step": 2383
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5184807669167635e-06,
+      "loss": 1.3172,
+      "step": 2384
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.515763370565218e-06,
+      "loss": 1.2053,
+      "step": 2385
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.513045481104277e-06,
+      "loss": 1.1224,
+      "step": 2386
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.510327099418125e-06,
+      "loss": 1.1852,
+      "step": 2387
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.50760822639111e-06,
+      "loss": 1.2093,
+      "step": 2388
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.504888862907735e-06,
+      "loss": 1.1731,
+      "step": 2389
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.502169009852669e-06,
+      "loss": 1.1648,
+      "step": 2390
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.499448668110735e-06,
+      "loss": 1.2403,
+      "step": 2391
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.496727838566916e-06,
+      "loss": 1.1928,
+      "step": 2392
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.494006522106357e-06,
+      "loss": 1.1247,
+      "step": 2393
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.4912847196143565e-06,
+      "loss": 1.2256,
+      "step": 2394
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.488562431976376e-06,
+      "loss": 1.2234,
+      "step": 2395
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.4858396600780305e-06,
+      "loss": 1.2344,
+      "step": 2396
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.483116404805094e-06,
+      "loss": 1.2448,
+      "step": 2397
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.4803926670435e-06,
+      "loss": 1.1898,
+      "step": 2398
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.477668447679336e-06,
+      "loss": 1.0576,
+      "step": 2399
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.474943747598847e-06,
+      "loss": 1.211,
+      "step": 2400
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.472218567688435e-06,
+      "loss": 1.2136,
+      "step": 2401
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.46949290883466e-06,
+      "loss": 1.1997,
+      "step": 2402
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.466766771924231e-06,
+      "loss": 1.2112,
+      "step": 2403
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.464040157844022e-06,
+      "loss": 1.209,
+      "step": 2404
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.461313067481052e-06,
+      "loss": 1.1584,
+      "step": 2405
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.4585855017225065e-06,
+      "loss": 1.0645,
+      "step": 2406
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.455857461455713e-06,
+      "loss": 1.1813,
+      "step": 2407
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.453128947568166e-06,
+      "loss": 1.1905,
+      "step": 2408
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.450399960947502e-06,
+      "loss": 1.2303,
+      "step": 2409
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.447670502481525e-06,
+      "loss": 1.2693,
+      "step": 2410
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.444940573058175e-06,
+      "loss": 1.2431,
+      "step": 2411
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.442210173565562e-06,
+      "loss": 1.2191,
+      "step": 2412
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.43947930489194e-06,
+      "loss": 0.9247,
+      "step": 2413
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.436747967925716e-06,
+      "loss": 1.2213,
+      "step": 2414
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.434016163555452e-06,
+      "loss": 1.1748,
+      "step": 2415
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.431283892669862e-06,
+      "loss": 1.2216,
+      "step": 2416
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.428551156157807e-06,
+      "loss": 1.1718,
+      "step": 2417
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.425817954908307e-06,
+      "loss": 1.2048,
+      "step": 2418
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.423084289810528e-06,
+      "loss": 1.2305,
+      "step": 2419
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.420350161753788e-06,
+      "loss": 1.2164,
+      "step": 2420
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.417615571627555e-06,
+      "loss": 1.2129,
+      "step": 2421
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.414880520321449e-06,
+      "loss": 1.0684,
+      "step": 2422
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.412145008725239e-06,
+      "loss": 1.1609,
+      "step": 2423
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.409409037728844e-06,
+      "loss": 1.2918,
+      "step": 2424
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.406672608222333e-06,
+      "loss": 1.1132,
+      "step": 2425
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.403935721095923e-06,
+      "loss": 1.2305,
+      "step": 2426
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.401198377239979e-06,
+      "loss": 1.2687,
+      "step": 2427
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3984605775450185e-06,
+      "loss": 0.9258,
+      "step": 2428
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.395722322901703e-06,
+      "loss": 1.1703,
+      "step": 2429
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.392983614200844e-06,
+      "loss": 1.1768,
+      "step": 2430
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.390244452333401e-06,
+      "loss": 1.2128,
+      "step": 2431
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.387504838190479e-06,
+      "loss": 1.2581,
+      "step": 2432
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.384764772663334e-06,
+      "loss": 1.2276,
+      "step": 2433
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.382024256643365e-06,
+      "loss": 1.2336,
+      "step": 2434
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.379283291022118e-06,
+      "loss": 0.9971,
+      "step": 2435
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.376541876691288e-06,
+      "loss": 1.2267,
+      "step": 2436
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.373800014542714e-06,
+      "loss": 1.2034,
+      "step": 2437
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.371057705468379e-06,
+      "loss": 1.217,
+      "step": 2438
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.368314950360416e-06,
+      "loss": 1.2185,
+      "step": 2439
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.365571750111097e-06,
+      "loss": 1.2912,
+      "step": 2440
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.362828105612846e-06,
+      "loss": 1.2218,
+      "step": 2441
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3600840177582235e-06,
+      "loss": 1.0676,
+      "step": 2442
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.357339487439943e-06,
+      "loss": 1.161,
+      "step": 2443
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3545945155508524e-06,
+      "loss": 1.2341,
+      "step": 2444
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.351849102983954e-06,
+      "loss": 1.2252,
+      "step": 2445
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3491032506323826e-06,
+      "loss": 1.1756,
+      "step": 2446
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3463569593894236e-06,
+      "loss": 1.1929,
+      "step": 2447
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.343610230148502e-06,
+      "loss": 1.1539,
+      "step": 2448
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.340863063803187e-06,
+      "loss": 1.2087,
+      "step": 2449
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.338115461247189e-06,
+      "loss": 1.1864,
+      "step": 2450
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3353674233743585e-06,
+      "loss": 1.2112,
+      "step": 2451
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.332618951078692e-06,
+      "loss": 1.1977,
+      "step": 2452
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.329870045254322e-06,
+      "loss": 1.1661,
+      "step": 2453
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.327120706795527e-06,
+      "loss": 1.2449,
+      "step": 2454
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.324370936596723e-06,
+      "loss": 1.1963,
+      "step": 2455
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.321620735552468e-06,
+      "loss": 1.2814,
+      "step": 2456
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.318870104557459e-06,
+      "loss": 1.1592,
+      "step": 2457
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3161190445065355e-06,
+      "loss": 0.9985,
+      "step": 2458
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.313367556294673e-06,
+      "loss": 1.238,
+      "step": 2459
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.310615640816987e-06,
+      "loss": 1.2002,
+      "step": 2460
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.307863298968735e-06,
+      "loss": 1.2267,
+      "step": 2461
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3051105316453105e-06,
+      "loss": 1.2265,
+      "step": 2462
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.302357339742245e-06,
+      "loss": 1.19,
+      "step": 2463
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.29960372415521e-06,
+      "loss": 1.0814,
+      "step": 2464
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.296849685780016e-06,
+      "loss": 1.2183,
+      "step": 2465
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.294095225512604e-06,
+      "loss": 1.192,
+      "step": 2466
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.291340344249062e-06,
+      "loss": 1.2527,
+      "step": 2467
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.288585042885608e-06,
+      "loss": 1.2504,
+      "step": 2468
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.2858293223186e-06,
+      "loss": 1.234,
+      "step": 2469
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.283073183444531e-06,
+      "loss": 1.2324,
+      "step": 2470
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.280316627160029e-06,
+      "loss": 1.1759,
+      "step": 2471
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.277559654361862e-06,
+      "loss": 1.251,
+      "step": 2472
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.274802265946927e-06,
+      "loss": 1.2415,
+      "step": 2473
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.272044462812263e-06,
+      "loss": 1.2031,
+      "step": 2474
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.269286245855039e-06,
+      "loss": 1.2385,
+      "step": 2475
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.266527615972564e-06,
+      "loss": 1.2391,
+      "step": 2476
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.263768574062271e-06,
+      "loss": 1.0259,
+      "step": 2477
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.26100912102174e-06,
+      "loss": 1.2074,
+      "step": 2478
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.258249257748675e-06,
+      "loss": 1.2473,
+      "step": 2479
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.255488985140921e-06,
+      "loss": 1.1554,
+      "step": 2480
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2527283040964445e-06,
+      "loss": 1.2031,
+      "step": 2481
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.249967215513363e-06,
+      "loss": 1.1809,
+      "step": 2482
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.247205720289907e-06,
+      "loss": 1.218,
+      "step": 2483
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.244443819324454e-06,
+      "loss": 0.9133,
+      "step": 2484
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.241681513515504e-06,
+      "loss": 1.1841,
+      "step": 2485
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2389188037616955e-06,
+      "loss": 1.1671,
+      "step": 2486
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.236155690961795e-06,
+      "loss": 1.2861,
+      "step": 2487
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2333921760147e-06,
+      "loss": 1.1549,
+      "step": 2488
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.230628259819439e-06,
+      "loss": 1.2829,
+      "step": 2489
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.227863943275173e-06,
+      "loss": 1.2116,
+      "step": 2490
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.22509922728119e-06,
+      "loss": 1.1542,
+      "step": 2491
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.222334112736912e-06,
+      "loss": 1.1869,
+      "step": 2492
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.219568600541886e-06,
+      "loss": 0.9878,
+      "step": 2493
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.216802691595794e-06,
+      "loss": 1.2137,
+      "step": 2494
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.214036386798438e-06,
+      "loss": 1.2018,
+      "step": 2495
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2112696870497615e-06,
+      "loss": 1.1901,
+      "step": 2496
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.208502593249823e-06,
+      "loss": 1.1686,
+      "step": 2497
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.20573510629882e-06,
+      "loss": 1.2497,
+      "step": 2498
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.202967227097073e-06,
+      "loss": 1.2177,
+      "step": 2499
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2001989565450305e-06,
+      "loss": 1.0038,
+      "step": 2500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.197430295543267e-06,
+      "loss": 1.2516,
+      "step": 2501
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.194661244992488e-06,
+      "loss": 1.2002,
+      "step": 2502
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.191891805793523e-06,
+      "loss": 1.2321,
+      "step": 2503
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1891219788473265e-06,
+      "loss": 1.1783,
+      "step": 2504
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1863517650549835e-06,
+      "loss": 1.2065,
+      "step": 2505
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.183581165317698e-06,
+      "loss": 1.0182,
+      "step": 2506
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.180810180536809e-06,
+      "loss": 1.1744,
+      "step": 2507
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.178038811613771e-06,
+      "loss": 1.1987,
+      "step": 2508
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.175267059450173e-06,
+      "loss": 1.1916,
+      "step": 2509
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.172494924947718e-06,
+      "loss": 1.2208,
+      "step": 2510
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.169722409008244e-06,
+      "loss": 1.1343,
+      "step": 2511
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.166949512533704e-06,
+      "loss": 1.2187,
+      "step": 2512
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.164176236426183e-06,
+      "loss": 1.0193,
+      "step": 2513
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.161402581587881e-06,
+      "loss": 1.1822,
+      "step": 2514
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.158628548921132e-06,
+      "loss": 1.1801,
+      "step": 2515
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.15585413932838e-06,
+      "loss": 1.1563,
+      "step": 2516
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.153079353712201e-06,
+      "loss": 1.2268,
+      "step": 2517
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1503041929752926e-06,
+      "loss": 1.1984,
+      "step": 2518
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.147528658020468e-06,
+      "loss": 1.2602,
+      "step": 2519
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.144752749750671e-06,
+      "loss": 0.9661,
+      "step": 2520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.14197646906896e-06,
+      "loss": 1.1903,
+      "step": 2521
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.139199816878516e-06,
+      "loss": 1.2023,
+      "step": 2522
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.136422794082645e-06,
+      "loss": 1.251,
+      "step": 2523
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.133645401584767e-06,
+      "loss": 1.2151,
+      "step": 2524
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.130867640288427e-06,
+      "loss": 1.1785,
+      "step": 2525
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.128089511097289e-06,
+      "loss": 1.1919,
+      "step": 2526
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.125311014915138e-06,
+      "loss": 1.2125,
+      "step": 2527
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.122532152645871e-06,
+      "loss": 1.2434,
+      "step": 2528
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.119752925193516e-06,
+      "loss": 1.0412,
+      "step": 2529
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.1169733334622096e-06,
+      "loss": 1.1971,
+      "step": 2530
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.114193378356213e-06,
+      "loss": 1.1889,
+      "step": 2531
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.111413060779903e-06,
+      "loss": 1.1615,
+      "step": 2532
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.108632381637775e-06,
+      "loss": 1.2744,
+      "step": 2533
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.105851341834439e-06,
+      "loss": 1.1718,
+      "step": 2534
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.10306994227463e-06,
+      "loss": 0.9924,
+      "step": 2535
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.1002881838631925e-06,
+      "loss": 1.2189,
+      "step": 2536
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.09750606750509e-06,
+      "loss": 1.1731,
+      "step": 2537
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.094723594105404e-06,
+      "loss": 1.1835,
+      "step": 2538
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.091940764569331e-06,
+      "loss": 1.2428,
+      "step": 2539
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.089157579802183e-06,
+      "loss": 1.176,
+      "step": 2540
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.086374040709387e-06,
+      "loss": 1.1877,
+      "step": 2541
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0835901481964895e-06,
+      "loss": 0.9226,
+      "step": 2542
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.080805903169145e-06,
+      "loss": 1.2108,
+      "step": 2543
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.078021306533129e-06,
+      "loss": 1.1757,
+      "step": 2544
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0752363591943265e-06,
+      "loss": 1.2696,
+      "step": 2545
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.072451062058742e-06,
+      "loss": 1.2068,
+      "step": 2546
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0696654160324875e-06,
+      "loss": 1.2041,
+      "step": 2547
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.066879422021794e-06,
+      "loss": 1.2528,
+      "step": 2548
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.064093080933e-06,
+      "loss": 0.9338,
+      "step": 2549
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.061306393672566e-06,
+      "loss": 1.1594,
+      "step": 2550
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.058519361147055e-06,
+      "loss": 1.2264,
+      "step": 2551
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.055731984263147e-06,
+      "loss": 1.1997,
+      "step": 2552
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.052944263927635e-06,
+      "loss": 1.1888,
+      "step": 2553
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.050156201047421e-06,
+      "loss": 1.1707,
+      "step": 2554
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.047367796529523e-06,
+      "loss": 0.9779,
+      "step": 2555
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.044579051281063e-06,
+      "loss": 1.2736,
+      "step": 2556
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.041789966209281e-06,
+      "loss": 1.2579,
+      "step": 2557
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.039000542221525e-06,
+      "loss": 1.2106,
+      "step": 2558
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0362107802252486e-06,
+      "loss": 1.2003,
+      "step": 2559
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.033420681128024e-06,
+      "loss": 1.2156,
+      "step": 2560
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.030630245837526e-06,
+      "loss": 1.204,
+      "step": 2561
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.027839475261542e-06,
+      "loss": 1.248,
+      "step": 2562
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.025048370307967e-06,
+      "loss": 1.227,
+      "step": 2563
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.022256931884807e-06,
+      "loss": 1.1947,
+      "step": 2564
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.019465160900173e-06,
+      "loss": 1.1795,
+      "step": 2565
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.01667305826229e-06,
+      "loss": 1.1996,
+      "step": 2566
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.013880624879482e-06,
+      "loss": 1.1866,
+      "step": 2567
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.011087861660191e-06,
+      "loss": 1.2464,
+      "step": 2568
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.008294769512957e-06,
+      "loss": 1.2211,
+      "step": 2569
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.005501349346434e-06,
+      "loss": 1.1507,
+      "step": 2570
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.002707602069377e-06,
+      "loss": 1.0372,
+      "step": 2571
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.999913528590652e-06,
+      "loss": 1.1918,
+      "step": 2572
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.997119129819229e-06,
+      "loss": 1.151,
+      "step": 2573
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.994324406664184e-06,
+      "loss": 1.1722,
+      "step": 2574
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9915293600346995e-06,
+      "loss": 1.2156,
+      "step": 2575
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.988733990840059e-06,
+      "loss": 1.203,
+      "step": 2576
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.98593829998966e-06,
+      "loss": 1.2092,
+      "step": 2577
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.983142288392992e-06,
+      "loss": 1.0222,
+      "step": 2578
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.980345956959663e-06,
+      "loss": 1.2523,
+      "step": 2579
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.977549306599373e-06,
+      "loss": 1.1368,
+      "step": 2580
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.974752338221933e-06,
+      "loss": 1.1792,
+      "step": 2581
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.971955052737253e-06,
+      "loss": 1.1128,
+      "step": 2582
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9691574510553505e-06,
+      "loss": 1.2371,
+      "step": 2583
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9663595340863424e-06,
+      "loss": 0.9991,
+      "step": 2584
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9635613027404495e-06,
+      "loss": 1.1826,
+      "step": 2585
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.960762757927993e-06,
+      "loss": 1.2327,
+      "step": 2586
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.957963900559403e-06,
+      "loss": 1.2047,
+      "step": 2587
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9551647315452e-06,
+      "loss": 1.2449,
+      "step": 2588
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.952365251796015e-06,
+      "loss": 1.1997,
+      "step": 2589
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.949565462222577e-06,
+      "loss": 1.1575,
+      "step": 2590
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.946765363735714e-06,
+      "loss": 0.9354,
+      "step": 2591
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9439649572463595e-06,
+      "loss": 1.2082,
+      "step": 2592
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9411642436655425e-06,
+      "loss": 1.1633,
+      "step": 2593
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.938363223904393e-06,
+      "loss": 1.2677,
+      "step": 2594
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.935561898874142e-06,
+      "loss": 1.1741,
+      "step": 2595
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.932760269486117e-06,
+      "loss": 1.1346,
+      "step": 2596
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.929958336651751e-06,
+      "loss": 1.1783,
+      "step": 2597
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.927156101282565e-06,
+      "loss": 1.2229,
+      "step": 2598
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9243535642901885e-06,
+      "loss": 1.1957,
+      "step": 2599
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.921550726586343e-06,
+      "loss": 1.0103,
+      "step": 2600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.918747589082853e-06,
+      "loss": 1.222,
+      "step": 2601
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.915944152691634e-06,
+      "loss": 1.2212,
+      "step": 2602
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.913140418324705e-06,
+      "loss": 1.1482,
+      "step": 2603
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.910336386894177e-06,
+      "loss": 1.2245,
+      "step": 2604
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9075320593122605e-06,
+      "loss": 1.2567,
+      "step": 2605
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.904727436491262e-06,
+      "loss": 1.2635,
+      "step": 2606
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.901922519343586e-06,
+      "loss": 1.0034,
+      "step": 2607
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.899117308781726e-06,
+      "loss": 1.1962,
+      "step": 2608
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.896311805718278e-06,
+      "loss": 1.2167,
+      "step": 2609
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.893506011065932e-06,
+      "loss": 1.2455,
+      "step": 2610
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8906999257374665e-06,
+      "loss": 1.1542,
+      "step": 2611
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.887893550645766e-06,
+      "loss": 1.2198,
+      "step": 2612
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.885086886703797e-06,
+      "loss": 0.9958,
+      "step": 2613
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8822799348246305e-06,
+      "loss": 1.1997,
+      "step": 2614
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.879472695921421e-06,
+      "loss": 1.195,
+      "step": 2615
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.876665170907429e-06,
+      "loss": 1.1735,
+      "step": 2616
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.873857360695993e-06,
+      "loss": 1.1946,
+      "step": 2617
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.871049266200558e-06,
+      "loss": 1.1394,
+      "step": 2618
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8682408883346535e-06,
+      "loss": 1.1734,
+      "step": 2619
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.865432228011901e-06,
+      "loss": 0.9788,
+      "step": 2620
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.862623286146021e-06,
+      "loss": 1.1385,
+      "step": 2621
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.859814063650817e-06,
+      "loss": 1.2275,
+      "step": 2622
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.85700456144019e-06,
+      "loss": 1.1904,
+      "step": 2623
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8541947804281285e-06,
+      "loss": 1.2235,
+      "step": 2624
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.851384721528712e-06,
+      "loss": 1.2051,
+      "step": 2625
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8485743856561125e-06,
+      "loss": 1.2424,
+      "step": 2626
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.845763773724591e-06,
+      "loss": 0.9713,
+      "step": 2627
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.842952886648496e-06,
+      "loss": 1.1854,
+      "step": 2628
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.84014172534227e-06,
+      "loss": 1.2517,
+      "step": 2629
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.837330290720442e-06,
+      "loss": 1.2123,
+      "step": 2630
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.834518583697628e-06,
+      "loss": 1.1736,
+      "step": 2631
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.831706605188538e-06,
+      "loss": 1.2069,
+      "step": 2632
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.828894356107963e-06,
+      "loss": 1.2288,
+      "step": 2633
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.826081837370793e-06,
+      "loss": 1.2072,
+      "step": 2634
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8232690498919906e-06,
+      "loss": 1.2109,
+      "step": 2635
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.820455994586621e-06,
+      "loss": 0.9697,
+      "step": 2636
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.817642672369825e-06,
+      "loss": 1.1599,
+      "step": 2637
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.814829084156839e-06,
+      "loss": 1.2171,
+      "step": 2638
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.812015230862978e-06,
+      "loss": 1.2548,
+      "step": 2639
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8092011134036505e-06,
+      "loss": 1.1828,
+      "step": 2640
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.806386732694346e-06,
+      "loss": 1.2024,
+      "step": 2641
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8035720896506396e-06,
+      "loss": 1.2043,
+      "step": 2642
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.800757185188195e-06,
+      "loss": 1.2094,
+      "step": 2643
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.79794202022276e-06,
+      "loss": 1.2052,
+      "step": 2644
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.795126595670164e-06,
+      "loss": 1.2171,
+      "step": 2645
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7923109124463264e-06,
+      "loss": 1.1446,
+      "step": 2646
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.789494971467244e-06,
+      "loss": 1.1897,
+      "step": 2647
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7866787736490026e-06,
+      "loss": 1.1193,
+      "step": 2648
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.783862319907773e-06,
+      "loss": 0.9509,
+      "step": 2649
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7810456111598e-06,
+      "loss": 1.2385,
+      "step": 2650
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.778228648321424e-06,
+      "loss": 1.1386,
+      "step": 2651
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.775411432309059e-06,
+      "loss": 1.1939,
+      "step": 2652
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.772593964039203e-06,
+      "loss": 1.1914,
+      "step": 2653
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.769776244428439e-06,
+      "loss": 1.2015,
+      "step": 2654
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.766958274393428e-06,
+      "loss": 1.2315,
+      "step": 2655
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.764140054850915e-06,
+      "loss": 1.1721,
+      "step": 2656
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.761321586717726e-06,
+      "loss": 1.2516,
+      "step": 2657
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.758502870910768e-06,
+      "loss": 1.183,
+      "step": 2658
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.755683908347026e-06,
+      "loss": 1.135,
+      "step": 2659
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.75286469994357e-06,
+      "loss": 1.2573,
+      "step": 2660
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.750045246617544e-06,
+      "loss": 1.1168,
+      "step": 2661
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.747225549286177e-06,
+      "loss": 1.0549,
+      "step": 2662
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.744405608866775e-06,
+      "loss": 1.2276,
+      "step": 2663
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.741585426276724e-06,
+      "loss": 1.2398,
+      "step": 2664
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.738765002433488e-06,
+      "loss": 1.2319,
+      "step": 2665
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7359443382546055e-06,
+      "loss": 1.1703,
+      "step": 2666
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.733123434657704e-06,
+      "loss": 1.1573,
+      "step": 2667
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.730302292560476e-06,
+      "loss": 1.2208,
+      "step": 2668
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.727480912880703e-06,
+      "loss": 1.1812,
+      "step": 2669
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.724659296536234e-06,
+      "loss": 1.235,
+      "step": 2670
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.721837444445004e-06,
+      "loss": 1.282,
+      "step": 2671
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.719015357525016e-06,
+      "loss": 1.0212,
+      "step": 2672
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.716193036694359e-06,
+      "loss": 1.1388,
+      "step": 2673
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.713370482871187e-06,
+      "loss": 1.2489,
+      "step": 2674
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.71054769697374e-06,
+      "loss": 1.1926,
+      "step": 2675
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.707724679920325e-06,
+      "loss": 1.1856,
+      "step": 2676
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7049014326293306e-06,
+      "loss": 1.2392,
+      "step": 2677
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.702077956019218e-06,
+      "loss": 0.9412,
+      "step": 2678
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.699254251008524e-06,
+      "loss": 1.2219,
+      "step": 2679
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.696430318515855e-06,
+      "loss": 1.1246,
+      "step": 2680
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.693606159459899e-06,
+      "loss": 1.2169,
+      "step": 2681
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.690781774759412e-06,
+      "loss": 1.2162,
+      "step": 2682
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.687957165333225e-06,
+      "loss": 1.2239,
+      "step": 2683
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.685132332100243e-06,
+      "loss": 1.2445,
+      "step": 2684
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.68230727597944e-06,
+      "loss": 0.9811,
+      "step": 2685
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.679481997889873e-06,
+      "loss": 1.1948,
+      "step": 2686
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6766564987506564e-06,
+      "loss": 1.1611,
+      "step": 2687
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6738307794809865e-06,
+      "loss": 1.2674,
+      "step": 2688
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.671004841000129e-06,
+      "loss": 1.2298,
+      "step": 2689
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.668178684227421e-06,
+      "loss": 1.2083,
+      "step": 2690
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.66535231008227e-06,
+      "loss": 0.9425,
+      "step": 2691
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.662525719484152e-06,
+      "loss": 1.2003,
+      "step": 2692
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.659698913352618e-06,
+      "loss": 1.1775,
+      "step": 2693
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.656871892607287e-06,
+      "loss": 1.2035,
+      "step": 2694
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.654044658167848e-06,
+      "loss": 1.1959,
+      "step": 2695
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.651217210954058e-06,
+      "loss": 1.2293,
+      "step": 2696
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.648389551885745e-06,
+      "loss": 1.1288,
+      "step": 2697
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.645561681882805e-06,
+      "loss": 0.9878,
+      "step": 2698
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.642733601865202e-06,
+      "loss": 1.2181,
+      "step": 2699
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.639905312752975e-06,
+      "loss": 1.2909,
+      "step": 2700
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.637076815466217e-06,
+      "loss": 1.0015,
+      "step": 2701
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6342481109251036e-06,
+      "loss": 1.2672,
+      "step": 2702
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.631419200049867e-06,
+      "loss": 1.1813,
+      "step": 2703
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.628590083760815e-06,
+      "loss": 1.2028,
+      "step": 2704
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.625760762978313e-06,
+      "loss": 1.1729,
+      "step": 2705
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.622931238622804e-06,
+      "loss": 1.2175,
+      "step": 2706
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.620101511614785e-06,
+      "loss": 0.9684,
+      "step": 2707
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.61727158287483e-06,
+      "loss": 1.1678,
+      "step": 2708
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.614441453323571e-06,
+      "loss": 1.1937,
+      "step": 2709
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.611611123881711e-06,
+      "loss": 1.2157,
+      "step": 2710
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6087805954700105e-06,
+      "loss": 1.1908,
+      "step": 2711
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6059498690093025e-06,
+      "loss": 1.1919,
+      "step": 2712
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.603118945420482e-06,
+      "loss": 1.1685,
+      "step": 2713
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6002878256245044e-06,
+      "loss": 1.0103,
+      "step": 2714
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.597456510542395e-06,
+      "loss": 1.1781,
+      "step": 2715
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.594625001095237e-06,
+      "loss": 1.1732,
+      "step": 2716
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.591793298204182e-06,
+      "loss": 1.2562,
+      "step": 2717
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.588961402790439e-06,
+      "loss": 1.1951,
+      "step": 2718
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.586129315775286e-06,
+      "loss": 1.1709,
+      "step": 2719
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5832970380800565e-06,
+      "loss": 0.9639,
+      "step": 2720
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5804645706261515e-06,
+      "loss": 1.1551,
+      "step": 2721
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.577631914335032e-06,
+      "loss": 1.2159,
+      "step": 2722
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5747990701282185e-06,
+      "loss": 1.1294,
+      "step": 2723
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.571966038927294e-06,
+      "loss": 1.1669,
+      "step": 2724
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.569132821653905e-06,
+      "loss": 1.1837,
+      "step": 2725
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5662994192297535e-06,
+      "loss": 1.1973,
+      "step": 2726
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5634658325766066e-06,
+      "loss": 1.0736,
+      "step": 2727
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.560632062616287e-06,
+      "loss": 1.2002,
+      "step": 2728
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.55779811027068e-06,
+      "loss": 1.2085,
+      "step": 2729
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5549639764617295e-06,
+      "loss": 1.2327,
+      "step": 2730
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.552129662111438e-06,
+      "loss": 1.2078,
+      "step": 2731
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.549295168141867e-06,
+      "loss": 1.1416,
+      "step": 2732
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.546460495475138e-06,
+      "loss": 1.2093,
+      "step": 2733
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.543625645033425e-06,
+      "loss": 1.1591,
+      "step": 2734
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.540790617738968e-06,
+      "loss": 1.334,
+      "step": 2735
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.537955414514058e-06,
+      "loss": 1.0029,
+      "step": 2736
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.535120036281047e-06,
+      "loss": 1.2361,
+      "step": 2737
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.532284483962341e-06,
+      "loss": 1.1978,
+      "step": 2738
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.529448758480408e-06,
+      "loss": 1.1394,
+      "step": 2739
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.526612860757763e-06,
+      "loss": 1.1547,
+      "step": 2740
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5237767917169885e-06,
+      "loss": 1.1878,
+      "step": 2741
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5209405522807115e-06,
+      "loss": 1.2109,
+      "step": 2742
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.518104143371623e-06,
+      "loss": 0.99,
+      "step": 2743
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.515267565912466e-06,
+      "loss": 1.1963,
+      "step": 2744
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.512430820826035e-06,
+      "loss": 1.1856,
+      "step": 2745
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.509593909035186e-06,
+      "loss": 1.162,
+      "step": 2746
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.506756831462824e-06,
+      "loss": 1.2441,
+      "step": 2747
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.503919589031908e-06,
+      "loss": 1.1649,
+      "step": 2748
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.501082182665456e-06,
+      "loss": 1.1562,
+      "step": 2749
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.498244613286531e-06,
+      "loss": 1.2267,
+      "step": 2750
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.495406881818256e-06,
+      "loss": 1.2216,
+      "step": 2751
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.492568989183803e-06,
+      "loss": 1.202,
+      "step": 2752
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.489730936306399e-06,
+      "loss": 1.1556,
+      "step": 2753
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.486892724109319e-06,
+      "loss": 1.2489,
+      "step": 2754
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.484054353515896e-06,
+      "loss": 1.1973,
+      "step": 2755
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.48121582544951e-06,
+      "loss": 0.7655,
+      "step": 2756
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.478377140833591e-06,
+      "loss": 1.1577,
+      "step": 2757
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.475538300591624e-06,
+      "loss": 1.1971,
+      "step": 2758
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.472699305647143e-06,
+      "loss": 1.1818,
+      "step": 2759
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.469860156923732e-06,
+      "loss": 1.2052,
+      "step": 2760
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.467020855345026e-06,
+      "loss": 1.1938,
+      "step": 2761
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.464181401834707e-06,
+      "loss": 1.2418,
+      "step": 2762
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.46134179731651e-06,
+      "loss": 0.9099,
+      "step": 2763
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.458502042714218e-06,
+      "loss": 1.2008,
+      "step": 2764
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.45566213895166e-06,
+      "loss": 1.2614,
+      "step": 2765
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4528220869527185e-06,
+      "loss": 1.1693,
+      "step": 2766
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.449981887641318e-06,
+      "loss": 1.1669,
+      "step": 2767
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4471415419414396e-06,
+      "loss": 1.2184,
+      "step": 2768
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.444301050777101e-06,
+      "loss": 0.9924,
+      "step": 2769
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.441460415072379e-06,
+      "loss": 1.3032,
+      "step": 2770
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.438619635751387e-06,
+      "loss": 1.1916,
+      "step": 2771
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.435778713738292e-06,
+      "loss": 1.0647,
+      "step": 2772
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.432937649957304e-06,
+      "loss": 1.1622,
+      "step": 2773
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.430096445332682e-06,
+      "loss": 1.2142,
+      "step": 2774
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.427255100788726e-06,
+      "loss": 1.2215,
+      "step": 2775
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.424413617249788e-06,
+      "loss": 1.2538,
+      "step": 2776
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.421571995640259e-06,
+      "loss": 1.182,
+      "step": 2777
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.41873023688458e-06,
+      "loss": 1.2093,
+      "step": 2778
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4158883419072325e-06,
+      "loss": 1.1535,
+      "step": 2779
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.413046311632745e-06,
+      "loss": 1.2034,
+      "step": 2780
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.41020414698569e-06,
+      "loss": 1.1743,
+      "step": 2781
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4073618488906795e-06,
+      "loss": 1.1989,
+      "step": 2782
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.404519418272376e-06,
+      "loss": 1.1764,
+      "step": 2783
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.401676856055479e-06,
+      "loss": 1.1334,
+      "step": 2784
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3988341631647354e-06,
+      "loss": 0.9086,
+      "step": 2785
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.395991340524931e-06,
+      "loss": 1.227,
+      "step": 2786
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.393148389060893e-06,
+      "loss": 1.1483,
+      "step": 2787
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.390305309697499e-06,
+      "loss": 1.1863,
+      "step": 2788
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.387462103359655e-06,
+      "loss": 1.2118,
+      "step": 2789
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3846187709723195e-06,
+      "loss": 1.2022,
+      "step": 2790
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3817753134604865e-06,
+      "loss": 1.1766,
+      "step": 2791
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.378931731749193e-06,
+      "loss": 1.0266,
+      "step": 2792
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.376088026763513e-06,
+      "loss": 1.1826,
+      "step": 2793
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.373244199428564e-06,
+      "loss": 1.169,
+      "step": 2794
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.370400250669504e-06,
+      "loss": 1.2715,
+      "step": 2795
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.367556181411526e-06,
+      "loss": 1.1646,
+      "step": 2796
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.364711992579868e-06,
+      "loss": 1.1627,
+      "step": 2797
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3618676850998e-06,
+      "loss": 0.9262,
+      "step": 2798
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.359023259896638e-06,
+      "loss": 1.2475,
+      "step": 2799
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.356178717895732e-06,
+      "loss": 1.21,
+      "step": 2800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.35333406002247e-06,
+      "loss": 1.1905,
+      "step": 2801
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.350489287202276e-06,
+      "loss": 1.1454,
+      "step": 2802
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.34764440036062e-06,
+      "loss": 1.1613,
+      "step": 2803
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.344799400422998e-06,
+      "loss": 1.1731,
+      "step": 2804
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.34195428831495e-06,
+      "loss": 1.2141,
+      "step": 2805
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.339109064962047e-06,
+      "loss": 1.2144,
+      "step": 2806
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3362637312899045e-06,
+      "loss": 1.1459,
+      "step": 2807
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.333418288224164e-06,
+      "loss": 1.0345,
+      "step": 2808
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.330572736690511e-06,
+      "loss": 1.2111,
+      "step": 2809
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.32772707761466e-06,
+      "loss": 1.1402,
+      "step": 2810
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3248813119223665e-06,
+      "loss": 1.2134,
+      "step": 2811
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.322035440539413e-06,
+      "loss": 1.1898,
+      "step": 2812
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.319189464391625e-06,
+      "loss": 1.1828,
+      "step": 2813
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.316343384404855e-06,
+      "loss": 1.0183,
+      "step": 2814
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.313497201504992e-06,
+      "loss": 1.2168,
+      "step": 2815
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.310650916617961e-06,
+      "loss": 1.169,
+      "step": 2816
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3078045306697154e-06,
+      "loss": 1.1308,
+      "step": 2817
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.304958044586245e-06,
+      "loss": 1.1935,
+      "step": 2818
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3021114592935695e-06,
+      "loss": 1.1909,
+      "step": 2819
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.299264775717745e-06,
+      "loss": 1.2244,
+      "step": 2820
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2964179947848544e-06,
+      "loss": 0.941,
+      "step": 2821
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.293571117421017e-06,
+      "loss": 1.2086,
+      "step": 2822
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.290724144552379e-06,
+      "loss": 1.1898,
+      "step": 2823
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.287877077105124e-06,
+      "loss": 1.1877,
+      "step": 2824
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2850299160054565e-06,
+      "loss": 1.18,
+      "step": 2825
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.282182662179623e-06,
+      "loss": 1.2708,
+      "step": 2826
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2793353165538895e-06,
+      "loss": 1.019,
+      "step": 2827
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.27648788005456e-06,
+      "loss": 1.1994,
+      "step": 2828
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.273640353607965e-06,
+      "loss": 1.2131,
+      "step": 2829
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.270792738140462e-06,
+      "loss": 1.1454,
+      "step": 2830
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.267945034578442e-06,
+      "loss": 1.1889,
+      "step": 2831
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2650972438483205e-06,
+      "loss": 1.1468,
+      "step": 2832
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.262249366876544e-06,
+      "loss": 1.2193,
+      "step": 2833
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.259401404589588e-06,
+      "loss": 0.9977,
+      "step": 2834
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2565533579139484e-06,
+      "loss": 1.1383,
+      "step": 2835
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2537052277761615e-06,
+      "loss": 1.2054,
+      "step": 2836
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.250857015102776e-06,
+      "loss": 1.1567,
+      "step": 2837
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.248008720820381e-06,
+      "loss": 1.198,
+      "step": 2838
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.245160345855581e-06,
+      "loss": 1.2041,
+      "step": 2839
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.242311891135016e-06,
+      "loss": 1.2687,
+      "step": 2840
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.239463357585343e-06,
+      "loss": 1.152,
+      "step": 2841
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.236614746133256e-06,
+      "loss": 1.1401,
+      "step": 2842
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2337660577054605e-06,
+      "loss": 1.0277,
+      "step": 2843
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.230917293228699e-06,
+      "loss": 1.1204,
+      "step": 2844
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.228068453629731e-06,
+      "loss": 1.1581,
+      "step": 2845
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.225219539835346e-06,
+      "loss": 1.2345,
+      "step": 2846
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.222370552772353e-06,
+      "loss": 1.1998,
+      "step": 2847
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2195214933675866e-06,
+      "loss": 1.2369,
+      "step": 2848
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2166723625479065e-06,
+      "loss": 1.2474,
+      "step": 2849
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2138231612401936e-06,
+      "loss": 1.037,
+      "step": 2850
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2109738903713525e-06,
+      "loss": 1.1892,
+      "step": 2851
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.208124550868311e-06,
+      "loss": 1.1526,
+      "step": 2852
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.205275143658018e-06,
+      "loss": 1.1988,
+      "step": 2853
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.202425669667444e-06,
+      "loss": 1.1834,
+      "step": 2854
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.199576129823585e-06,
+      "loss": 1.2163,
+      "step": 2855
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.196726525053453e-06,
+      "loss": 0.9272,
+      "step": 2856
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.193876856284085e-06,
+      "loss": 1.2293,
+      "step": 2857
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.191027124442537e-06,
+      "loss": 1.1698,
+      "step": 2858
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.188177330455886e-06,
+      "loss": 1.1831,
+      "step": 2859
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.185327475251231e-06,
+      "loss": 1.1428,
+      "step": 2860
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.182477559755688e-06,
+      "loss": 1.184,
+      "step": 2861
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1796275848963926e-06,
+      "loss": 1.2306,
+      "step": 2862
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.176777551600504e-06,
+      "loss": 0.9808,
+      "step": 2863
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.173927460795195e-06,
+      "loss": 1.2677,
+      "step": 2864
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1710773134076605e-06,
+      "loss": 1.2318,
+      "step": 2865
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.168227110365113e-06,
+      "loss": 1.2304,
+      "step": 2866
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1653768525947815e-06,
+      "loss": 1.233,
+      "step": 2867
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.162526541023916e-06,
+      "loss": 1.1963,
+      "step": 2868
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.159676176579782e-06,
+      "loss": 1.1933,
+      "step": 2869
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.15682576018966e-06,
+      "loss": 0.9126,
+      "step": 2870
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.153975292780852e-06,
+      "loss": 1.1408,
+      "step": 2871
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.151124775280675e-06,
+      "loss": 1.2264,
+      "step": 2872
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.148274208616462e-06,
+      "loss": 1.136,
+      "step": 2873
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.145423593715558e-06,
+      "loss": 1.227,
+      "step": 2874
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.14257293150533e-06,
+      "loss": 1.2032,
+      "step": 2875
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.139722222913158e-06,
+      "loss": 1.1789,
+      "step": 2876
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.136871468866439e-06,
+      "loss": 1.2117,
+      "step": 2877
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.134020670292578e-06,
+      "loss": 1.1785,
+      "step": 2878
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1311698281190025e-06,
+      "loss": 1.001,
+      "step": 2879
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.12831894327315e-06,
+      "loss": 1.2056,
+      "step": 2880
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.125468016682471e-06,
+      "loss": 1.1328,
+      "step": 2881
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.122617049274433e-06,
+      "loss": 1.1893,
+      "step": 2882
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.119766041976516e-06,
+      "loss": 1.2561,
+      "step": 2883
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.11691499571621e-06,
+      "loss": 1.1968,
+      "step": 2884
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.114063911421021e-06,
+      "loss": 1.1656,
+      "step": 2885
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.111212790018466e-06,
+      "loss": 0.8851,
+      "step": 2886
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.108361632436073e-06,
+      "loss": 1.1436,
+      "step": 2887
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.105510439601384e-06,
+      "loss": 1.2196,
+      "step": 2888
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.102659212441953e-06,
+      "loss": 1.1552,
+      "step": 2889
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0998079518853404e-06,
+      "loss": 1.1852,
+      "step": 2890
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.096956658859122e-06,
+      "loss": 1.2077,
+      "step": 2891
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.094105334290884e-06,
+      "loss": 0.9668,
+      "step": 2892
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.09125397910822e-06,
+      "loss": 1.2145,
+      "step": 2893
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.088402594238736e-06,
+      "loss": 1.1394,
+      "step": 2894
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.085551180610046e-06,
+      "loss": 1.1835,
+      "step": 2895
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.082699739149777e-06,
+      "loss": 1.2355,
+      "step": 2896
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.079848270785558e-06,
+      "loss": 1.2282,
+      "step": 2897
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0769967764450345e-06,
+      "loss": 1.1586,
+      "step": 2898
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0741452570558555e-06,
+      "loss": 0.8238,
+      "step": 2899
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.071293713545681e-06,
+      "loss": 1.1863,
+      "step": 2900
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.068442146842175e-06,
+      "loss": 1.1161,
+      "step": 2901
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.065590557873014e-06,
+      "loss": 1.2378,
+      "step": 2902
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.062738947565879e-06,
+      "loss": 1.197,
+      "step": 2903
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.059887316848457e-06,
+      "loss": 1.2689,
+      "step": 2904
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0570356666484414e-06,
+      "loss": 0.8379,
+      "step": 2905
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.054183997893538e-06,
+      "loss": 1.203,
+      "step": 2906
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.05133231151145e-06,
+      "loss": 1.0859,
+      "step": 2907
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.048480608429893e-06,
+      "loss": 1.1806,
+      "step": 2908
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.045628889576583e-06,
+      "loss": 1.1538,
+      "step": 2909
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.042777155879246e-06,
+      "loss": 1.2119,
+      "step": 2910
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.039925408265608e-06,
+      "loss": 1.1953,
+      "step": 2911
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.037073647663405e-06,
+      "loss": 1.1221,
+      "step": 2912
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.03422187500037e-06,
+      "loss": 1.2291,
+      "step": 2913
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.031370091204248e-06,
+      "loss": 1.2026,
+      "step": 2914
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.028518297202781e-06,
+      "loss": 1.0151,
+      "step": 2915
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0256664939237186e-06,
+      "loss": 1.1586,
+      "step": 2916
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.02281468229481e-06,
+      "loss": 1.1595,
+      "step": 2917
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0199628632438115e-06,
+      "loss": 1.2687,
+      "step": 2918
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.017111037698477e-06,
+      "loss": 1.1503,
+      "step": 2919
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0142592065865644e-06,
+      "loss": 1.1703,
+      "step": 2920
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0114073708358354e-06,
+      "loss": 1.0707,
+      "step": 2921
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.008555531374052e-06,
+      "loss": 1.1561,
+      "step": 2922
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0057036891289755e-06,
+      "loss": 1.1492,
+      "step": 2923
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.00285184502837e-06,
+      "loss": 1.1954,
+      "step": 2924
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5e-06,
+      "loss": 1.2332,
+      "step": 2925
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9971481549716315e-06,
+      "loss": 1.1987,
+      "step": 2926
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.994296310871027e-06,
+      "loss": 1.2356,
+      "step": 2927
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.99144446862595e-06,
+      "loss": 1.0352,
+      "step": 2928
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.988592629164165e-06,
+      "loss": 1.2017,
+      "step": 2929
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.985740793413436e-06,
+      "loss": 1.1564,
+      "step": 2930
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9828889623015265e-06,
+      "loss": 1.2359,
+      "step": 2931
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.980037136756191e-06,
+      "loss": 1.1169,
+      "step": 2932
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.977185317705191e-06,
+      "loss": 1.1591,
+      "step": 2933
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.974333506076283e-06,
+      "loss": 1.2165,
+      "step": 2934
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.971481702797221e-06,
+      "loss": 1.1072,
+      "step": 2935
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.968629908795753e-06,
+      "loss": 1.1525,
+      "step": 2936
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.965778124999631e-06,
+      "loss": 1.1456,
+      "step": 2937
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.962926352336597e-06,
+      "loss": 1.165,
+      "step": 2938
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.960074591734394e-06,
+      "loss": 1.0998,
+      "step": 2939
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9572228441207556e-06,
+      "loss": 1.1816,
+      "step": 2940
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.954371110423417e-06,
+      "loss": 0.9019,
+      "step": 2941
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.951519391570108e-06,
+      "loss": 1.125,
+      "step": 2942
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.948667688488552e-06,
+      "loss": 1.1921,
+      "step": 2943
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.945816002106465e-06,
+      "loss": 1.0822,
+      "step": 2944
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9429643333515585e-06,
+      "loss": 1.1559,
+      "step": 2945
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.940112683151545e-06,
+      "loss": 1.2134,
+      "step": 2946
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.937261052434124e-06,
+      "loss": 1.2216,
+      "step": 2947
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.934409442126988e-06,
+      "loss": 1.1423,
+      "step": 2948
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.931557853157826e-06,
+      "loss": 1.2467,
+      "step": 2949
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.92870628645432e-06,
+      "loss": 1.0065,
+      "step": 2950
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.925854742944145e-06,
+      "loss": 1.0973,
+      "step": 2951
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.923003223554967e-06,
+      "loss": 1.1949,
+      "step": 2952
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.920151729214443e-06,
+      "loss": 1.1591,
+      "step": 2953
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.917300260850224e-06,
+      "loss": 1.1968,
+      "step": 2954
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9144488193899546e-06,
+      "loss": 1.1916,
+      "step": 2955
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.911597405761265e-06,
+      "loss": 1.1361,
+      "step": 2956
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9087460208917815e-06,
+      "loss": 1.0262,
+      "step": 2957
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.905894665709116e-06,
+      "loss": 1.1459,
+      "step": 2958
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.903043341140879e-06,
+      "loss": 1.1883,
+      "step": 2959
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.900192048114661e-06,
+      "loss": 1.1274,
+      "step": 2960
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8973407875580485e-06,
+      "loss": 1.1822,
+      "step": 2961
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.894489560398618e-06,
+      "loss": 1.1601,
+      "step": 2962
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.891638367563928e-06,
+      "loss": 0.9576,
+      "step": 2963
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.888787209981536e-06,
+      "loss": 1.2168,
+      "step": 2964
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.885936088578981e-06,
+      "loss": 1.1505,
+      "step": 2965
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.883085004283793e-06,
+      "loss": 1.2441,
+      "step": 2966
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.880233958023486e-06,
+      "loss": 1.1474,
+      "step": 2967
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.877382950725568e-06,
+      "loss": 1.2755,
+      "step": 2968
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.87453198331753e-06,
+      "loss": 1.1678,
+      "step": 2969
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.871681056726853e-06,
+      "loss": 1.0701,
+      "step": 2970
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.868830171881e-06,
+      "loss": 1.1584,
+      "step": 2971
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.865979329707423e-06,
+      "loss": 1.2069,
+      "step": 2972
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.863128531133562e-06,
+      "loss": 1.1651,
+      "step": 2973
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.860277777086843e-06,
+      "loss": 1.2432,
+      "step": 2974
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.857427068494671e-06,
+      "loss": 1.1919,
+      "step": 2975
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.854576406284443e-06,
+      "loss": 1.1846,
+      "step": 2976
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8517257913835405e-06,
+      "loss": 1.016,
+      "step": 2977
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.848875224719327e-06,
+      "loss": 1.2079,
+      "step": 2978
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.846024707219149e-06,
+      "loss": 1.1563,
+      "step": 2979
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.843174239810341e-06,
+      "loss": 1.2707,
+      "step": 2980
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.84032382342022e-06,
+      "loss": 1.197,
+      "step": 2981
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.837473458976086e-06,
+      "loss": 1.2286,
+      "step": 2982
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.83462314740522e-06,
+      "loss": 1.1586,
+      "step": 2983
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.83177288963489e-06,
+      "loss": 1.1655,
+      "step": 2984
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.82892268659234e-06,
+      "loss": 1.1851,
+      "step": 2985
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8260725392048065e-06,
+      "loss": 0.9567,
+      "step": 2986
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.823222448399498e-06,
+      "loss": 1.1788,
+      "step": 2987
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.820372415103608e-06,
+      "loss": 1.1343,
+      "step": 2988
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.817522440244312e-06,
+      "loss": 1.1756,
+      "step": 2989
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.814672524748771e-06,
+      "loss": 1.2024,
+      "step": 2990
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.811822669544115e-06,
+      "loss": 1.1708,
+      "step": 2991
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8089728755574646e-06,
+      "loss": 1.1873,
+      "step": 2992
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.806123143715916e-06,
+      "loss": 0.979,
+      "step": 2993
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8032734749465495e-06,
+      "loss": 1.1539,
+      "step": 2994
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.800423870176417e-06,
+      "loss": 1.2553,
+      "step": 2995
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.797574330332557e-06,
+      "loss": 1.2055,
+      "step": 2996
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.794724856341985e-06,
+      "loss": 1.1944,
+      "step": 2997
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.791875449131691e-06,
+      "loss": 1.2023,
+      "step": 2998
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.789026109628649e-06,
+      "loss": 1.1323,
+      "step": 2999
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.786176838759808e-06,
+      "loss": 1.1966,
+      "step": 3000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.783327637452097e-06,
+      "loss": 1.1126,
+      "step": 3001
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.780478506632415e-06,
+      "loss": 1.1916,
+      "step": 3002
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.777629447227649e-06,
+      "loss": 1.1703,
+      "step": 3003
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.774780460164656e-06,
+      "loss": 1.2101,
+      "step": 3004
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.771931546370271e-06,
+      "loss": 1.0875,
+      "step": 3005
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7690827067713035e-06,
+      "loss": 0.9059,
+      "step": 3006
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.76623394229454e-06,
+      "loss": 1.2055,
+      "step": 3007
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.763385253866746e-06,
+      "loss": 1.0946,
+      "step": 3008
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7605366424146575e-06,
+      "loss": 1.1993,
+      "step": 3009
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7576881088649865e-06,
+      "loss": 1.1256,
+      "step": 3010
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.75483965414442e-06,
+      "loss": 1.1521,
+      "step": 3011
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.751991279179621e-06,
+      "loss": 1.1656,
+      "step": 3012
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.749142984897226e-06,
+      "loss": 1.2259,
+      "step": 3013
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.746294772223842e-06,
+      "loss": 1.1294,
+      "step": 3014
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7434466420860515e-06,
+      "loss": 1.1599,
+      "step": 3015
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.740598595410414e-06,
+      "loss": 1.2029,
+      "step": 3016
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7377506331234565e-06,
+      "loss": 1.1269,
+      "step": 3017
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.734902756151681e-06,
+      "loss": 1.1989,
+      "step": 3018
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.73205496542156e-06,
+      "loss": 1.1697,
+      "step": 3019
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.729207261859538e-06,
+      "loss": 1.2004,
+      "step": 3020
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.726359646392037e-06,
+      "loss": 1.154,
+      "step": 3021
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7235121199454414e-06,
+      "loss": 0.9116,
+      "step": 3022
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.720664683446111e-06,
+      "loss": 1.2071,
+      "step": 3023
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.717817337820378e-06,
+      "loss": 1.1629,
+      "step": 3024
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.714970083994544e-06,
+      "loss": 1.2381,
+      "step": 3025
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7121229228948785e-06,
+      "loss": 1.111,
+      "step": 3026
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7092758554476215e-06,
+      "loss": 1.2986,
+      "step": 3027
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.706428882578986e-06,
+      "loss": 0.9209,
+      "step": 3028
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.703582005215146e-06,
+      "loss": 1.1407,
+      "step": 3029
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.700735224282256e-06,
+      "loss": 1.2228,
+      "step": 3030
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.697888540706432e-06,
+      "loss": 1.1709,
+      "step": 3031
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6950419554137584e-06,
+      "loss": 1.2197,
+      "step": 3032
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.692195469330286e-06,
+      "loss": 1.1659,
+      "step": 3033
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.689349083382041e-06,
+      "loss": 1.201,
+      "step": 3034
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.686502798495009e-06,
+      "loss": 0.8811,
+      "step": 3035
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.683656615595148e-06,
+      "loss": 1.2102,
+      "step": 3036
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.680810535608376e-06,
+      "loss": 1.1658,
+      "step": 3037
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6779645594605875e-06,
+      "loss": 1.2052,
+      "step": 3038
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.675118688077634e-06,
+      "loss": 1.1608,
+      "step": 3039
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.672272922385341e-06,
+      "loss": 1.1686,
+      "step": 3040
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.669427263309491e-06,
+      "loss": 1.143,
+      "step": 3041
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.666581711775836e-06,
+      "loss": 1.1432,
+      "step": 3042
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.663736268710097e-06,
+      "loss": 1.1808,
+      "step": 3043
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.660890935037954e-06,
+      "loss": 1.1402,
+      "step": 3044
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.658045711685053e-06,
+      "loss": 1.1874,
+      "step": 3045
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.655200599577003e-06,
+      "loss": 1.1057,
+      "step": 3046
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.652355599639381e-06,
+      "loss": 1.2121,
+      "step": 3047
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.649510712797725e-06,
+      "loss": 0.9927,
+      "step": 3048
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.646665939977533e-06,
+      "loss": 1.2119,
+      "step": 3049
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.643821282104269e-06,
+      "loss": 1.3546,
+      "step": 3050
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.640976740103363e-06,
+      "loss": 1.0564,
+      "step": 3051
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.638132314900201e-06,
+      "loss": 1.2501,
+      "step": 3052
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.635288007420135e-06,
+      "loss": 1.1842,
+      "step": 3053
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6324438185884755e-06,
+      "loss": 1.1915,
+      "step": 3054
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6295997493304965e-06,
+      "loss": 1.1008,
+      "step": 3055
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.626755800571437e-06,
+      "loss": 1.1862,
+      "step": 3056
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.623911973236489e-06,
+      "loss": 1.155,
+      "step": 3057
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.621068268250809e-06,
+      "loss": 1.1176,
+      "step": 3058
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6182246865395135e-06,
+      "loss": 1.1956,
+      "step": 3059
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.615381229027681e-06,
+      "loss": 1.2216,
+      "step": 3060
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6125378966403465e-06,
+      "loss": 1.1713,
+      "step": 3061
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.609694690302503e-06,
+      "loss": 1.117,
+      "step": 3062
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.606851610939108e-06,
+      "loss": 1.1743,
+      "step": 3063
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.604008659475071e-06,
+      "loss": 0.9871,
+      "step": 3064
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.601165836835266e-06,
+      "loss": 1.1663,
+      "step": 3065
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.598323143944522e-06,
+      "loss": 1.1658,
+      "step": 3066
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.595480581727627e-06,
+      "loss": 1.1942,
+      "step": 3067
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.592638151109322e-06,
+      "loss": 1.2002,
+      "step": 3068
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.589795853014313e-06,
+      "loss": 1.217,
+      "step": 3069
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5869536883672566e-06,
+      "loss": 0.9809,
+      "step": 3070
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.58411165809277e-06,
+      "loss": 1.171,
+      "step": 3071
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.581269763115421e-06,
+      "loss": 1.2052,
+      "step": 3072
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.578428004359742e-06,
+      "loss": 1.2482,
+      "step": 3073
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5755863827502135e-06,
+      "loss": 1.1768,
+      "step": 3074
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.572744899211275e-06,
+      "loss": 1.1844,
+      "step": 3075
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5699035546673206e-06,
+      "loss": 1.1542,
+      "step": 3076
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.567062350042697e-06,
+      "loss": 0.9423,
+      "step": 3077
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.564221286261709e-06,
+      "loss": 1.1581,
+      "step": 3078
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.561380364248616e-06,
+      "loss": 1.1709,
+      "step": 3079
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.558539584927624e-06,
+      "loss": 1.1167,
+      "step": 3080
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.555698949222899e-06,
+      "loss": 1.1295,
+      "step": 3081
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.552858458058562e-06,
+      "loss": 1.2005,
+      "step": 3082
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.550018112358683e-06,
+      "loss": 1.1936,
+      "step": 3083
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.547177913047284e-06,
+      "loss": 0.9601,
+      "step": 3084
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5443378610483426e-06,
+      "loss": 1.3406,
+      "step": 3085
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.541497957285783e-06,
+      "loss": 0.9761,
+      "step": 3086
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.53865820268349e-06,
+      "loss": 1.1812,
+      "step": 3087
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.535818598165294e-06,
+      "loss": 1.1212,
+      "step": 3088
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.532979144654976e-06,
+      "loss": 1.3066,
+      "step": 3089
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5301398430762675e-06,
+      "loss": 1.1204,
+      "step": 3090
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.527300694352858e-06,
+      "loss": 1.2133,
+      "step": 3091
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.524461699408377e-06,
+      "loss": 1.1514,
+      "step": 3092
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.521622859166411e-06,
+      "loss": 1.0003,
+      "step": 3093
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.518784174550491e-06,
+      "loss": 1.1732,
+      "step": 3094
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.515945646484105e-06,
+      "loss": 1.1897,
+      "step": 3095
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.513107275890682e-06,
+      "loss": 1.1394,
+      "step": 3096
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.510269063693602e-06,
+      "loss": 1.1768,
+      "step": 3097
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.507431010816199e-06,
+      "loss": 1.1464,
+      "step": 3098
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.504593118181745e-06,
+      "loss": 1.2155,
+      "step": 3099
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.50175538671347e-06,
+      "loss": 0.9613,
+      "step": 3100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.498917817334546e-06,
+      "loss": 1.1565,
+      "step": 3101
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.496080410968093e-06,
+      "loss": 1.2084,
+      "step": 3102
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.493243168537178e-06,
+      "loss": 1.1718,
+      "step": 3103
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.490406090964815e-06,
+      "loss": 1.2779,
+      "step": 3104
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4875691791739655e-06,
+      "loss": 1.1456,
+      "step": 3105
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.484732434087537e-06,
+      "loss": 0.9524,
+      "step": 3106
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.481895856628378e-06,
+      "loss": 1.1932,
+      "step": 3107
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.479059447719289e-06,
+      "loss": 1.1527,
+      "step": 3108
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.476223208283013e-06,
+      "loss": 1.2227,
+      "step": 3109
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.473387139242238e-06,
+      "loss": 1.1827,
+      "step": 3110
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.470551241519594e-06,
+      "loss": 1.191,
+      "step": 3111
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.467715516037659e-06,
+      "loss": 1.1629,
+      "step": 3112
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.464879963718954e-06,
+      "loss": 1.0069,
+      "step": 3113
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.462044585485944e-06,
+      "loss": 1.2073,
+      "step": 3114
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.459209382261034e-06,
+      "loss": 1.1822,
+      "step": 3115
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.456374354966576e-06,
+      "loss": 1.2115,
+      "step": 3116
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.453539504524864e-06,
+      "loss": 1.2342,
+      "step": 3117
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.450704831858135e-06,
+      "loss": 1.1728,
+      "step": 3118
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4478703378885635e-06,
+      "loss": 1.1271,
+      "step": 3119
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.445036023538273e-06,
+      "loss": 1.2967,
+      "step": 3120
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.442201889729321e-06,
+      "loss": 1.189,
+      "step": 3121
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.439367937383715e-06,
+      "loss": 0.8921,
+      "step": 3122
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.436534167423395e-06,
+      "loss": 1.186,
+      "step": 3123
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.433700580770249e-06,
+      "loss": 1.1066,
+      "step": 3124
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4308671783460955e-06,
+      "loss": 1.1722,
+      "step": 3125
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.428033961072707e-06,
+      "loss": 1.1948,
+      "step": 3126
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.425200929871783e-06,
+      "loss": 1.2004,
+      "step": 3127
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.422368085664971e-06,
+      "loss": 1.157,
+      "step": 3128
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4195354293738484e-06,
+      "loss": 0.939,
+      "step": 3129
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.416702961919945e-06,
+      "loss": 1.2763,
+      "step": 3130
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4138706842247155e-06,
+      "loss": 1.1661,
+      "step": 3131
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.411038597209562e-06,
+      "loss": 1.2553,
+      "step": 3132
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4082067017958205e-06,
+      "loss": 1.1444,
+      "step": 3133
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.405374998904764e-06,
+      "loss": 1.1766,
+      "step": 3134
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.402543489457607e-06,
+      "loss": 1.1288,
+      "step": 3135
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.399712174375496e-06,
+      "loss": 1.2169,
+      "step": 3136
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.396881054579521e-06,
+      "loss": 1.235,
+      "step": 3137
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.394050130990698e-06,
+      "loss": 1.1884,
+      "step": 3138
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.391219404529991e-06,
+      "loss": 1.2438,
+      "step": 3139
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.388388876118292e-06,
+      "loss": 1.1798,
+      "step": 3140
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3855585466764305e-06,
+      "loss": 1.2629,
+      "step": 3141
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3827284171251725e-06,
+      "loss": 0.9758,
+      "step": 3142
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.379898488385215e-06,
+      "loss": 1.2196,
+      "step": 3143
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.377068761377198e-06,
+      "loss": 1.1585,
+      "step": 3144
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.374239237021688e-06,
+      "loss": 1.1716,
+      "step": 3145
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.371409916239188e-06,
+      "loss": 1.2589,
+      "step": 3146
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.368580799950133e-06,
+      "loss": 1.2222,
+      "step": 3147
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.365751889074898e-06,
+      "loss": 0.9576,
+      "step": 3148
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3629231845337845e-06,
+      "loss": 1.1592,
+      "step": 3149
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.360094687247028e-06,
+      "loss": 1.3163,
+      "step": 3150
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.357266398134798e-06,
+      "loss": 1.1728,
+      "step": 3151
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.354438318117196e-06,
+      "loss": 1.2353,
+      "step": 3152
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.351610448114258e-06,
+      "loss": 1.1507,
+      "step": 3153
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.348782789045945e-06,
+      "loss": 1.1932,
+      "step": 3154
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.345955341832156e-06,
+      "loss": 1.2084,
+      "step": 3155
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.343128107392714e-06,
+      "loss": 1.2107,
+      "step": 3156
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3403010866473836e-06,
+      "loss": 1.1553,
+      "step": 3157
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.33747428051585e-06,
+      "loss": 1.0163,
+      "step": 3158
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.334647689917734e-06,
+      "loss": 1.1888,
+      "step": 3159
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.331821315772579e-06,
+      "loss": 1.1622,
+      "step": 3160
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3289951589998716e-06,
+      "loss": 1.1956,
+      "step": 3161
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.326169220519015e-06,
+      "loss": 1.1421,
+      "step": 3162
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.323343501249346e-06,
+      "loss": 1.1886,
+      "step": 3163
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.32051800211013e-06,
+      "loss": 1.1255,
+      "step": 3164
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.31769272402056e-06,
+      "loss": 1.1488,
+      "step": 3165
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.314867667899759e-06,
+      "loss": 1.2256,
+      "step": 3166
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.312042834666776e-06,
+      "loss": 1.2116,
+      "step": 3167
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.309218225240591e-06,
+      "loss": 1.1913,
+      "step": 3168
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.306393840540103e-06,
+      "loss": 1.1896,
+      "step": 3169
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.303569681484146e-06,
+      "loss": 1.1993,
+      "step": 3170
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.300745748991478e-06,
+      "loss": 1.0179,
+      "step": 3171
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.297922043980784e-06,
+      "loss": 1.1959,
+      "step": 3172
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.29509856737067e-06,
+      "loss": 1.2093,
+      "step": 3173
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.292275320079677e-06,
+      "loss": 1.0977,
+      "step": 3174
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.289452303026263e-06,
+      "loss": 1.1656,
+      "step": 3175
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.286629517128815e-06,
+      "loss": 1.1651,
+      "step": 3176
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.283806963305644e-06,
+      "loss": 0.9429,
+      "step": 3177
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.280984642474983e-06,
+      "loss": 1.2293,
+      "step": 3178
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.278162555554997e-06,
+      "loss": 1.1874,
+      "step": 3179
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.275340703463767e-06,
+      "loss": 1.1829,
+      "step": 3180
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2725190871193e-06,
+      "loss": 1.2313,
+      "step": 3181
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.269697707439525e-06,
+      "loss": 1.1609,
+      "step": 3182
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.266876565342298e-06,
+      "loss": 1.1641,
+      "step": 3183
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.264055661745396e-06,
+      "loss": 0.8665,
+      "step": 3184
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.261234997566516e-06,
+      "loss": 1.143,
+      "step": 3185
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.258414573723277e-06,
+      "loss": 1.2435,
+      "step": 3186
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.255594391133226e-06,
+      "loss": 1.2007,
+      "step": 3187
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.252774450713824e-06,
+      "loss": 1.2153,
+      "step": 3188
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.249954753382457e-06,
+      "loss": 1.1375,
+      "step": 3189
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.247135300056433e-06,
+      "loss": 1.1605,
+      "step": 3190
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.244316091652974e-06,
+      "loss": 1.2226,
+      "step": 3191
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.241497129089234e-06,
+      "loss": 1.1631,
+      "step": 3192
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.238678413282275e-06,
+      "loss": 0.9463,
+      "step": 3193
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.235859945149087e-06,
+      "loss": 1.1845,
+      "step": 3194
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.233041725606573e-06,
+      "loss": 1.1809,
+      "step": 3195
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.230223755571563e-06,
+      "loss": 1.2267,
+      "step": 3196
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.227406035960798e-06,
+      "loss": 1.1742,
+      "step": 3197
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.224588567690943e-06,
+      "loss": 1.2081,
+      "step": 3198
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2217713516785775e-06,
+      "loss": 1.1254,
+      "step": 3199
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2189543888402e-06,
+      "loss": 0.9286,
+      "step": 3200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.216137680092229e-06,
+      "loss": 1.1734,
+      "step": 3201
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.213321226350999e-06,
+      "loss": 1.1492,
+      "step": 3202
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.210505028532759e-06,
+      "loss": 1.1166,
+      "step": 3203
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.207689087553676e-06,
+      "loss": 1.1589,
+      "step": 3204
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.204873404329837e-06,
+      "loss": 1.1877,
+      "step": 3205
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.202057979777242e-06,
+      "loss": 1.2217,
+      "step": 3206
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.199242814811807e-06,
+      "loss": 0.8835,
+      "step": 3207
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.196427910349362e-06,
+      "loss": 1.1042,
+      "step": 3208
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.193613267305656e-06,
+      "loss": 1.2704,
+      "step": 3209
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.19079888659635e-06,
+      "loss": 1.101,
+      "step": 3210
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.187984769137023e-06,
+      "loss": 1.1931,
+      "step": 3211
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.185170915843163e-06,
+      "loss": 1.1877,
+      "step": 3212
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.182357327630175e-06,
+      "loss": 1.0712,
+      "step": 3213
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.17954400541338e-06,
+      "loss": 1.1611,
+      "step": 3214
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.17673095010801e-06,
+      "loss": 1.2015,
+      "step": 3215
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.17391816262921e-06,
+      "loss": 1.1735,
+      "step": 3216
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.171105643892037e-06,
+      "loss": 1.1839,
+      "step": 3217
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1682933948114634e-06,
+      "loss": 1.2097,
+      "step": 3218
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1654814163023735e-06,
+      "loss": 1.077,
+      "step": 3219
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.162669709279561e-06,
+      "loss": 0.952,
+      "step": 3220
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.159858274657733e-06,
+      "loss": 1.2088,
+      "step": 3221
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.157047113351504e-06,
+      "loss": 1.1447,
+      "step": 3222
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.154236226275411e-06,
+      "loss": 1.1238,
+      "step": 3223
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.151425614343888e-06,
+      "loss": 1.1916,
+      "step": 3224
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.14861527847129e-06,
+      "loss": 1.2151,
+      "step": 3225
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1458052195718715e-06,
+      "loss": 1.2012,
+      "step": 3226
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.142995438559811e-06,
+      "loss": 1.2839,
+      "step": 3227
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.140185936349184e-06,
+      "loss": 1.1276,
+      "step": 3228
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.137376713853981e-06,
+      "loss": 0.988,
+      "step": 3229
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.134567771988098e-06,
+      "loss": 1.153,
+      "step": 3230
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.131759111665349e-06,
+      "loss": 1.1978,
+      "step": 3231
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.128950733799444e-06,
+      "loss": 1.1766,
+      "step": 3232
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.126142639304009e-06,
+      "loss": 1.1773,
+      "step": 3233
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.123334829092575e-06,
+      "loss": 1.1826,
+      "step": 3234
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.120527304078579e-06,
+      "loss": 1.1527,
+      "step": 3235
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.117720065175371e-06,
+      "loss": 0.9824,
+      "step": 3236
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.114913113296205e-06,
+      "loss": 1.2338,
+      "step": 3237
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.112106449354236e-06,
+      "loss": 1.1461,
+      "step": 3238
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1093000742625335e-06,
+      "loss": 1.1952,
+      "step": 3239
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.106493988934069e-06,
+      "loss": 1.1353,
+      "step": 3240
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.103688194281723e-06,
+      "loss": 1.189,
+      "step": 3241
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.100882691218276e-06,
+      "loss": 1.1772,
+      "step": 3242
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.098077480656415e-06,
+      "loss": 1.2045,
+      "step": 3243
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.095272563508737e-06,
+      "loss": 1.15,
+      "step": 3244
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.09246794068774e-06,
+      "loss": 1.2165,
+      "step": 3245
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.089663613105826e-06,
+      "loss": 1.1251,
+      "step": 3246
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0868595816752975e-06,
+      "loss": 1.1813,
+      "step": 3247
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.084055847308367e-06,
+      "loss": 1.1493,
+      "step": 3248
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.081252410917148e-06,
+      "loss": 0.8481,
+      "step": 3249
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.078449273413658e-06,
+      "loss": 1.1735,
+      "step": 3250
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.075646435709813e-06,
+      "loss": 1.1747,
+      "step": 3251
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.072843898717436e-06,
+      "loss": 1.2047,
+      "step": 3252
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.07004166334825e-06,
+      "loss": 1.1197,
+      "step": 3253
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0672397305138836e-06,
+      "loss": 1.1667,
+      "step": 3254
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.064438101125859e-06,
+      "loss": 0.9581,
+      "step": 3255
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0616367760956095e-06,
+      "loss": 1.1367,
+      "step": 3256
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.058835756334457e-06,
+      "loss": 1.2084,
+      "step": 3257
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.056035042753641e-06,
+      "loss": 1.1494,
+      "step": 3258
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.053234636264286e-06,
+      "loss": 1.2341,
+      "step": 3259
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.050434537777426e-06,
+      "loss": 1.24,
+      "step": 3260
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.047634748203986e-06,
+      "loss": 1.1652,
+      "step": 3261
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.044835268454802e-06,
+      "loss": 1.244,
+      "step": 3262
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0420360994406e-06,
+      "loss": 1.1789,
+      "step": 3263
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.039237242072008e-06,
+      "loss": 1.1956,
+      "step": 3264
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.036438697259551e-06,
+      "loss": 0.8427,
+      "step": 3265
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.033640465913658e-06,
+      "loss": 1.2057,
+      "step": 3266
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.03084254894465e-06,
+      "loss": 1.1923,
+      "step": 3267
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.028044947262749e-06,
+      "loss": 1.1782,
+      "step": 3268
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.025247661778069e-06,
+      "loss": 1.2157,
+      "step": 3269
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.022450693400628e-06,
+      "loss": 1.1926,
+      "step": 3270
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.019654043040338e-06,
+      "loss": 1.046,
+      "step": 3271
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.016857711607009e-06,
+      "loss": 1.1898,
+      "step": 3272
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.014061700010343e-06,
+      "loss": 1.2351,
+      "step": 3273
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.011266009159942e-06,
+      "loss": 1.108,
+      "step": 3274
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.008470639965303e-06,
+      "loss": 1.1715,
+      "step": 3275
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.005675593335818e-06,
+      "loss": 1.1993,
+      "step": 3276
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0028808701807734e-06,
+      "loss": 1.1715,
+      "step": 3277
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0000864714093505e-06,
+      "loss": 0.8649,
+      "step": 3278
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.997292397930624e-06,
+      "loss": 1.1964,
+      "step": 3279
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.994498650653568e-06,
+      "loss": 1.1931,
+      "step": 3280
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.991705230487045e-06,
+      "loss": 1.1296,
+      "step": 3281
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.988912138339812e-06,
+      "loss": 1.1779,
+      "step": 3282
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.986119375120519e-06,
+      "loss": 1.1311,
+      "step": 3283
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.983326941737712e-06,
+      "loss": 0.9627,
+      "step": 3284
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.980534839099829e-06,
+      "loss": 1.2311,
+      "step": 3285
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.977743068115196e-06,
+      "loss": 1.1116,
+      "step": 3286
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.974951629692034e-06,
+      "loss": 1.1523,
+      "step": 3287
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.97216052473846e-06,
+      "loss": 1.2057,
+      "step": 3288
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.969369754162477e-06,
+      "loss": 1.1428,
+      "step": 3289
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.966579318871978e-06,
+      "loss": 1.274,
+      "step": 3290
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.963789219774753e-06,
+      "loss": 0.9848,
+      "step": 3291
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.960999457778476e-06,
+      "loss": 1.1101,
+      "step": 3292
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.95821003379072e-06,
+      "loss": 1.1417,
+      "step": 3293
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9554209487189375e-06,
+      "loss": 1.2279,
+      "step": 3294
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.95263220347048e-06,
+      "loss": 1.2064,
+      "step": 3295
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.949843798952578e-06,
+      "loss": 1.1583,
+      "step": 3296
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9470557360723665e-06,
+      "loss": 1.1078,
+      "step": 3297
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9442680157368545e-06,
+      "loss": 1.1751,
+      "step": 3298
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.941480638852948e-06,
+      "loss": 1.2057,
+      "step": 3299
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.938693606327434e-06,
+      "loss": 1.0679,
+      "step": 3300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9359069190669995e-06,
+      "loss": 1.1524,
+      "step": 3301
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.933120577978207e-06,
+      "loss": 1.1312,
+      "step": 3302
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.930334583967514e-06,
+      "loss": 1.1787,
+      "step": 3303
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.92754893794126e-06,
+      "loss": 1.1609,
+      "step": 3304
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9247636408056735e-06,
+      "loss": 1.2299,
+      "step": 3305
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.921978693466872e-06,
+      "loss": 1.0715,
+      "step": 3306
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.919194096830856e-06,
+      "loss": 0.8462,
+      "step": 3307
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.916409851803513e-06,
+      "loss": 1.1561,
+      "step": 3308
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.913625959290613e-06,
+      "loss": 1.203,
+      "step": 3309
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.910842420197818e-06,
+      "loss": 1.2175,
+      "step": 3310
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.908059235430671e-06,
+      "loss": 1.2098,
+      "step": 3311
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.905276405894598e-06,
+      "loss": 1.1434,
+      "step": 3312
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.902493932494912e-06,
+      "loss": 0.8802,
+      "step": 3313
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.899711816136808e-06,
+      "loss": 1.1906,
+      "step": 3314
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.896930057725372e-06,
+      "loss": 1.1184,
+      "step": 3315
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.894148658165562e-06,
+      "loss": 1.1976,
+      "step": 3316
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8913676183622275e-06,
+      "loss": 1.2428,
+      "step": 3317
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.888586939220098e-06,
+      "loss": 1.1686,
+      "step": 3318
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.885806621643788e-06,
+      "loss": 1.138,
+      "step": 3319
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.883026666537792e-06,
+      "loss": 1.0997,
+      "step": 3320
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8802470748064855e-06,
+      "loss": 1.1767,
+      "step": 3321
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.877467847354129e-06,
+      "loss": 1.2425,
+      "step": 3322
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.874688985084863e-06,
+      "loss": 1.1809,
+      "step": 3323
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.871910488902712e-06,
+      "loss": 1.1806,
+      "step": 3324
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.869132359711574e-06,
+      "loss": 1.1641,
+      "step": 3325
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.866354598415236e-06,
+      "loss": 1.0857,
+      "step": 3326
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.863577205917356e-06,
+      "loss": 0.9578,
+      "step": 3327
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.860800183121485e-06,
+      "loss": 1.1823,
+      "step": 3328
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8580235309310414e-06,
+      "loss": 1.22,
+      "step": 3329
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.855247250249331e-06,
+      "loss": 1.2073,
+      "step": 3330
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.852471341979531e-06,
+      "loss": 1.192,
+      "step": 3331
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.849695807024709e-06,
+      "loss": 1.137,
+      "step": 3332
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8469206462878e-06,
+      "loss": 1.192,
+      "step": 3333
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.844145860671623e-06,
+      "loss": 1.2138,
+      "step": 3334
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8413714510788715e-06,
+      "loss": 1.1631,
+      "step": 3335
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8385974184121185e-06,
+      "loss": 1.1321,
+      "step": 3336
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.835823763573819e-06,
+      "loss": 1.1378,
+      "step": 3337
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.833050487466298e-06,
+      "loss": 1.1202,
+      "step": 3338
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8302775909917585e-06,
+      "loss": 1.209,
+      "step": 3339
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.827505075052282e-06,
+      "loss": 1.1706,
+      "step": 3340
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.824732940549828e-06,
+      "loss": 1.233,
+      "step": 3341
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8219611883862295e-06,
+      "loss": 1.2058,
+      "step": 3342
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8191898194631926e-06,
+      "loss": 0.9257,
+      "step": 3343
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.816418834682302e-06,
+      "loss": 1.2198,
+      "step": 3344
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8136482349450178e-06,
+      "loss": 1.2046,
+      "step": 3345
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8108780211526747e-06,
+      "loss": 1.1751,
+      "step": 3346
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8081081942064786e-06,
+      "loss": 1.1859,
+      "step": 3347
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8053387550075126e-06,
+      "loss": 1.2277,
+      "step": 3348
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.802569704456733e-06,
+      "loss": 0.9775,
+      "step": 3349
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7998010434549716e-06,
+      "loss": 1.2248,
+      "step": 3350
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7970327729029288e-06,
+      "loss": 1.1674,
+      "step": 3351
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7942648937011818e-06,
+      "loss": 1.2301,
+      "step": 3352
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7914974067501774e-06,
+      "loss": 1.1843,
+      "step": 3353
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7887303129502406e-06,
+      "loss": 1.2034,
+      "step": 3354
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7859636132015632e-06,
+      "loss": 1.1442,
+      "step": 3355
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.783197308404209e-06,
+      "loss": 0.9607,
+      "step": 3356
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7804313994581143e-06,
+      "loss": 1.2089,
+      "step": 3357
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7776658872630883e-06,
+      "loss": 1.0603,
+      "step": 3358
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7749007727188107e-06,
+      "loss": 1.1809,
+      "step": 3359
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7721360567248283e-06,
+      "loss": 1.1871,
+      "step": 3360
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.769371740180563e-06,
+      "loss": 1.1838,
+      "step": 3361
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7666078239853002e-06,
+      "loss": 0.8537,
+      "step": 3362
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7638443090382067e-06,
+      "loss": 1.1551,
+      "step": 3363
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7610811962383054e-06,
+      "loss": 1.1695,
+      "step": 3364
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7583184864844978e-06,
+      "loss": 1.1688,
+      "step": 3365
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.755556180675547e-06,
+      "loss": 1.2056,
+      "step": 3366
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.752794279710094e-06,
+      "loss": 1.1691,
+      "step": 3367
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.750032784486639e-06,
+      "loss": 1.1029,
+      "step": 3368
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.747271695903556e-06,
+      "loss": 1.1828,
+      "step": 3369
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7445110148590822e-06,
+      "loss": 1.0986,
+      "step": 3370
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7417507422513254e-06,
+      "loss": 1.1817,
+      "step": 3371
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7389908789782606e-06,
+      "loss": 0.9212,
+      "step": 3372
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.73623142593773e-06,
+      "loss": 1.1461,
+      "step": 3373
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.733472384027439e-06,
+      "loss": 1.1148,
+      "step": 3374
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.730713754144961e-06,
+      "loss": 1.1491,
+      "step": 3375
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7279555371877373e-06,
+      "loss": 1.1976,
+      "step": 3376
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.725197734053074e-06,
+      "loss": 1.1788,
+      "step": 3377
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.722440345638141e-06,
+      "loss": 0.9017,
+      "step": 3378
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7196833728399716e-06,
+      "loss": 1.1575,
+      "step": 3379
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.71692681655547e-06,
+      "loss": 1.1888,
+      "step": 3380
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7141706776814014e-06,
+      "loss": 1.1533,
+      "step": 3381
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.711414957114393e-06,
+      "loss": 1.2172,
+      "step": 3382
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7086596557509396e-06,
+      "loss": 1.1346,
+      "step": 3383
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.705904774487396e-06,
+      "loss": 1.2229,
+      "step": 3384
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7031503142199867e-06,
+      "loss": 0.9285,
+      "step": 3385
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7003962758447908e-06,
+      "loss": 1.1319,
+      "step": 3386
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6976426602577565e-06,
+      "loss": 1.1881,
+      "step": 3387
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.69488946835469e-06,
+      "loss": 1.1643,
+      "step": 3388
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.692136701031266e-06,
+      "loss": 1.1713,
+      "step": 3389
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6893843591830136e-06,
+      "loss": 1.1109,
+      "step": 3390
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6866324437053286e-06,
+      "loss": 0.8027,
+      "step": 3391
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6838809554934666e-06,
+      "loss": 1.1405,
+      "step": 3392
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.68112989544254e-06,
+      "loss": 1.1843,
+      "step": 3393
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6783792644475336e-06,
+      "loss": 1.1143,
+      "step": 3394
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.675629063403278e-06,
+      "loss": 1.1374,
+      "step": 3395
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6728792932044756e-06,
+      "loss": 1.1785,
+      "step": 3396
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.670129954745678e-06,
+      "loss": 1.252,
+      "step": 3397
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6673810489213102e-06,
+      "loss": 0.9168,
+      "step": 3398
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6646325766256423e-06,
+      "loss": 1.1432,
+      "step": 3399
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6618845387528133e-06,
+      "loss": 1.1914,
+      "step": 3400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6591369361968127e-06,
+      "loss": 1.1533,
+      "step": 3401
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6563897698514984e-06,
+      "loss": 1.1883,
+      "step": 3402
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6536430406105772e-06,
+      "loss": 1.254,
+      "step": 3403
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6508967493676195e-06,
+      "loss": 1.1482,
+      "step": 3404
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6481508970160485e-06,
+      "loss": 1.1642,
+      "step": 3405
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6454054844491475e-06,
+      "loss": 1.07,
+      "step": 3406
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.642660512560059e-06,
+      "loss": 0.8208,
+      "step": 3407
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6399159822417786e-06,
+      "loss": 1.1906,
+      "step": 3408
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.637171894387157e-06,
+      "loss": 1.0871,
+      "step": 3409
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6344282498889037e-06,
+      "loss": 1.206,
+      "step": 3410
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6316850496395863e-06,
+      "loss": 1.1249,
+      "step": 3411
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6289422945316233e-06,
+      "loss": 1.1697,
+      "step": 3412
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.626199985457288e-06,
+      "loss": 1.2835,
+      "step": 3413
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.623458123308713e-06,
+      "loss": 0.9547,
+      "step": 3414
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6207167089778826e-06,
+      "loss": 1.1613,
+      "step": 3415
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.617975743356637e-06,
+      "loss": 1.1862,
+      "step": 3416
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6152352273366673e-06,
+      "loss": 1.1445,
+      "step": 3417
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6124951618095224e-06,
+      "loss": 1.1521,
+      "step": 3418
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6097555476666e-06,
+      "loss": 1.0688,
+      "step": 3419
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.607016385799158e-06,
+      "loss": 0.8567,
+      "step": 3420
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.604277677098299e-06,
+      "loss": 1.1758,
+      "step": 3421
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.601539422454983e-06,
+      "loss": 1.22,
+      "step": 3422
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.598801622760021e-06,
+      "loss": 1.196,
+      "step": 3423
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.596064278904079e-06,
+      "loss": 1.1482,
+      "step": 3424
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.593327391777668e-06,
+      "loss": 1.2042,
+      "step": 3425
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5905909622711564e-06,
+      "loss": 1.1866,
+      "step": 3426
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.587854991274763e-06,
+      "loss": 0.9056,
+      "step": 3427
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.585119479678553e-06,
+      "loss": 1.2738,
+      "step": 3428
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5823844283724464e-06,
+      "loss": 1.109,
+      "step": 3429
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.579649838246214e-06,
+      "loss": 1.157,
+      "step": 3430
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5769157101894745e-06,
+      "loss": 1.1596,
+      "step": 3431
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.574182045091693e-06,
+      "loss": 1.1776,
+      "step": 3432
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5714488438421935e-06,
+      "loss": 1.1059,
+      "step": 3433
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5687161073301403e-06,
+      "loss": 0.8484,
+      "step": 3434
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5659838364445505e-06,
+      "loss": 1.1436,
+      "step": 3435
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5632520320742847e-06,
+      "loss": 1.2033,
+      "step": 3436
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5605206951080618e-06,
+      "loss": 1.1862,
+      "step": 3437
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.557789826434439e-06,
+      "loss": 1.1072,
+      "step": 3438
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.555059426941827e-06,
+      "loss": 1.197,
+      "step": 3439
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5523294975184786e-06,
+      "loss": 1.1795,
+      "step": 3440
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5496000390524975e-06,
+      "loss": 1.2148,
+      "step": 3441
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.546871052431835e-06,
+      "loss": 1.0979,
+      "step": 3442
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5441425385442886e-06,
+      "loss": 1.1639,
+      "step": 3443
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.541414498277497e-06,
+      "loss": 1.1867,
+      "step": 3444
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.538686932518949e-06,
+      "loss": 1.1413,
+      "step": 3445
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5359598421559806e-06,
+      "loss": 1.201,
+      "step": 3446
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5332332280757706e-06,
+      "loss": 1.1141,
+      "step": 3447
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.530507091165343e-06,
+      "loss": 1.1551,
+      "step": 3448
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5277814323115657e-06,
+      "loss": 1.1298,
+      "step": 3449
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.525056252401153e-06,
+      "loss": 1.0022,
+      "step": 3450
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5223315523206657e-06,
+      "loss": 1.196,
+      "step": 3451
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.519607332956502e-06,
+      "loss": 1.0841,
+      "step": 3452
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.516883595194907e-06,
+      "loss": 1.1561,
+      "step": 3453
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.51416033992197e-06,
+      "loss": 1.1812,
+      "step": 3454
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5114375680236253e-06,
+      "loss": 1.1917,
+      "step": 3455
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.508715280385644e-06,
+      "loss": 0.8982,
+      "step": 3456
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.505993477893644e-06,
+      "loss": 1.1754,
+      "step": 3457
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5032721614330834e-06,
+      "loss": 1.0937,
+      "step": 3458
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5005513318892666e-06,
+      "loss": 1.2384,
+      "step": 3459
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.497830990147332e-06,
+      "loss": 1.0774,
+      "step": 3460
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.495111137092265e-06,
+      "loss": 1.1321,
+      "step": 3461
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.492391773608893e-06,
+      "loss": 1.1387,
+      "step": 3462
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.489672900581876e-06,
+      "loss": 0.9594,
+      "step": 3463
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4869545188957244e-06,
+      "loss": 1.2105,
+      "step": 3464
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.484236629434783e-06,
+      "loss": 1.1182,
+      "step": 3465
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4815192330832394e-06,
+      "loss": 1.1418,
+      "step": 3466
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4788023307251147e-06,
+      "loss": 1.1589,
+      "step": 3467
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.47608592324428e-06,
+      "loss": 1.167,
+      "step": 3468
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.473370011524435e-06,
+      "loss": 0.9052,
+      "step": 3469
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4706545964491245e-06,
+      "loss": 1.1939,
+      "step": 3470
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4679396789017263e-06,
+      "loss": 1.3476,
+      "step": 3471
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4652252597654646e-06,
+      "loss": 1.0178,
+      "step": 3472
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.462511339923392e-06,
+      "loss": 1.0853,
+      "step": 3473
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4597979202584077e-06,
+      "loss": 1.2555,
+      "step": 3474
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4570850016532386e-06,
+      "loss": 1.1798,
+      "step": 3475
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.454372584990455e-06,
+      "loss": 1.1774,
+      "step": 3476
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.451660671152465e-06,
+      "loss": 1.1899,
+      "step": 3477
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4489492610215104e-06,
+      "loss": 1.1165,
+      "step": 3478
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4462383554796676e-06,
+      "loss": 0.8408,
+      "step": 3479
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4435279554088484e-06,
+      "loss": 1.2308,
+      "step": 3480
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4408180616908082e-06,
+      "loss": 1.0746,
+      "step": 3481
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.438108675207129e-06,
+      "loss": 1.1764,
+      "step": 3482
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4353997968392295e-06,
+      "loss": 1.1428,
+      "step": 3483
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4326914274683653e-06,
+      "loss": 1.1674,
+      "step": 3484
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4299835679756244e-06,
+      "loss": 0.9031,
+      "step": 3485
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.427276219241933e-06,
+      "loss": 1.1552,
+      "step": 3486
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4245693821480443e-06,
+      "loss": 1.2021,
+      "step": 3487
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4218630575745504e-06,
+      "loss": 1.1168,
+      "step": 3488
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.419157246401875e-06,
+      "loss": 1.2553,
+      "step": 3489
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.416451949510276e-06,
+      "loss": 1.1655,
+      "step": 3490
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4137471677798413e-06,
+      "loss": 1.1478,
+      "step": 3491
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4110429020904924e-06,
+      "loss": 0.8884,
+      "step": 3492
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4083391533219836e-06,
+      "loss": 1.2045,
+      "step": 3493
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4056359223539028e-06,
+      "loss": 1.1595,
+      "step": 3494
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.402933210065665e-06,
+      "loss": 1.1341,
+      "step": 3495
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.400231017336519e-06,
+      "loss": 1.1668,
+      "step": 3496
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.397529345045546e-06,
+      "loss": 1.1718,
+      "step": 3497
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3948281940716527e-06,
+      "loss": 0.8113,
+      "step": 3498
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3921275652935838e-06,
+      "loss": 1.1741,
+      "step": 3499
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.389427459589908e-06,
+      "loss": 1.1814,
+      "step": 3500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.386727877839027e-06,
+      "loss": 1.1933,
+      "step": 3501
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3840288209191685e-06,
+      "loss": 1.1497,
+      "step": 3502
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3813302897083955e-06,
+      "loss": 1.1839,
+      "step": 3503
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3786322850845943e-06,
+      "loss": 1.1796,
+      "step": 3504
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.375934807925483e-06,
+      "loss": 1.085,
+      "step": 3505
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3732378591086057e-06,
+      "loss": 1.2856,
+      "step": 3506
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3705414395113354e-06,
+      "loss": 1.164,
+      "step": 3507
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3678455500108764e-06,
+      "loss": 0.9485,
+      "step": 3508
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.365150191484258e-06,
+      "loss": 1.1873,
+      "step": 3509
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3624553648083323e-06,
+      "loss": 1.1837,
+      "step": 3510
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.359761070859785e-06,
+      "loss": 1.1544,
+      "step": 3511
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3570673105151267e-06,
+      "loss": 1.1918,
+      "step": 3512
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3543740846506946e-06,
+      "loss": 1.1511,
+      "step": 3513
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.351681394142648e-06,
+      "loss": 0.9216,
+      "step": 3514
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.348989239866976e-06,
+      "loss": 1.1609,
+      "step": 3515
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3462976226994947e-06,
+      "loss": 1.1982,
+      "step": 3516
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.343606543515843e-06,
+      "loss": 1.1678,
+      "step": 3517
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.340916003191482e-06,
+      "loss": 1.148,
+      "step": 3518
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3382260026017027e-06,
+      "loss": 1.1748,
+      "step": 3519
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.335536542621617e-06,
+      "loss": 1.2541,
+      "step": 3520
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.332847624126165e-06,
+      "loss": 0.9066,
+      "step": 3521
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3301592479901057e-06,
+      "loss": 1.0986,
+      "step": 3522
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3274714150880237e-06,
+      "loss": 1.198,
+      "step": 3523
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3247841262943276e-06,
+      "loss": 1.1674,
+      "step": 3524
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3220973824832496e-06,
+      "loss": 1.1978,
+      "step": 3525
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.319411184528842e-06,
+      "loss": 1.1572,
+      "step": 3526
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3167255333049807e-06,
+      "loss": 1.094,
+      "step": 3527
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3140404296853635e-06,
+      "loss": 1.1574,
+      "step": 3528
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3113558745435135e-06,
+      "loss": 1.1939,
+      "step": 3529
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.30867186875277e-06,
+      "loss": 1.1712,
+      "step": 3530
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.305988413186295e-06,
+      "loss": 1.0917,
+      "step": 3531
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3033055087170763e-06,
+      "loss": 1.206,
+      "step": 3532
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3006231562179136e-06,
+      "loss": 1.1293,
+      "step": 3533
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.297941356561436e-06,
+      "loss": 0.8587,
+      "step": 3534
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2952601106200865e-06,
+      "loss": 1.1723,
+      "step": 3535
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.292579419266133e-06,
+      "loss": 1.1365,
+      "step": 3536
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.289899283371657e-06,
+      "loss": 1.2434,
+      "step": 3537
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.287219703808564e-06,
+      "loss": 1.1575,
+      "step": 3538
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2845406814485774e-06,
+      "loss": 1.2225,
+      "step": 3539
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2818622171632397e-06,
+      "loss": 1.1686,
+      "step": 3540
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2791843118239088e-06,
+      "loss": 1.2016,
+      "step": 3541
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2765069663017635e-06,
+      "loss": 1.095,
+      "step": 3542
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2738301814678015e-06,
+      "loss": 0.9237,
+      "step": 3543
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2711539581928374e-06,
+      "loss": 1.1948,
+      "step": 3544
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.268478297347499e-06,
+      "loss": 1.1383,
+      "step": 3545
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2658031998022368e-06,
+      "loss": 1.1509,
+      "step": 3546
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.263128666427316e-06,
+      "loss": 1.128,
+      "step": 3547
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.260454698092819e-06,
+      "loss": 1.1531,
+      "step": 3548
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2577812956686393e-06,
+      "loss": 1.1516,
+      "step": 3549
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.255108460024493e-06,
+      "loss": 0.8922,
+      "step": 3550
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.252436192029911e-06,
+      "loss": 1.2462,
+      "step": 3551
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2497644925542354e-06,
+      "loss": 1.1217,
+      "step": 3552
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.247093362466626e-06,
+      "loss": 1.1984,
+      "step": 3553
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.244422802636057e-06,
+      "loss": 1.1411,
+      "step": 3554
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.241752813931316e-06,
+      "loss": 1.1805,
+      "step": 3555
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2390833972210102e-06,
+      "loss": 1.1521,
+      "step": 3556
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.236414553373552e-06,
+      "loss": 0.8883,
+      "step": 3557
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.233746283257173e-06,
+      "loss": 1.1415,
+      "step": 3558
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2310785877399174e-06,
+      "loss": 1.1503,
+      "step": 3559
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2284114676896437e-06,
+      "loss": 1.1845,
+      "step": 3560
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2257449239740196e-06,
+      "loss": 1.1804,
+      "step": 3561
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2230789574605275e-06,
+      "loss": 1.1653,
+      "step": 3562
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2204135690164635e-06,
+      "loss": 0.8375,
+      "step": 3563
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.217748759508931e-06,
+      "loss": 1.1902,
+      "step": 3564
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2150845298048495e-06,
+      "loss": 1.1435,
+      "step": 3565
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2124208807709494e-06,
+      "loss": 1.2047,
+      "step": 3566
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2097578132737716e-06,
+      "loss": 1.136,
+      "step": 3567
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2070953281796635e-06,
+      "loss": 1.1351,
+      "step": 3568
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2044334263547917e-06,
+      "loss": 1.0882,
+      "step": 3569
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2017721086651255e-06,
+      "loss": 0.9541,
+      "step": 3570
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1991113759764493e-06,
+      "loss": 1.1801,
+      "step": 3571
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1964512291543517e-06,
+      "loss": 1.1222,
+      "step": 3572
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1937916690642356e-06,
+      "loss": 1.1416,
+      "step": 3573
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1911326965713117e-06,
+      "loss": 1.0869,
+      "step": 3574
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1884743125405993e-06,
+      "loss": 1.1752,
+      "step": 3575
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.185816517836925e-06,
+      "loss": 1.1915,
+      "step": 3576
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.183159313324924e-06,
+      "loss": 1.153,
+      "step": 3577
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1805026998690423e-06,
+      "loss": 1.1938,
+      "step": 3578
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1778466783335328e-06,
+      "loss": 0.8926,
+      "step": 3579
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1751912495824512e-06,
+      "loss": 1.0995,
+      "step": 3580
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.172536414479663e-06,
+      "loss": 1.2326,
+      "step": 3581
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1698821738888463e-06,
+      "loss": 1.2368,
+      "step": 3582
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.167228528673478e-06,
+      "loss": 1.1651,
+      "step": 3583
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1645754796968437e-06,
+      "loss": 1.1438,
+      "step": 3584
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.161923027822035e-06,
+      "loss": 1.1554,
+      "step": 3585
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.159271173911952e-06,
+      "loss": 0.9209,
+      "step": 3586
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.156619918829298e-06,
+      "loss": 1.1535,
+      "step": 3587
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1539692634365788e-06,
+      "loss": 1.1052,
+      "step": 3588
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1513192085961096e-06,
+      "loss": 1.1626,
+      "step": 3589
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.148669755170007e-06,
+      "loss": 1.1066,
+      "step": 3590
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1460209040201967e-06,
+      "loss": 1.2245,
+      "step": 3591
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1433726560084023e-06,
+      "loss": 0.9244,
+      "step": 3592
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.140725011996154e-06,
+      "loss": 1.1693,
+      "step": 3593
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1380779728447864e-06,
+      "loss": 1.1738,
+      "step": 3594
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1354315394154377e-06,
+      "loss": 1.0917,
+      "step": 3595
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1327857125690454e-06,
+      "loss": 1.2133,
+      "step": 3596
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1301404931663536e-06,
+      "loss": 1.1541,
+      "step": 3597
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1274958820679073e-06,
+      "loss": 1.2141,
+      "step": 3598
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1248518801340517e-06,
+      "loss": 0.883,
+      "step": 3599
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1222084882249375e-06,
+      "loss": 1.1112,
+      "step": 3600
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1195657072005155e-06,
+      "loss": 1.2321,
+      "step": 3601
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.116923537920537e-06,
+      "loss": 1.1311,
+      "step": 3602
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.114281981244553e-06,
+      "loss": 1.1909,
+      "step": 3603
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.111641038031919e-06,
+      "loss": 1.1727,
+      "step": 3604
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1090007091417884e-06,
+      "loss": 1.0091,
+      "step": 3605
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.106360995433115e-06,
+      "loss": 1.1669,
+      "step": 3606
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1037218977646515e-06,
+      "loss": 1.2028,
+      "step": 3607
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.101083416994953e-06,
+      "loss": 1.1438,
+      "step": 3608
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.098445553982372e-06,
+      "loss": 1.1126,
+      "step": 3609
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.095808309585061e-06,
+      "loss": 1.098,
+      "step": 3610
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0931716846609683e-06,
+      "loss": 1.1693,
+      "step": 3611
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.090535680067843e-06,
+      "loss": 1.2268,
+      "step": 3612
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0879002966632343e-06,
+      "loss": 1.1481,
+      "step": 3613
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.085265535304487e-06,
+      "loss": 1.2116,
+      "step": 3614
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.082631396848743e-06,
+      "loss": 0.8684,
+      "step": 3615
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.079997882152941e-06,
+      "loss": 1.1276,
+      "step": 3616
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0773649920738215e-06,
+      "loss": 1.1497,
+      "step": 3617
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0747327274679173e-06,
+      "loss": 1.1341,
+      "step": 3618
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.072101089191558e-06,
+      "loss": 1.1414,
+      "step": 3619
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0694700781008703e-06,
+      "loss": 1.1611,
+      "step": 3620
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0668396950517763e-06,
+      "loss": 0.9952,
+      "step": 3621
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0642099408999982e-06,
+      "loss": 1.2029,
+      "step": 3622
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0615808165010453e-06,
+      "loss": 1.2345,
+      "step": 3623
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.058952322710228e-06,
+      "loss": 1.1615,
+      "step": 3624
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.05632446038265e-06,
+      "loss": 1.1346,
+      "step": 3625
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.053697230373212e-06,
+      "loss": 1.1401,
+      "step": 3626
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0510706335366034e-06,
+      "loss": 1.1841,
+      "step": 3627
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.048444670727313e-06,
+      "loss": 1.0883,
+      "step": 3628
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0458193427996184e-06,
+      "loss": 1.1206,
+      "step": 3629
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.043194650607598e-06,
+      "loss": 1.1705,
+      "step": 3630
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0405705950051147e-06,
+      "loss": 1.2288,
+      "step": 3631
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0379471768458304e-06,
+      "loss": 1.1727,
+      "step": 3632
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.035324396983198e-06,
+      "loss": 1.157,
+      "step": 3633
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.03270225627046e-06,
+      "loss": 0.7983,
+      "step": 3634
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.030080755560656e-06,
+      "loss": 1.1681,
+      "step": 3635
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0274598957066132e-06,
+      "loss": 1.1523,
+      "step": 3636
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.024839677560954e-06,
+      "loss": 1.1523,
+      "step": 3637
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0222201019760865e-06,
+      "loss": 1.0873,
+      "step": 3638
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.019601169804216e-06,
+      "loss": 1.1977,
+      "step": 3639
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.016982881897336e-06,
+      "loss": 1.1441,
+      "step": 3640
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0143652391072285e-06,
+      "loss": 0.8868,
+      "step": 3641
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0117482422854672e-06,
+      "loss": 1.1748,
+      "step": 3642
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.009131892283415e-06,
+      "loss": 1.1667,
+      "step": 3643
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0065161899522278e-06,
+      "loss": 1.3012,
+      "step": 3644
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0039011361428466e-06,
+      "loss": 1.1453,
+      "step": 3645
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.001286731706002e-06,
+      "loss": 1.1916,
+      "step": 3646
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.998672977492214e-06,
+      "loss": 1.1558,
+      "step": 3647
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.996059874351794e-06,
+      "loss": 1.2393,
+      "step": 3648
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9934474231348377e-06,
+      "loss": 1.1811,
+      "step": 3649
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9908356246912275e-06,
+      "loss": 0.9027,
+      "step": 3650
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9882244798706372e-06,
+      "loss": 1.1971,
+      "step": 3651
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9856139895225284e-06,
+      "loss": 1.1997,
+      "step": 3652
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9830041544961475e-06,
+      "loss": 1.2192,
+      "step": 3653
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.980394975640526e-06,
+      "loss": 1.1248,
+      "step": 3654
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.977786453804485e-06,
+      "loss": 1.1576,
+      "step": 3655
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.975178589836632e-06,
+      "loss": 1.1985,
+      "step": 3656
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.97257138458536e-06,
+      "loss": 0.9043,
+      "step": 3657
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.969964838898845e-06,
+      "loss": 1.218,
+      "step": 3658
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9673589536250517e-06,
+      "loss": 1.109,
+      "step": 3659
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9647537296117278e-06,
+      "loss": 1.2533,
+      "step": 3660
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.962149167706411e-06,
+      "loss": 1.0526,
+      "step": 3661
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.959545268756414e-06,
+      "loss": 1.1593,
+      "step": 3662
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.956942033608843e-06,
+      "loss": 1.1351,
+      "step": 3663
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.954339463110582e-06,
+      "loss": 0.8861,
+      "step": 3664
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.951737558108305e-06,
+      "loss": 1.1892,
+      "step": 3665
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.949136319448463e-06,
+      "loss": 1.1326,
+      "step": 3666
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9465357479772937e-06,
+      "loss": 1.1368,
+      "step": 3667
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9439358445408186e-06,
+      "loss": 1.1948,
+      "step": 3668
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9413366099848366e-06,
+      "loss": 1.1283,
+      "step": 3669
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.938738045154937e-06,
+      "loss": 0.9723,
+      "step": 3670
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9361401508964856e-06,
+      "loss": 1.1613,
+      "step": 3671
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.933542928054632e-06,
+      "loss": 1.1061,
+      "step": 3672
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9309463774743047e-06,
+      "loss": 1.1793,
+      "step": 3673
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.928350500000218e-06,
+      "loss": 1.087,
+      "step": 3674
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9257552964768644e-06,
+      "loss": 1.2581,
+      "step": 3675
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9231607677485187e-06,
+      "loss": 1.1346,
+      "step": 3676
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9205669146592323e-06,
+      "loss": 0.866,
+      "step": 3677
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.91797373805284e-06,
+      "loss": 1.2073,
+      "step": 3678
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9153812387729603e-06,
+      "loss": 1.1297,
+      "step": 3679
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.912789417662984e-06,
+      "loss": 1.1862,
+      "step": 3680
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.910198275566085e-06,
+      "loss": 1.1437,
+      "step": 3681
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9076078133252117e-06,
+      "loss": 1.1542,
+      "step": 3682
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9050180317831036e-06,
+      "loss": 1.1777,
+      "step": 3683
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.902428931782266e-06,
+      "loss": 1.1246,
+      "step": 3684
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.899840514164987e-06,
+      "loss": 1.211,
+      "step": 3685
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8972527797733298e-06,
+      "loss": 0.89,
+      "step": 3686
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8946657294491452e-06,
+      "loss": 1.1237,
+      "step": 3687
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8920793640340515e-06,
+      "loss": 1.169,
+      "step": 3688
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8894936843694465e-06,
+      "loss": 1.2429,
+      "step": 3689
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.886908691296504e-06,
+      "loss": 1.1191,
+      "step": 3690
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8843243856561777e-06,
+      "loss": 1.1849,
+      "step": 3691
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8817407682891986e-06,
+      "loss": 1.1545,
+      "step": 3692
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8791578400360686e-06,
+      "loss": 0.9174,
+      "step": 3693
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8765756017370673e-06,
+      "loss": 1.2068,
+      "step": 3694
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8739940542322507e-06,
+      "loss": 1.1267,
+      "step": 3695
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8714131983614535e-06,
+      "loss": 1.1996,
+      "step": 3696
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8688330349642794e-06,
+      "loss": 1.0822,
+      "step": 3697
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8662535648801076e-06,
+      "loss": 1.1643,
+      "step": 3698
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.863674788948097e-06,
+      "loss": 0.9736,
+      "step": 3699
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8610967080071726e-06,
+      "loss": 1.144,
+      "step": 3700
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8585193228960438e-06,
+      "loss": 1.22,
+      "step": 3701
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8559426344531816e-06,
+      "loss": 1.1433,
+      "step": 3702
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.853366643516841e-06,
+      "loss": 1.1236,
+      "step": 3703
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8507913509250418e-06,
+      "loss": 1.1042,
+      "step": 3704
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8482167575155845e-06,
+      "loss": 1.175,
+      "step": 3705
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.845642864126034e-06,
+      "loss": 1.1345,
+      "step": 3706
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.843069671593734e-06,
+      "loss": 1.1132,
+      "step": 3707
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8404971807557957e-06,
+      "loss": 1.1601,
+      "step": 3708
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8379253924491047e-06,
+      "loss": 1.1008,
+      "step": 3709
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.835354307510319e-06,
+      "loss": 1.1798,
+      "step": 3710
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.832783926775865e-06,
+      "loss": 1.1869,
+      "step": 3711
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.830214251081941e-06,
+      "loss": 0.9361,
+      "step": 3712
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8276452812645115e-06,
+      "loss": 1.1306,
+      "step": 3713
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8250770181593245e-06,
+      "loss": 1.1873,
+      "step": 3714
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.822509462601886e-06,
+      "loss": 1.088,
+      "step": 3715
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.819942615427473e-06,
+      "loss": 1.1909,
+      "step": 3716
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.817376477471132e-06,
+      "loss": 1.1896,
+      "step": 3717
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8148110495676894e-06,
+      "loss": 1.1012,
+      "step": 3718
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.812246332551728e-06,
+      "loss": 1.1941,
+      "step": 3719
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8096823272576034e-06,
+      "loss": 1.1507,
+      "step": 3720
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8071190345194353e-06,
+      "loss": 1.233,
+      "step": 3721
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8045564551711256e-06,
+      "loss": 0.8663,
+      "step": 3722
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8019945900463307e-06,
+      "loss": 1.1633,
+      "step": 3723
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7994334399784773e-06,
+      "loss": 1.1753,
+      "step": 3724
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7968730058007605e-06,
+      "loss": 1.1361,
+      "step": 3725
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7943132883461434e-06,
+      "loss": 1.1461,
+      "step": 3726
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7917542884473582e-06,
+      "loss": 1.1259,
+      "step": 3727
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7891960069368985e-06,
+      "loss": 0.8934,
+      "step": 3728
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.786638444647025e-06,
+      "loss": 1.1779,
+      "step": 3729
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7840816024097676e-06,
+      "loss": 1.1414,
+      "step": 3730
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7815254810569227e-06,
+      "loss": 1.0939,
+      "step": 3731
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.778970081420048e-06,
+      "loss": 1.2079,
+      "step": 3732
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7764154043304654e-06,
+      "loss": 1.0707,
+      "step": 3733
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.773861450619269e-06,
+      "loss": 1.1904,
+      "step": 3734
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.771308221117309e-06,
+      "loss": 0.919,
+      "step": 3735
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7687557166552087e-06,
+      "loss": 1.0955,
+      "step": 3736
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7662039380633466e-06,
+      "loss": 1.2439,
+      "step": 3737
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7636528861718726e-06,
+      "loss": 1.1103,
+      "step": 3738
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7611025618106944e-06,
+      "loss": 1.1922,
+      "step": 3739
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7585529658094887e-06,
+      "loss": 1.1612,
+      "step": 3740
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7560040989976894e-06,
+      "loss": 0.9384,
+      "step": 3741
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.753455962204499e-06,
+      "loss": 1.2173,
+      "step": 3742
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7509085562588743e-06,
+      "loss": 1.1776,
+      "step": 3743
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.748361881989546e-06,
+      "loss": 1.1824,
+      "step": 3744
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7458159402249947e-06,
+      "loss": 1.1366,
+      "step": 3745
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7432707317934724e-06,
+      "loss": 1.1392,
+      "step": 3746
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.740726257522987e-06,
+      "loss": 1.1249,
+      "step": 3747
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7381825182413047e-06,
+      "loss": 0.8168,
+      "step": 3748
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7356395147759644e-06,
+      "loss": 1.2081,
+      "step": 3749
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.733097247954255e-06,
+      "loss": 1.1051,
+      "step": 3750
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7305557186032283e-06,
+      "loss": 1.2959,
+      "step": 3751
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7280149275496926e-06,
+      "loss": 1.1697,
+      "step": 3752
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.725474875620228e-06,
+      "loss": 1.1798,
+      "step": 3753
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.722935563641163e-06,
+      "loss": 1.1832,
+      "step": 3754
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7203969924385885e-06,
+      "loss": 1.1927,
+      "step": 3755
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7178591628383523e-06,
+      "loss": 1.1564,
+      "step": 3756
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7153220756660654e-06,
+      "loss": 0.9701,
+      "step": 3757
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7127857317470967e-06,
+      "loss": 1.1687,
+      "step": 3758
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7102501319065706e-06,
+      "loss": 1.2039,
+      "step": 3759
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7077152769693677e-06,
+      "loss": 1.1676,
+      "step": 3760
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7051811677601315e-06,
+      "loss": 1.1214,
+      "step": 3761
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7026478051032625e-06,
+      "loss": 1.173,
+      "step": 3762
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.700115189822914e-06,
+      "loss": 1.1001,
+      "step": 3763
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.697583322742997e-06,
+      "loss": 0.969,
+      "step": 3764
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.695052204687182e-06,
+      "loss": 1.1517,
+      "step": 3765
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6925218364788972e-06,
+      "loss": 1.0725,
+      "step": 3766
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6899922189413213e-06,
+      "loss": 1.2489,
+      "step": 3767
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6874633528973905e-06,
+      "loss": 1.2004,
+      "step": 3768
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6849352391698004e-06,
+      "loss": 1.1614,
+      "step": 3769
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.682407878580996e-06,
+      "loss": 1.1357,
+      "step": 3770
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6798812719531843e-06,
+      "loss": 1.0118,
+      "step": 3771
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6773554201083186e-06,
+      "loss": 1.1641,
+      "step": 3772
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.674830323868115e-06,
+      "loss": 1.1457,
+      "step": 3773
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.672305984054037e-06,
+      "loss": 1.1762,
+      "step": 3774
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6697824014873076e-06,
+      "loss": 1.1701,
+      "step": 3775
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6672595769888963e-06,
+      "loss": 1.1253,
+      "step": 3776
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.664737511379537e-06,
+      "loss": 0.8713,
+      "step": 3777
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.662216205479704e-06,
+      "loss": 1.1862,
+      "step": 3778
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6596956601096358e-06,
+      "loss": 1.1283,
+      "step": 3779
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6571758760893127e-06,
+      "loss": 1.1784,
+      "step": 3780
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.654656854238479e-06,
+      "loss": 1.1667,
+      "step": 3781
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.652138595376622e-06,
+      "loss": 1.17,
+      "step": 3782
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6496211003229795e-06,
+      "loss": 1.1628,
+      "step": 3783
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6471043698965527e-06,
+      "loss": 1.0389,
+      "step": 3784
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6445884049160843e-06,
+      "loss": 1.2138,
+      "step": 3785
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6420732062000676e-06,
+      "loss": 1.1844,
+      "step": 3786
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.639558774566747e-06,
+      "loss": 1.1361,
+      "step": 3787
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6370451108341266e-06,
+      "loss": 1.1892,
+      "step": 3788
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6345322158199503e-06,
+      "loss": 1.1214,
+      "step": 3789
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.632020090341715e-06,
+      "loss": 1.2053,
+      "step": 3790
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6295087352166643e-06,
+      "loss": 1.1407,
+      "step": 3791
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.626998151261798e-06,
+      "loss": 1.1853,
+      "step": 3792
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6244883392938618e-06,
+      "loss": 0.9408,
+      "step": 3793
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.621979300129349e-06,
+      "loss": 1.1814,
+      "step": 3794
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6194710345845e-06,
+      "loss": 1.133,
+      "step": 3795
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6169635434753076e-06,
+      "loss": 1.1828,
+      "step": 3796
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.614456827617513e-06,
+      "loss": 1.1382,
+      "step": 3797
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.611950887826601e-06,
+      "loss": 1.2469,
+      "step": 3798
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.609445724917804e-06,
+      "loss": 1.1245,
+      "step": 3799
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6069413397061063e-06,
+      "loss": 0.9018,
+      "step": 3800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6044377330062375e-06,
+      "loss": 1.1703,
+      "step": 3801
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6019349056326717e-06,
+      "loss": 1.1011,
+      "step": 3802
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.599432858399628e-06,
+      "loss": 1.1471,
+      "step": 3803
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5969315921210803e-06,
+      "loss": 1.0943,
+      "step": 3804
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5944311076107366e-06,
+      "loss": 1.2054,
+      "step": 3805
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.591931405682061e-06,
+      "loss": 0.931,
+      "step": 3806
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5894324871482557e-06,
+      "loss": 1.1478,
+      "step": 3807
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5869343528222744e-06,
+      "loss": 1.1785,
+      "step": 3808
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5844370035168077e-06,
+      "loss": 1.0984,
+      "step": 3809
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5819404400442998e-06,
+      "loss": 1.1366,
+      "step": 3810
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5794446632169312e-06,
+      "loss": 1.1049,
+      "step": 3811
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.576949673846633e-06,
+      "loss": 1.1914,
+      "step": 3812
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.574455472745076e-06,
+      "loss": 1.0888,
+      "step": 3813
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5719620607236737e-06,
+      "loss": 1.1284,
+      "step": 3814
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.569469438593587e-06,
+      "loss": 1.1873,
+      "step": 3815
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5669776071657194e-06,
+      "loss": 1.1356,
+      "step": 3816
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.564486567250714e-06,
+      "loss": 1.1129,
+      "step": 3817
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5619963196589557e-06,
+      "loss": 1.1493,
+      "step": 3818
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.559506865200576e-06,
+      "loss": 0.8955,
+      "step": 3819
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5570182046854485e-06,
+      "loss": 1.1015,
+      "step": 3820
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5545303389231846e-06,
+      "loss": 1.1875,
+      "step": 3821
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5520432687231333e-06,
+      "loss": 1.1303,
+      "step": 3822
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5495569948944e-06,
+      "loss": 1.1318,
+      "step": 3823
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5470715182458165e-06,
+      "loss": 1.1773,
+      "step": 3824
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.544586839585961e-06,
+      "loss": 1.103,
+      "step": 3825
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5421029597231476e-06,
+      "loss": 1.1612,
+      "step": 3826
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5396198794654383e-06,
+      "loss": 1.1169,
+      "step": 3827
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5371375996206303e-06,
+      "loss": 1.203,
+      "step": 3828
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5346561209962616e-06,
+      "loss": 1.0832,
+      "step": 3829
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.532175444399605e-06,
+      "loss": 1.0903,
+      "step": 3830
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.529695570637679e-06,
+      "loss": 1.1819,
+      "step": 3831
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.52721650051724e-06,
+      "loss": 1.1866,
+      "step": 3832
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.524738234844779e-06,
+      "loss": 1.205,
+      "step": 3833
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5222607744265266e-06,
+      "loss": 1.125,
+      "step": 3834
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5197841200684525e-06,
+      "loss": 0.9258,
+      "step": 3835
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.517308272576267e-06,
+      "loss": 1.1519,
+      "step": 3836
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5148332327554127e-06,
+      "loss": 1.2091,
+      "step": 3837
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5123590014110694e-06,
+      "loss": 1.1151,
+      "step": 3838
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5098855793481604e-06,
+      "loss": 1.1862,
+      "step": 3839
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.507412967371337e-06,
+      "loss": 1.0889,
+      "step": 3840
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.504941166284996e-06,
+      "loss": 1.1801,
+      "step": 3841
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5024701768932615e-06,
+      "loss": 0.8541,
+      "step": 3842
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5000000000000015e-06,
+      "loss": 1.1814,
+      "step": 3843
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.497530636408812e-06,
+      "loss": 1.2053,
+      "step": 3844
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.495062086923033e-06,
+      "loss": 1.138,
+      "step": 3845
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4925943523457304e-06,
+      "loss": 1.2348,
+      "step": 3846
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.490127433479714e-06,
+      "loss": 1.1237,
+      "step": 3847
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.487661331127522e-06,
+      "loss": 1.0033,
+      "step": 3848
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4851960460914264e-06,
+      "loss": 1.147,
+      "step": 3849
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4827315791734387e-06,
+      "loss": 1.0625,
+      "step": 3850
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4802679311753024e-06,
+      "loss": 1.1651,
+      "step": 3851
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4778051028984918e-06,
+      "loss": 1.1168,
+      "step": 3852
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.475343095144214e-06,
+      "loss": 1.2022,
+      "step": 3853
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4728819087134144e-06,
+      "loss": 1.111,
+      "step": 3854
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4704215444067684e-06,
+      "loss": 0.926,
+      "step": 3855
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4679620030246836e-06,
+      "loss": 1.1708,
+      "step": 3856
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.465503285367297e-06,
+      "loss": 1.2782,
+      "step": 3857
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.463045392234483e-06,
+      "loss": 0.785,
+      "step": 3858
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4605883244258477e-06,
+      "loss": 1.1487,
+      "step": 3859
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.458132082740724e-06,
+      "loss": 1.2064,
+      "step": 3860
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4556766679781763e-06,
+      "loss": 1.17,
+      "step": 3861
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4532220809370045e-06,
+      "loss": 1.1889,
+      "step": 3862
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.450768322415739e-06,
+      "loss": 1.1514,
+      "step": 3863
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4483153932126364e-06,
+      "loss": 0.8915,
+      "step": 3864
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4458632941256834e-06,
+      "loss": 1.1735,
+      "step": 3865
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4434120259526013e-06,
+      "loss": 1.0964,
+      "step": 3866
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4409615894908407e-06,
+      "loss": 1.1699,
+      "step": 3867
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4385119855375773e-06,
+      "loss": 1.0686,
+      "step": 3868
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4360632148897162e-06,
+      "loss": 1.157,
+      "step": 3869
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4336152783438984e-06,
+      "loss": 1.1836,
+      "step": 3870
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4311681766964833e-06,
+      "loss": 1.0052,
+      "step": 3871
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4287219107435694e-06,
+      "loss": 1.1835,
+      "step": 3872
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.426276481280973e-06,
+      "loss": 1.1686,
+      "step": 3873
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4238318891042477e-06,
+      "loss": 1.1648,
+      "step": 3874
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.421388135008666e-06,
+      "loss": 1.1428,
+      "step": 3875
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4189452197892366e-06,
+      "loss": 1.2646,
+      "step": 3876
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4165031442406857e-06,
+      "loss": 1.0984,
+      "step": 3877
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.414061909157476e-06,
+      "loss": 0.9544,
+      "step": 3878
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.411621515333788e-06,
+      "loss": 1.1632,
+      "step": 3879
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4091819635635366e-06,
+      "loss": 1.1586,
+      "step": 3880
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4067432546403545e-06,
+      "loss": 1.1635,
+      "step": 3881
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4043053893576097e-06,
+      "loss": 1.2176,
+      "step": 3882
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.401868368508387e-06,
+      "loss": 1.1345,
+      "step": 3883
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.399432192885499e-06,
+      "loss": 0.9584,
+      "step": 3884
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3969968632814862e-06,
+      "loss": 1.19,
+      "step": 3885
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3945623804886137e-06,
+      "loss": 1.0833,
+      "step": 3886
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.392128745298868e-06,
+      "loss": 1.1796,
+      "step": 3887
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3896959585039585e-06,
+      "loss": 1.1116,
+      "step": 3888
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3872640208953248e-06,
+      "loss": 1.1892,
+      "step": 3889
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3848329332641273e-06,
+      "loss": 1.161,
+      "step": 3890
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3824026964012487e-06,
+      "loss": 1.0907,
+      "step": 3891
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3799733110972927e-06,
+      "loss": 1.2927,
+      "step": 3892
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.377544778142592e-06,
+      "loss": 0.9581,
+      "step": 3893
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3751170983272e-06,
+      "loss": 1.1262,
+      "step": 3894
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3726902724408896e-06,
+      "loss": 1.1695,
+      "step": 3895
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3702643012731562e-06,
+      "loss": 1.0906,
+      "step": 3896
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3678391856132203e-06,
+      "loss": 1.1563,
+      "step": 3897
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3654149262500255e-06,
+      "loss": 1.0904,
+      "step": 3898
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3629915239722307e-06,
+      "loss": 1.1446,
+      "step": 3899
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3605689795682178e-06,
+      "loss": 0.9605,
+      "step": 3900
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.358147293826093e-06,
+      "loss": 1.2228,
+      "step": 3901
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3557264675336826e-06,
+      "loss": 1.1139,
+      "step": 3902
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.35330650147853e-06,
+      "loss": 1.1501,
+      "step": 3903
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.350887396447899e-06,
+      "loss": 1.1075,
+      "step": 3904
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3484691532287784e-06,
+      "loss": 1.1582,
+      "step": 3905
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3460517726078696e-06,
+      "loss": 1.1292,
+      "step": 3906
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3436352553716e-06,
+      "loss": 1.1024,
+      "step": 3907
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.34121960230611e-06,
+      "loss": 1.194,
+      "step": 3908
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3388048141972645e-06,
+      "loss": 1.2181,
+      "step": 3909
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3363908918306423e-06,
+      "loss": 1.2149,
+      "step": 3910
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.333977835991545e-06,
+      "loss": 1.0951,
+      "step": 3911
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3315656474649866e-06,
+      "loss": 1.136,
+      "step": 3912
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.329154327035706e-06,
+      "loss": 0.8657,
+      "step": 3913
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.326743875488152e-06,
+      "loss": 1.0583,
+      "step": 3914
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.324334293606499e-06,
+      "loss": 1.1656,
+      "step": 3915
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3219255821746302e-06,
+      "loss": 1.1319,
+      "step": 3916
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.319517741976153e-06,
+      "loss": 1.1875,
+      "step": 3917
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.317110773794386e-06,
+      "loss": 1.0916,
+      "step": 3918
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.314704678412365e-06,
+      "loss": 1.1799,
+      "step": 3919
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3122994566128444e-06,
+      "loss": 0.9093,
+      "step": 3920
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3098951091782946e-06,
+      "loss": 1.2259,
+      "step": 3921
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3074916368908983e-06,
+      "loss": 1.1236,
+      "step": 3922
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.305089040532553e-06,
+      "loss": 1.1748,
+      "step": 3923
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.302687320884876e-06,
+      "loss": 1.0923,
+      "step": 3924
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3002864787291972e-06,
+      "loss": 1.1799,
+      "step": 3925
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.29788651484656e-06,
+      "loss": 0.8234,
+      "step": 3926
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2954874300177197e-06,
+      "loss": 1.2529,
+      "step": 3927
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.293089225023152e-06,
+      "loss": 1.1418,
+      "step": 3928
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2906919006430426e-06,
+      "loss": 0.9035,
+      "step": 3929
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.288295457657291e-06,
+      "loss": 1.1635,
+      "step": 3930
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.285899896845507e-06,
+      "loss": 1.181,
+      "step": 3931
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.283505218987019e-06,
+      "loss": 1.1444,
+      "step": 3932
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2811114248608675e-06,
+      "loss": 1.1502,
+      "step": 3933
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.278718515245801e-06,
+      "loss": 1.1461,
+      "step": 3934
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.276326490920282e-06,
+      "loss": 1.2034,
+      "step": 3935
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2739353526624868e-06,
+      "loss": 0.9876,
+      "step": 3936
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.271545101250305e-06,
+      "loss": 1.0791,
+      "step": 3937
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2691557374613328e-06,
+      "loss": 1.2248,
+      "step": 3938
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.266767262072878e-06,
+      "loss": 1.0776,
+      "step": 3939
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2643796758619657e-06,
+      "loss": 1.1667,
+      "step": 3940
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2619929796053238e-06,
+      "loss": 1.0908,
+      "step": 3941
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2596071740793974e-06,
+      "loss": 0.8386,
+      "step": 3942
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.257222260060335e-06,
+      "loss": 1.1408,
+      "step": 3943
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2548382383240037e-06,
+      "loss": 1.1209,
+      "step": 3944
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2524551096459703e-06,
+      "loss": 1.1332,
+      "step": 3945
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.25007287480152e-06,
+      "loss": 1.1808,
+      "step": 3946
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2476915345656406e-06,
+      "loss": 1.0859,
+      "step": 3947
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2453110897130355e-06,
+      "loss": 1.1558,
+      "step": 3948
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.242931541018108e-06,
+      "loss": 0.8532,
+      "step": 3949
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2405528892549792e-06,
+      "loss": 1.0751,
+      "step": 3950
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.238175135197471e-06,
+      "loss": 1.1908,
+      "step": 3951
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2357982796191193e-06,
+      "loss": 1.1133,
+      "step": 3952
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.233422323293163e-06,
+      "loss": 1.2014,
+      "step": 3953
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.231047266992548e-06,
+      "loss": 1.1521,
+      "step": 3954
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2286731114899322e-06,
+      "loss": 0.8274,
+      "step": 3955
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2262998575576793e-06,
+      "loss": 1.1578,
+      "step": 3956
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.223927505967857e-06,
+      "loss": 1.1437,
+      "step": 3957
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2215560574922373e-06,
+      "loss": 1.1527,
+      "step": 3958
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.219185512902305e-06,
+      "loss": 1.1656,
+      "step": 3959
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2168158729692484e-06,
+      "loss": 1.1526,
+      "step": 3960
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2144471384639606e-06,
+      "loss": 1.1004,
+      "step": 3961
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2120793101570366e-06,
+      "loss": 1.1072,
+      "step": 3962
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2097123888187825e-06,
+      "loss": 1.1595,
+      "step": 3963
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2073463752192102e-06,
+      "loss": 1.1575,
+      "step": 3964
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2049812701280304e-06,
+      "loss": 0.8905,
+      "step": 3965
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.202617074314659e-06,
+      "loss": 1.114,
+      "step": 3966
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.20025378854822e-06,
+      "loss": 1.2133,
+      "step": 3967
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.197891413597542e-06,
+      "loss": 1.1665,
+      "step": 3968
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1955299502311523e-06,
+      "loss": 1.1898,
+      "step": 3969
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1931693992172827e-06,
+      "loss": 1.1614,
+      "step": 3970
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1908097613238715e-06,
+      "loss": 0.975,
+      "step": 3971
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.188451037318559e-06,
+      "loss": 1.13,
+      "step": 3972
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1860932279686866e-06,
+      "loss": 1.1339,
+      "step": 3973
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1837363340412965e-06,
+      "loss": 1.1487,
+      "step": 3974
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.181380356303139e-06,
+      "loss": 1.1338,
+      "step": 3975
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1790252955206586e-06,
+      "loss": 1.2142,
+      "step": 3976
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.17667115246001e-06,
+      "loss": 1.1399,
+      "step": 3977
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.174317927887041e-06,
+      "loss": 0.9088,
+      "step": 3978
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.171965622567308e-06,
+      "loss": 1.1624,
+      "step": 3979
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1696142372660618e-06,
+      "loss": 1.1882,
+      "step": 3980
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1672637727482603e-06,
+      "loss": 1.1768,
+      "step": 3981
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.164914229778555e-06,
+      "loss": 1.1193,
+      "step": 3982
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1625656091213056e-06,
+      "loss": 1.1562,
+      "step": 3983
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1602179115405637e-06,
+      "loss": 1.1593,
+      "step": 3984
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.157871137800084e-06,
+      "loss": 1.0699,
+      "step": 3985
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1555252886633223e-06,
+      "loss": 1.1552,
+      "step": 3986
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1531803648934333e-06,
+      "loss": 1.1906,
+      "step": 3987
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1508363672532685e-06,
+      "loss": 1.0801,
+      "step": 3988
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1484932965053772e-06,
+      "loss": 1.1585,
+      "step": 3989
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1461511534120104e-06,
+      "loss": 1.1154,
+      "step": 3990
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.143809938735119e-06,
+      "loss": 0.9734,
+      "step": 3991
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.141469653236346e-06,
+      "loss": 1.1527,
+      "step": 3992
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1391302976770335e-06,
+      "loss": 1.1485,
+      "step": 3993
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1367918728182247e-06,
+      "loss": 1.1657,
+      "step": 3994
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.134454379420659e-06,
+      "loss": 1.1087,
+      "step": 3995
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.132117818244771e-06,
+      "loss": 1.1921,
+      "step": 3996
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1297821900506897e-06,
+      "loss": 1.1883,
+      "step": 3997
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1274474955982464e-06,
+      "loss": 1.1237,
+      "step": 3998
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1251137356469677e-06,
+      "loss": 1.2396,
+      "step": 3999
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1227809109560717e-06,
+      "loss": 0.9131,
+      "step": 4000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1204490222844733e-06,
+      "loss": 1.0651,
+      "step": 4001
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.118118070390786e-06,
+      "loss": 1.1484,
+      "step": 4002
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1157880560333197e-06,
+      "loss": 1.1208,
+      "step": 4003
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1134589799700744e-06,
+      "loss": 1.151,
+      "step": 4004
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1111308429587446e-06,
+      "loss": 1.0956,
+      "step": 4005
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1088036457567244e-06,
+      "loss": 1.1352,
+      "step": 4006
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1064773891211017e-06,
+      "loss": 0.9486,
+      "step": 4007
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.104152073808653e-06,
+      "loss": 1.1723,
+      "step": 4008
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1018277005758507e-06,
+      "loss": 1.1699,
+      "step": 4009
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.099504270178866e-06,
+      "loss": 1.2027,
+      "step": 4010
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0971817833735548e-06,
+      "loss": 1.2054,
+      "step": 4011
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.094860240915474e-06,
+      "loss": 1.1866,
+      "step": 4012
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0925396435598665e-06,
+      "loss": 1.0841,
+      "step": 4013
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.090219992061675e-06,
+      "loss": 1.081,
+      "step": 4014
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.087901287175526e-06,
+      "loss": 1.1904,
+      "step": 4015
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.085583529655747e-06,
+      "loss": 1.1077,
+      "step": 4016
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.083266720256348e-06,
+      "loss": 1.2041,
+      "step": 4017
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.08095085973104e-06,
+      "loss": 1.1207,
+      "step": 4018
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0786359488332185e-06,
+      "loss": 1.193,
+      "step": 4019
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.07632198831597e-06,
+      "loss": 0.8844,
+      "step": 4020
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0740089789320765e-06,
+      "loss": 1.1133,
+      "step": 4021
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.07169692143401e-06,
+      "loss": 1.1274,
+      "step": 4022
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.069385816573928e-06,
+      "loss": 1.1215,
+      "step": 4023
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0670756651036813e-06,
+      "loss": 1.1794,
+      "step": 4024
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0647664677748095e-06,
+      "loss": 1.1576,
+      "step": 4025
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0624582253385473e-06,
+      "loss": 1.1783,
+      "step": 4026
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0601509385458097e-06,
+      "loss": 0.8215,
+      "step": 4027
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0578446081472046e-06,
+      "loss": 1.1271,
+      "step": 4028
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.055539234893031e-06,
+      "loss": 1.1523,
+      "step": 4029
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.053234819533276e-06,
+      "loss": 1.2703,
+      "step": 4030
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0509313628176136e-06,
+      "loss": 1.1356,
+      "step": 4031
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.048628865495403e-06,
+      "loss": 1.1263,
+      "step": 4032
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0463273283156966e-06,
+      "loss": 1.1439,
+      "step": 4033
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0440267520272345e-06,
+      "loss": 1.0877,
+      "step": 4034
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0417271373784403e-06,
+      "loss": 1.1571,
+      "step": 4035
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0394284851174252e-06,
+      "loss": 0.868,
+      "step": 4036
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.037130795991989e-06,
+      "loss": 1.144,
+      "step": 4037
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0348340707496213e-06,
+      "loss": 1.1961,
+      "step": 4038
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.032538310137492e-06,
+      "loss": 1.0939,
+      "step": 4039
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0302435149024574e-06,
+      "loss": 1.166,
+      "step": 4040
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0279496857910667e-06,
+      "loss": 1.1292,
+      "step": 4041
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.025656823549547e-06,
+      "loss": 1.1346,
+      "step": 4042
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.023364928923817e-06,
+      "loss": 0.8476,
+      "step": 4043
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.021074002659475e-06,
+      "loss": 1.1861,
+      "step": 4044
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.018784045501809e-06,
+      "loss": 1.2513,
+      "step": 4045
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.016495058195788e-06,
+      "loss": 1.0883,
+      "step": 4046
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0142070414860704e-06,
+      "loss": 1.197,
+      "step": 4047
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.011919996116991e-06,
+      "loss": 1.1119,
+      "step": 4048
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0096339228325777e-06,
+      "loss": 0.8431,
+      "step": 4049
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0073488223765346e-06,
+      "loss": 1.0895,
+      "step": 4050
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0050646954922553e-06,
+      "loss": 1.1607,
+      "step": 4051
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0027815429228105e-06,
+      "loss": 1.0582,
+      "step": 4052
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.000499365410961e-06,
+      "loss": 1.1843,
+      "step": 4053
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.998218163699145e-06,
+      "loss": 1.1464,
+      "step": 4054
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9959379385294824e-06,
+      "loss": 1.1815,
+      "step": 4055
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9936586906437805e-06,
+      "loss": 0.8622,
+      "step": 4056
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.991380420783528e-06,
+      "loss": 1.0456,
+      "step": 4057
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.989103129689891e-06,
+      "loss": 1.1608,
+      "step": 4058
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9868268181037186e-06,
+      "loss": 1.1349,
+      "step": 4059
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.984551486765544e-06,
+      "loss": 1.1674,
+      "step": 4060
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.982277136415582e-06,
+      "loss": 1.0924,
+      "step": 4061
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9800037677937227e-06,
+      "loss": 1.095,
+      "step": 4062
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.977731381639541e-06,
+      "loss": 1.1805,
+      "step": 4063
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9754599786922913e-06,
+      "loss": 1.177,
+      "step": 4064
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.973189559690911e-06,
+      "loss": 1.1411,
+      "step": 4065
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.970920125374012e-06,
+      "loss": 1.1149,
+      "step": 4066
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9686516764798886e-06,
+      "loss": 1.1568,
+      "step": 4067
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9663842137465144e-06,
+      "loss": 1.1981,
+      "step": 4068
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9641177379115445e-06,
+      "loss": 1.1427,
+      "step": 4069
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9618522497123095e-06,
+      "loss": 1.1702,
+      "step": 4070
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9595877498858175e-06,
+      "loss": 1.1871,
+      "step": 4071
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9573242391687597e-06,
+      "loss": 0.8003,
+      "step": 4072
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9550617182975046e-06,
+      "loss": 1.0913,
+      "step": 4073
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.952800188008096e-06,
+      "loss": 1.1686,
+      "step": 4074
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.950539649036255e-06,
+      "loss": 1.1635,
+      "step": 4075
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9482801021173853e-06,
+      "loss": 1.2128,
+      "step": 4076
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9460215479865613e-06,
+      "loss": 1.1581,
+      "step": 4077
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.943763987378541e-06,
+      "loss": 1.0019,
+      "step": 4078
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9415074210277523e-06,
+      "loss": 1.1505,
+      "step": 4079
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9392518496683072e-06,
+      "loss": 1.1256,
+      "step": 4080
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.936997274033986e-06,
+      "loss": 1.1542,
+      "step": 4081
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9347436948582525e-06,
+      "loss": 1.1212,
+      "step": 4082
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9324911128742406e-06,
+      "loss": 1.1615,
+      "step": 4083
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.930239528814765e-06,
+      "loss": 1.1738,
+      "step": 4084
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9279889434123083e-06,
+      "loss": 0.8403,
+      "step": 4085
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.925739357399038e-06,
+      "loss": 1.1806,
+      "step": 4086
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.923490771506787e-06,
+      "loss": 1.1263,
+      "step": 4087
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.92124318646707e-06,
+      "loss": 1.1965,
+      "step": 4088
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.918996603011073e-06,
+      "loss": 1.1258,
+      "step": 4089
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9167510218696535e-06,
+      "loss": 1.1575,
+      "step": 4090
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9145064437733472e-06,
+      "loss": 1.2244,
+      "step": 4091
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9122628694523653e-06,
+      "loss": 0.9202,
+      "step": 4092
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9100202996365867e-06,
+      "loss": 1.1019,
+      "step": 4093
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9077787350555643e-06,
+      "loss": 1.1361,
+      "step": 4094
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9055381764385272e-06,
+      "loss": 1.1713,
+      "step": 4095
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.903298624514378e-06,
+      "loss": 1.1879,
+      "step": 4096
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.901060080011688e-06,
+      "loss": 1.0646,
+      "step": 4097
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8988225436587005e-06,
+      "loss": 0.7739,
+      "step": 4098
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8965860161833338e-06,
+      "loss": 1.1612,
+      "step": 4099
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8943504983131788e-06,
+      "loss": 1.1315,
+      "step": 4100
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8921159907754956e-06,
+      "loss": 1.1954,
+      "step": 4101
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8898824942972121e-06,
+      "loss": 1.0634,
+      "step": 4102
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.887650009604935e-06,
+      "loss": 1.0401,
+      "step": 4103
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8854185374249384e-06,
+      "loss": 1.153,
+      "step": 4104
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8831880784831658e-06,
+      "loss": 1.1137,
+      "step": 4105
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8809586335052294e-06,
+      "loss": 1.165,
+      "step": 4106
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8787302032164168e-06,
+      "loss": 1.1361,
+      "step": 4107
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8765027883416843e-06,
+      "loss": 1.1179,
+      "step": 4108
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8742763896056538e-06,
+      "loss": 1.1468,
+      "step": 4109
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8720510077326182e-06,
+      "loss": 1.1133,
+      "step": 4110
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.869826643446544e-06,
+      "loss": 1.1929,
+      "step": 4111
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8676032974710589e-06,
+      "loss": 1.1673,
+      "step": 4112
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.865380970529469e-06,
+      "loss": 1.0844,
+      "step": 4113
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8631596633447379e-06,
+      "loss": 0.9244,
+      "step": 4114
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8609393766395083e-06,
+      "loss": 1.1626,
+      "step": 4115
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8587201111360808e-06,
+      "loss": 1.0863,
+      "step": 4116
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.856501867556434e-06,
+      "loss": 1.1664,
+      "step": 4117
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.854284646622203e-06,
+      "loss": 1.1046,
+      "step": 4118
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8520684490547014e-06,
+      "loss": 1.176,
+      "step": 4119
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8498532755749017e-06,
+      "loss": 1.1072,
+      "step": 4120
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8476391269034438e-06,
+      "loss": 0.8918,
+      "step": 4121
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8454260037606391e-06,
+      "loss": 1.251,
+      "step": 4122
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8432139068664646e-06,
+      "loss": 1.171,
+      "step": 4123
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8410028369405587e-06,
+      "loss": 1.2094,
+      "step": 4124
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8387927947022271e-06,
+      "loss": 1.139,
+      "step": 4125
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8365837808704445e-06,
+      "loss": 1.1719,
+      "step": 4126
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.834375796163851e-06,
+      "loss": 0.9213,
+      "step": 4127
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8321688413007483e-06,
+      "loss": 1.1712,
+      "step": 4128
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8299629169991023e-06,
+      "loss": 1.1725,
+      "step": 4129
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8277580239765485e-06,
+      "loss": 1.1289,
+      "step": 4130
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8255541629503865e-06,
+      "loss": 1.172,
+      "step": 4131
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.823351334637576e-06,
+      "loss": 1.1291,
+      "step": 4132
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8211495397547413e-06,
+      "loss": 1.1485,
+      "step": 4133
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8189487790181736e-06,
+      "loss": 0.9003,
+      "step": 4134
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8167490531438286e-06,
+      "loss": 1.1751,
+      "step": 4135
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8145503628473204e-06,
+      "loss": 1.1466,
+      "step": 4136
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.812352708843928e-06,
+      "loss": 1.1631,
+      "step": 4137
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.810156091848595e-06,
+      "loss": 1.1408,
+      "step": 4138
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8079605125759286e-06,
+      "loss": 1.1125,
+      "step": 4139
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8057659717401948e-06,
+      "loss": 1.153,
+      "step": 4140
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8035724700553215e-06,
+      "loss": 1.2019,
+      "step": 4141
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8013800082349025e-06,
+      "loss": 1.1178,
+      "step": 4142
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7991885869921928e-06,
+      "loss": 0.9733,
+      "step": 4143
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7969982070401054e-06,
+      "loss": 1.1053,
+      "step": 4144
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7948088690912157e-06,
+      "loss": 1.1298,
+      "step": 4145
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7926205738577634e-06,
+      "loss": 1.1663,
+      "step": 4146
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.790433322051644e-06,
+      "loss": 1.1448,
+      "step": 4147
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7882471143844188e-06,
+      "loss": 1.1186,
+      "step": 4148
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7860619515673034e-06,
+      "loss": 1.1735,
+      "step": 4149
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7838778343111807e-06,
+      "loss": 0.9942,
+      "step": 4150
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7816947633265852e-06,
+      "loss": 1.0974,
+      "step": 4151
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7795127393237193e-06,
+      "loss": 1.1922,
+      "step": 4152
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7773317630124377e-06,
+      "loss": 1.1125,
+      "step": 4153
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.77515183510226e-06,
+      "loss": 1.2325,
+      "step": 4154
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7729729563023613e-06,
+      "loss": 1.1915,
+      "step": 4155
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7707951273215746e-06,
+      "loss": 0.7581,
+      "step": 4156
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7686183488683938e-06,
+      "loss": 1.102,
+      "step": 4157
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7664426216509728e-06,
+      "loss": 1.1825,
+      "step": 4158
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7642679463771196e-06,
+      "loss": 1.1576,
+      "step": 4159
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7620943237542982e-06,
+      "loss": 1.1808,
+      "step": 4160
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7599217544896363e-06,
+      "loss": 1.2114,
+      "step": 4161
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7577502392899177e-06,
+      "loss": 1.172,
+      "step": 4162
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7555797788615785e-06,
+      "loss": 0.8295,
+      "step": 4163
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7534103739107144e-06,
+      "loss": 1.1102,
+      "step": 4164
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.751242025143079e-06,
+      "loss": 1.1632,
+      "step": 4165
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7490747332640833e-06,
+      "loss": 1.1458,
+      "step": 4166
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.746908498978791e-06,
+      "loss": 1.1511,
+      "step": 4167
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.744743322991921e-06,
+      "loss": 1.0839,
+      "step": 4168
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7425792060078518e-06,
+      "loss": 1.0693,
+      "step": 4169
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7404161487306176e-06,
+      "loss": 1.1956,
+      "step": 4170
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7382541518639045e-06,
+      "loss": 1.175,
+      "step": 4171
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7360932161110538e-06,
+      "loss": 1.1321,
+      "step": 4172
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7339333421750638e-06,
+      "loss": 1.134,
+      "step": 4173
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7317745307585888e-06,
+      "loss": 1.1374,
+      "step": 4174
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7296167825639325e-06,
+      "loss": 1.2331,
+      "step": 4175
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7274600982930544e-06,
+      "loss": 1.1359,
+      "step": 4176
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.725304478647572e-06,
+      "loss": 1.0928,
+      "step": 4177
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7231499243287503e-06,
+      "loss": 1.1759,
+      "step": 4178
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7209964360375137e-06,
+      "loss": 0.8566,
+      "step": 4179
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7188440144744335e-06,
+      "loss": 1.1237,
+      "step": 4180
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.71669266033974e-06,
+      "loss": 1.1664,
+      "step": 4181
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.714542374333311e-06,
+      "loss": 1.0999,
+      "step": 4182
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7123931571546826e-06,
+      "loss": 1.195,
+      "step": 4183
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7102450095030364e-06,
+      "loss": 1.1345,
+      "step": 4184
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.708097932077213e-06,
+      "loss": 1.1207,
+      "step": 4185
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7059519255756974e-06,
+      "loss": 1.1993,
+      "step": 4186
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7038069906966342e-06,
+      "loss": 1.0809,
+      "step": 4187
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.701663128137812e-06,
+      "loss": 1.1655,
+      "step": 4188
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.699520338596678e-06,
+      "loss": 1.1327,
+      "step": 4189
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.697378622770323e-06,
+      "loss": 1.1691,
+      "step": 4190
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6952379813554914e-06,
+      "loss": 1.0708,
+      "step": 4191
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.693098415048579e-06,
+      "loss": 0.9183,
+      "step": 4192
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6909599245456344e-06,
+      "loss": 1.1941,
+      "step": 4193
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6888225105423505e-06,
+      "loss": 1.1348,
+      "step": 4194
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6866861737340705e-06,
+      "loss": 1.1767,
+      "step": 4195
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6845509148157924e-06,
+      "loss": 1.1478,
+      "step": 4196
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6824167344821607e-06,
+      "loss": 1.1603,
+      "step": 4197
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6802836334274675e-06,
+      "loss": 1.1539,
+      "step": 4198
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.678151612345653e-06,
+      "loss": 1.1539,
+      "step": 4199
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6760206719303107e-06,
+      "loss": 1.0843,
+      "step": 4200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6738908128746806e-06,
+      "loss": 1.1391,
+      "step": 4201
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.671762035871649e-06,
+      "loss": 1.155,
+      "step": 4202
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6696343416137495e-06,
+      "loss": 1.2014,
+      "step": 4203
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6675077307931676e-06,
+      "loss": 1.139,
+      "step": 4204
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6653822041017348e-06,
+      "loss": 0.8693,
+      "step": 4205
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6632577622309282e-06,
+      "loss": 1.162,
+      "step": 4206
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6611344058718714e-06,
+      "loss": 1.245,
+      "step": 4207
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6590121357153377e-06,
+      "loss": 0.9797,
+      "step": 4208
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6568909524517478e-06,
+      "loss": 1.1866,
+      "step": 4209
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.654770856771165e-06,
+      "loss": 1.0834,
+      "step": 4210
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6526518493632988e-06,
+      "loss": 1.1793,
+      "step": 4211
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.65053393091751e-06,
+      "loss": 1.1004,
+      "step": 4212
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6484171021227986e-06,
+      "loss": 1.1934,
+      "step": 4213
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.646301363667815e-06,
+      "loss": 1.0436,
+      "step": 4214
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6441867162408514e-06,
+      "loss": 1.1182,
+      "step": 4215
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6420731605298485e-06,
+      "loss": 1.1842,
+      "step": 4216
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.639960697222388e-06,
+      "loss": 1.1282,
+      "step": 4217
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6378493270057004e-06,
+      "loss": 1.1927,
+      "step": 4218
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6357390505666564e-06,
+      "loss": 1.1839,
+      "step": 4219
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6336298685917752e-06,
+      "loss": 1.1712,
+      "step": 4220
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6315217817672142e-06,
+      "loss": 0.9503,
+      "step": 4221
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6294147907787817e-06,
+      "loss": 1.154,
+      "step": 4222
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6273088963119222e-06,
+      "loss": 1.0514,
+      "step": 4223
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6252040990517303e-06,
+      "loss": 1.1666,
+      "step": 4224
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6231003996829392e-06,
+      "loss": 1.1133,
+      "step": 4225
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6209977988899244e-06,
+      "loss": 1.1292,
+      "step": 4226
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6188962973567068e-06,
+      "loss": 1.1237,
+      "step": 4227
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6167958957669506e-06,
+      "loss": 0.8781,
+      "step": 4228
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6146965948039589e-06,
+      "loss": 1.1817,
+      "step": 4229
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6125983951506758e-06,
+      "loss": 1.0592,
+      "step": 4230
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6105012974896916e-06,
+      "loss": 1.2236,
+      "step": 4231
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.608405302503237e-06,
+      "loss": 1.096,
+      "step": 4232
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.606310410873181e-06,
+      "loss": 1.192,
+      "step": 4233
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6042166232810346e-06,
+      "loss": 0.9827,
+      "step": 4234
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6021239404079513e-06,
+      "loss": 1.1304,
+      "step": 4235
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.600032362934727e-06,
+      "loss": 1.1923,
+      "step": 4236
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5979418915417928e-06,
+      "loss": 1.1263,
+      "step": 4237
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5958525269092217e-06,
+      "loss": 1.1777,
+      "step": 4238
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5937642697167288e-06,
+      "loss": 1.1177,
+      "step": 4239
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5916771206436694e-06,
+      "loss": 1.1855,
+      "step": 4240
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5895910803690351e-06,
+      "loss": 0.8583,
+      "step": 4241
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.587506149571456e-06,
+      "loss": 1.2951,
+      "step": 4242
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5854223289292064e-06,
+      "loss": 1.0943,
+      "step": 4243
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5833396191201967e-06,
+      "loss": 0.9747,
+      "step": 4244
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5812580208219751e-06,
+      "loss": 1.2254,
+      "step": 4245
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5791775347117267e-06,
+      "loss": 1.1381,
+      "step": 4246
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5770981614662812e-06,
+      "loss": 1.1855,
+      "step": 4247
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.575019901762097e-06,
+      "loss": 1.1127,
+      "step": 4248
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5729427562752803e-06,
+      "loss": 1.1876,
+      "step": 4249
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5708667256815657e-06,
+      "loss": 0.8947,
+      "step": 4250
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5687918106563326e-06,
+      "loss": 1.14,
+      "step": 4251
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5667180118745911e-06,
+      "loss": 1.118,
+      "step": 4252
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5646453300109942e-06,
+      "loss": 1.1596,
+      "step": 4253
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.562573765739825e-06,
+      "loss": 1.1281,
+      "step": 4254
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5605033197350095e-06,
+      "loss": 1.1467,
+      "step": 4255
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5584339926701041e-06,
+      "loss": 1.1605,
+      "step": 4256
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5563657852183072e-06,
+      "loss": 0.833,
+      "step": 4257
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5542986980524473e-06,
+      "loss": 1.1682,
+      "step": 4258
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5522327318449926e-06,
+      "loss": 1.1515,
+      "step": 4259
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5501678872680443e-06,
+      "loss": 1.2041,
+      "step": 4260
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.548104164993337e-06,
+      "loss": 1.1019,
+      "step": 4261
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5460415656922455e-06,
+      "loss": 1.1574,
+      "step": 4262
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5439800900357765e-06,
+      "loss": 0.897,
+      "step": 4263
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5419197386945707e-06,
+      "loss": 1.1425,
+      "step": 4264
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5398605123389004e-06,
+      "loss": 1.1466,
+      "step": 4265
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.537802411638677e-06,
+      "loss": 1.094,
+      "step": 4266
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.535745437263445e-06,
+      "loss": 1.1976,
+      "step": 4267
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5336895898823801e-06,
+      "loss": 1.1238,
+      "step": 4268
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5316348701642896e-06,
+      "loss": 1.1647,
+      "step": 4269
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5295812787776192e-06,
+      "loss": 0.9111,
+      "step": 4270
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5275288163904457e-06,
+      "loss": 1.0785,
+      "step": 4271
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5254774836704766e-06,
+      "loss": 1.1415,
+      "step": 4272
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.523427281285052e-06,
+      "loss": 1.1289,
+      "step": 4273
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5213782099011465e-06,
+      "loss": 1.2128,
+      "step": 4274
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5193302701853674e-06,
+      "loss": 1.1039,
+      "step": 4275
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5172834628039507e-06,
+      "loss": 0.8379,
+      "step": 4276
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5152377884227632e-06,
+      "loss": 1.3064,
+      "step": 4277
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.513193247707308e-06,
+      "loss": 1.1209,
+      "step": 4278
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5111498413227182e-06,
+      "loss": 0.8589,
+      "step": 4279
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5091075699337542e-06,
+      "loss": 1.1691,
+      "step": 4280
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5070664342048085e-06,
+      "loss": 1.1698,
+      "step": 4281
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5050264347999078e-06,
+      "loss": 1.0885,
+      "step": 4282
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5029875723827037e-06,
+      "loss": 1.1754,
+      "step": 4283
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.500949847616483e-06,
+      "loss": 1.0492,
+      "step": 4284
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4989132611641576e-06,
+      "loss": 1.1254,
+      "step": 4285
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4968778136882745e-06,
+      "loss": 0.9381,
+      "step": 4286
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4948435058510036e-06,
+      "loss": 1.1433,
+      "step": 4287
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4928103383141513e-06,
+      "loss": 1.1364,
+      "step": 4288
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4907783117391455e-06,
+      "loss": 1.1113,
+      "step": 4289
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4887474267870504e-06,
+      "loss": 1.1835,
+      "step": 4290
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4867176841185544e-06,
+      "loss": 1.0922,
+      "step": 4291
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4846890843939721e-06,
+      "loss": 1.1171,
+      "step": 4292
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4826616282732509e-06,
+      "loss": 1.1438,
+      "step": 4293
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.480635316415967e-06,
+      "loss": 1.0652,
+      "step": 4294
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4786101494813203e-06,
+      "loss": 1.1537,
+      "step": 4295
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4765861281281375e-06,
+      "loss": 1.0891,
+      "step": 4296
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4745632530148763e-06,
+      "loss": 1.1663,
+      "step": 4297
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4725415247996221e-06,
+      "loss": 1.1228,
+      "step": 4298
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4705209441400841e-06,
+      "loss": 0.9727,
+      "step": 4299
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4685015116935968e-06,
+      "loss": 1.145,
+      "step": 4300
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4664832281171264e-06,
+      "loss": 1.0724,
+      "step": 4301
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4644660940672628e-06,
+      "loss": 1.1492,
+      "step": 4302
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4624501102002209e-06,
+      "loss": 1.1479,
+      "step": 4303
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4604352771718405e-06,
+      "loss": 1.1038,
+      "step": 4304
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4584215956375903e-06,
+      "loss": 1.1393,
+      "step": 4305
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.456409066252565e-06,
+      "loss": 0.9109,
+      "step": 4306
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4543976896714796e-06,
+      "loss": 1.0561,
+      "step": 4307
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4523874665486759e-06,
+      "loss": 1.2807,
+      "step": 4308
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.450378397538123e-06,
+      "loss": 1.0971,
+      "step": 4309
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4483704832934143e-06,
+      "loss": 1.1434,
+      "step": 4310
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4463637244677648e-06,
+      "loss": 1.1026,
+      "step": 4311
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4443581217140129e-06,
+      "loss": 1.1592,
+      "step": 4312
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4423536756846251e-06,
+      "loss": 1.1657,
+      "step": 4313
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4403503870316904e-06,
+      "loss": 1.1485,
+      "step": 4314
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4383482564069195e-06,
+      "loss": 0.9525,
+      "step": 4315
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.436347284461645e-06,
+      "loss": 1.1716,
+      "step": 4316
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4343474718468274e-06,
+      "loss": 1.1398,
+      "step": 4317
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4323488192130457e-06,
+      "loss": 1.1321,
+      "step": 4318
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4303513272105057e-06,
+      "loss": 1.1902,
+      "step": 4319
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4283549964890298e-06,
+      "loss": 1.1477,
+      "step": 4320
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.426359827698069e-06,
+      "loss": 0.838,
+      "step": 4321
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4243658214866906e-06,
+      "loss": 1.066,
+      "step": 4322
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.422372978503589e-06,
+      "loss": 1.1914,
+      "step": 4323
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4203812993970745e-06,
+      "loss": 1.2202,
+      "step": 4324
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4183907848150857e-06,
+      "loss": 1.1571,
+      "step": 4325
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4164014354051758e-06,
+      "loss": 1.149,
+      "step": 4326
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4144132518145203e-06,
+      "loss": 1.1172,
+      "step": 4327
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4124262346899192e-06,
+      "loss": 0.848,
+      "step": 4328
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.410440384677791e-06,
+      "loss": 1.1453,
+      "step": 4329
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.408455702424173e-06,
+      "loss": 1.0952,
+      "step": 4330
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4064721885747217e-06,
+      "loss": 1.17,
+      "step": 4331
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4044898437747179e-06,
+      "loss": 1.1119,
+      "step": 4332
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4025086686690603e-06,
+      "loss": 1.1874,
+      "step": 4333
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.400528663902266e-06,
+      "loss": 1.111,
+      "step": 4334
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3985498301184685e-06,
+      "loss": 0.8968,
+      "step": 4335
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.396572167961427e-06,
+      "loss": 1.1709,
+      "step": 4336
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3945956780745162e-06,
+      "loss": 1.0844,
+      "step": 4337
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3926203611007294e-06,
+      "loss": 1.2228,
+      "step": 4338
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3906462176826756e-06,
+      "loss": 1.0872,
+      "step": 4339
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3886732484625864e-06,
+      "loss": 1.1718,
+      "step": 4340
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3867014540823125e-06,
+      "loss": 0.981,
+      "step": 4341
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3847308351833166e-06,
+      "loss": 1.1764,
+      "step": 4342
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3827613924066817e-06,
+      "loss": 1.1605,
+      "step": 4343
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3807931263931095e-06,
+      "loss": 1.0805,
+      "step": 4344
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3788260377829193e-06,
+      "loss": 1.1828,
+      "step": 4345
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3768601272160453e-06,
+      "loss": 1.1895,
+      "step": 4346
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.374895395332037e-06,
+      "loss": 1.1652,
+      "step": 4347
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3729318427700655e-06,
+      "loss": 1.1122,
+      "step": 4348
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3709694701689124e-06,
+      "loss": 1.151,
+      "step": 4349
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3690082781669823e-06,
+      "loss": 1.0914,
+      "step": 4350
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3670482674022871e-06,
+      "loss": 0.9452,
+      "step": 4351
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3650894385124641e-06,
+      "loss": 1.2024,
+      "step": 4352
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3631317921347564e-06,
+      "loss": 1.1679,
+      "step": 4353
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3611753289060314e-06,
+      "loss": 1.1659,
+      "step": 4354
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3592200494627634e-06,
+      "loss": 1.1336,
+      "step": 4355
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3572659544410493e-06,
+      "loss": 1.1517,
+      "step": 4356
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3553130444765933e-06,
+      "loss": 0.8403,
+      "step": 4357
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3533613202047208e-06,
+      "loss": 1.117,
+      "step": 4358
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.351410782260366e-06,
+      "loss": 1.1505,
+      "step": 4359
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3494614312780819e-06,
+      "loss": 1.0904,
+      "step": 4360
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3475132678920311e-06,
+      "loss": 1.1392,
+      "step": 4361
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3455662927359914e-06,
+      "loss": 1.118,
+      "step": 4362
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.343620506443355e-06,
+      "loss": 1.1758,
+      "step": 4363
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3416759096471283e-06,
+      "loss": 0.9781,
+      "step": 4364
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.339732502979928e-06,
+      "loss": 1.1463,
+      "step": 4365
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3377902870739828e-06,
+      "loss": 1.162,
+      "step": 4366
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3358492625611375e-06,
+      "loss": 1.1892,
+      "step": 4367
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.33390943007285e-06,
+      "loss": 1.0784,
+      "step": 4368
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3319707902401858e-06,
+      "loss": 1.216,
+      "step": 4369
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.330033343693824e-06,
+      "loss": 0.829,
+      "step": 4370
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3280970910640573e-06,
+      "loss": 1.0833,
+      "step": 4371
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3261620329807911e-06,
+      "loss": 1.1689,
+      "step": 4372
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3242281700735382e-06,
+      "loss": 1.07,
+      "step": 4373
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.322295502971423e-06,
+      "loss": 1.1857,
+      "step": 4374
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.320364032303184e-06,
+      "loss": 1.1411,
+      "step": 4375
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3184337586971702e-06,
+      "loss": 1.1413,
+      "step": 4376
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3165046827813388e-06,
+      "loss": 0.7996,
+      "step": 4377
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.314576805183256e-06,
+      "loss": 1.0847,
+      "step": 4378
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3126501265301023e-06,
+      "loss": 1.202,
+      "step": 4379
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3107246474486684e-06,
+      "loss": 1.1264,
+      "step": 4380
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.308800368565351e-06,
+      "loss": 1.1681,
+      "step": 4381
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3068772905061565e-06,
+      "loss": 1.1215,
+      "step": 4382
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3049554138967052e-06,
+      "loss": 1.1561,
+      "step": 4383
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3030347393622205e-06,
+      "loss": 1.1872,
+      "step": 4384
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3011152675275418e-06,
+      "loss": 1.1737,
+      "step": 4385
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2991969990171088e-06,
+      "loss": 1.0541,
+      "step": 4386
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.297279934454978e-06,
+      "loss": 1.1602,
+      "step": 4387
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2953640744648072e-06,
+      "loss": 1.1358,
+      "step": 4388
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2934494196698688e-06,
+      "loss": 1.1631,
+      "step": 4389
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.291535970693037e-06,
+      "loss": 1.1588,
+      "step": 4390
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2896237281567991e-06,
+      "loss": 1.2126,
+      "step": 4391
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2877126926832446e-06,
+      "loss": 1.1399,
+      "step": 4392
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2858028648940762e-06,
+      "loss": 0.8519,
+      "step": 4393
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2838942454105978e-06,
+      "loss": 1.1379,
+      "step": 4394
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2819868348537263e-06,
+      "loss": 1.1029,
+      "step": 4395
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2800806338439798e-06,
+      "loss": 1.09,
+      "step": 4396
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2781756430014846e-06,
+      "loss": 1.1685,
+      "step": 4397
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2762718629459746e-06,
+      "loss": 1.0886,
+      "step": 4398
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.274369294296791e-06,
+      "loss": 0.8258,
+      "step": 4399
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2724679376728776e-06,
+      "loss": 1.2223,
+      "step": 4400
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2705677936927841e-06,
+      "loss": 1.1396,
+      "step": 4401
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2686688629746674e-06,
+      "loss": 1.1973,
+      "step": 4402
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2667711461362914e-06,
+      "loss": 1.0669,
+      "step": 4403
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.264874643795021e-06,
+      "loss": 1.1911,
+      "step": 4404
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2629793565678273e-06,
+      "loss": 1.1238,
+      "step": 4405
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2610852850712873e-06,
+      "loss": 0.7773,
+      "step": 4406
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.259192429921584e-06,
+      "loss": 1.1173,
+      "step": 4407
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2573007917345008e-06,
+      "loss": 1.1254,
+      "step": 4408
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2554103711254256e-06,
+      "loss": 1.1761,
+      "step": 4409
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2535211687093535e-06,
+      "loss": 1.2287,
+      "step": 4410
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2516331851008816e-06,
+      "loss": 1.0996,
+      "step": 4411
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2497464209142102e-06,
+      "loss": 1.1905,
+      "step": 4412
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2478608767631412e-06,
+      "loss": 0.9538,
+      "step": 4413
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2459765532610818e-06,
+      "loss": 1.0946,
+      "step": 4414
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2440934510210445e-06,
+      "loss": 1.1802,
+      "step": 4415
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2422115706556397e-06,
+      "loss": 1.1241,
+      "step": 4416
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2403309127770813e-06,
+      "loss": 1.1791,
+      "step": 4417
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2384514779971885e-06,
+      "loss": 1.1968,
+      "step": 4418
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2365732669273778e-06,
+      "loss": 1.0776,
+      "step": 4419
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2346962801786738e-06,
+      "loss": 1.1504,
+      "step": 4420
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2328205183616964e-06,
+      "loss": 1.083,
+      "step": 4421
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2309459820866725e-06,
+      "loss": 0.8189,
+      "step": 4422
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2290726719634255e-06,
+      "loss": 1.1806,
+      "step": 4423
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2272005886013844e-06,
+      "loss": 1.0711,
+      "step": 4424
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2253297326095743e-06,
+      "loss": 1.1627,
+      "step": 4425
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.223460104596627e-06,
+      "loss": 1.0926,
+      "step": 4426
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2215917051707677e-06,
+      "loss": 1.1623,
+      "step": 4427
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2197245349398296e-06,
+      "loss": 0.9091,
+      "step": 4428
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2178585945112381e-06,
+      "loss": 1.0999,
+      "step": 4429
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2159938844920255e-06,
+      "loss": 1.1648,
+      "step": 4430
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2141304054888204e-06,
+      "loss": 1.2145,
+      "step": 4431
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2122681581078488e-06,
+      "loss": 1.1541,
+      "step": 4432
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2104071429549402e-06,
+      "loss": 1.2095,
+      "step": 4433
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2085473606355235e-06,
+      "loss": 1.1032,
+      "step": 4434
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2066888117546227e-06,
+      "loss": 0.8838,
+      "step": 4435
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2048314969168613e-06,
+      "loss": 1.1874,
+      "step": 4436
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.202975416726464e-06,
+      "loss": 1.0417,
+      "step": 4437
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2011205717872538e-06,
+      "loss": 1.1441,
+      "step": 4438
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1992669627026482e-06,
+      "loss": 1.1129,
+      "step": 4439
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1974145900756646e-06,
+      "loss": 1.1744,
+      "step": 4440
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.195563454508919e-06,
+      "loss": 1.1087,
+      "step": 4441
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.193713556604626e-06,
+      "loss": 0.9282,
+      "step": 4442
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1918648969645947e-06,
+      "loss": 1.1796,
+      "step": 4443
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1900174761902306e-06,
+      "loss": 1.0683,
+      "step": 4444
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1881712948825403e-06,
+      "loss": 1.1771,
+      "step": 4445
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1863263536421261e-06,
+      "loss": 1.1453,
+      "step": 4446
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1844826530691845e-06,
+      "loss": 1.2156,
+      "step": 4447
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.182640193763508e-06,
+      "loss": 1.0125,
+      "step": 4448
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1807989763244887e-06,
+      "loss": 1.1488,
+      "step": 4449
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1789590013511143e-06,
+      "loss": 1.1681,
+      "step": 4450
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.177120269441966e-06,
+      "loss": 1.1571,
+      "step": 4451
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.175282781195219e-06,
+      "loss": 1.0994,
+      "step": 4452
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1734465372086507e-06,
+      "loss": 1.1436,
+      "step": 4453
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1716115380796256e-06,
+      "loss": 1.1444,
+      "step": 4454
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1697777844051105e-06,
+      "loss": 1.1223,
+      "step": 4455
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1679452767816606e-06,
+      "loss": 1.1535,
+      "step": 4456
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1661140158054325e-06,
+      "loss": 1.1043,
+      "step": 4457
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1642840020721695e-06,
+      "loss": 0.8773,
+      "step": 4458
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1624552361772168e-06,
+      "loss": 1.1265,
+      "step": 4459
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1606277187155062e-06,
+      "loss": 1.1137,
+      "step": 4460
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1588014502815715e-06,
+      "loss": 1.2267,
+      "step": 4461
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1569764314695337e-06,
+      "loss": 1.1556,
+      "step": 4462
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1551526628731074e-06,
+      "loss": 1.1756,
+      "step": 4463
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1533301450856054e-06,
+      "loss": 0.8171,
+      "step": 4464
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1515088786999306e-06,
+      "loss": 1.1475,
+      "step": 4465
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1496888643085786e-06,
+      "loss": 1.1472,
+      "step": 4466
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1478701025036359e-06,
+      "loss": 1.159,
+      "step": 4467
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1460525938767852e-06,
+      "loss": 1.1674,
+      "step": 4468
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1442363390193011e-06,
+      "loss": 1.1448,
+      "step": 4469
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1424213385220484e-06,
+      "loss": 1.0973,
+      "step": 4470
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1406075929754823e-06,
+      "loss": 0.9671,
+      "step": 4471
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1387951029696543e-06,
+      "loss": 1.1999,
+      "step": 4472
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1369838690942059e-06,
+      "loss": 1.2112,
+      "step": 4473
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1351738919383677e-06,
+      "loss": 1.1062,
+      "step": 4474
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1333651720909621e-06,
+      "loss": 1.1424,
+      "step": 4475
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1315577101404046e-06,
+      "loss": 1.1514,
+      "step": 4476
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1297515066747016e-06,
+      "loss": 1.1292,
+      "step": 4477
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1279465622814472e-06,
+      "loss": 1.1464,
+      "step": 4478
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.126142877547826e-06,
+      "loss": 1.1797,
+      "step": 4479
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1243404530606157e-06,
+      "loss": 1.0945,
+      "step": 4480
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1225392894061838e-06,
+      "loss": 1.161,
+      "step": 4481
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1207393871704852e-06,
+      "loss": 1.1387,
+      "step": 4482
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1189407469390635e-06,
+      "loss": 1.1702,
+      "step": 4483
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1171433692970563e-06,
+      "loss": 0.9269,
+      "step": 4484
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1153472548291888e-06,
+      "loss": 1.1063,
+      "step": 4485
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1135524041197732e-06,
+      "loss": 1.1687,
+      "step": 4486
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.11175881775271e-06,
+      "loss": 1.1409,
+      "step": 4487
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1099664963114932e-06,
+      "loss": 1.0879,
+      "step": 4488
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1081754403792e-06,
+      "loss": 1.1072,
+      "step": 4489
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1063856505385007e-06,
+      "loss": 1.1004,
+      "step": 4490
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1045971273716476e-06,
+      "loss": 1.19,
+      "step": 4491
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.102809871460489e-06,
+      "loss": 1.176,
+      "step": 4492
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1010238833864523e-06,
+      "loss": 1.0988,
+      "step": 4493
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0992391637305594e-06,
+      "loss": 1.1151,
+      "step": 4494
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0974557130734154e-06,
+      "loss": 1.1807,
+      "step": 4495
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.095673531995215e-06,
+      "loss": 1.094,
+      "step": 4496
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0938926210757383e-06,
+      "loss": 1.1694,
+      "step": 4497
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.092112980894352e-06,
+      "loss": 1.1356,
+      "step": 4498
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0903346120300101e-06,
+      "loss": 1.192,
+      "step": 4499
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0885575150612549e-06,
+      "loss": 0.8739,
+      "step": 4500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.086781690566212e-06,
+      "loss": 1.116,
+      "step": 4501
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.085007139122593e-06,
+      "loss": 1.1583,
+      "step": 4502
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.083233861307697e-06,
+      "loss": 1.1685,
+      "step": 4503
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0814618576984104e-06,
+      "loss": 1.1719,
+      "step": 4504
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0796911288712003e-06,
+      "loss": 1.1659,
+      "step": 4505
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0779216754021215e-06,
+      "loss": 0.8033,
+      "step": 4506
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0761534978668153e-06,
+      "loss": 1.1798,
+      "step": 4507
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0743865968405075e-06,
+      "loss": 1.1283,
+      "step": 4508
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.072620972898007e-06,
+      "loss": 1.1936,
+      "step": 4509
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.070856626613706e-06,
+      "loss": 1.0823,
+      "step": 4510
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0690935585615847e-06,
+      "loss": 1.1314,
+      "step": 4511
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.067331769315208e-06,
+      "loss": 1.1174,
+      "step": 4512
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0655712594477203e-06,
+      "loss": 0.8018,
+      "step": 4513
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0638120295318511e-06,
+      "loss": 1.1589,
+      "step": 4514
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.062054080139916e-06,
+      "loss": 1.1846,
+      "step": 4515
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0602974118438137e-06,
+      "loss": 1.1368,
+      "step": 4516
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0585420252150235e-06,
+      "loss": 1.1609,
+      "step": 4517
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0567879208246084e-06,
+      "loss": 1.1334,
+      "step": 4518
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.055035099243218e-06,
+      "loss": 1.1389,
+      "step": 4519
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.053283561041078e-06,
+      "loss": 0.9153,
+      "step": 4520
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0515333067880036e-06,
+      "loss": 1.123,
+      "step": 4521
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0497843370533857e-06,
+      "loss": 1.1837,
+      "step": 4522
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0480366524062041e-06,
+      "loss": 1.1655,
+      "step": 4523
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0462902534150137e-06,
+      "loss": 1.1045,
+      "step": 4524
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.044545140647958e-06,
+      "loss": 1.1396,
+      "step": 4525
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.042801314672755e-06,
+      "loss": 1.1158,
+      "step": 4526
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0410587760567104e-06,
+      "loss": 1.1506,
+      "step": 4527
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0393175253667054e-06,
+      "loss": 1.0818,
+      "step": 4528
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0375775631692087e-06,
+      "loss": 0.8799,
+      "step": 4529
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.035838890030263e-06,
+      "loss": 1.1679,
+      "step": 4530
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.034101506515497e-06,
+      "loss": 1.1003,
+      "step": 4531
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0323654131901172e-06,
+      "loss": 1.1558,
+      "step": 4532
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.03063061061891e-06,
+      "loss": 1.0599,
+      "step": 4533
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0288970993662428e-06,
+      "loss": 1.1701,
+      "step": 4534
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0271648799960655e-06,
+      "loss": 0.8308,
+      "step": 4535
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0254339530719031e-06,
+      "loss": 1.1462,
+      "step": 4536
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0237043191568607e-06,
+      "loss": 1.112,
+      "step": 4537
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0219759788136264e-06,
+      "loss": 1.2177,
+      "step": 4538
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0202489326044663e-06,
+      "loss": 1.0962,
+      "step": 4539
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0185231810912223e-06,
+      "loss": 1.1701,
+      "step": 4540
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0167987248353172e-06,
+      "loss": 1.0545,
+      "step": 4541
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0150755643977534e-06,
+      "loss": 0.8232,
+      "step": 4542
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.013353700339112e-06,
+      "loss": 1.1282,
+      "step": 4543
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.01163313321955e-06,
+      "loss": 1.1417,
+      "step": 4544
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0099138635988026e-06,
+      "loss": 1.1433,
+      "step": 4545
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0081958920361855e-06,
+      "loss": 1.1291,
+      "step": 4546
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0064792190905914e-06,
+      "loss": 1.1757,
+      "step": 4547
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0047638453204888e-06,
+      "loss": 1.1123,
+      "step": 4548
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0030497712839233e-06,
+      "loss": 0.9438,
+      "step": 4549
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0013369975385202e-06,
+      "loss": 1.1677,
+      "step": 4550
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.99625524641481e-07,
+      "loss": 1.0817,
+      "step": 4551
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.979153531495834e-07,
+      "loss": 1.1691,
+      "step": 4552
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.962064836191792e-07,
+      "loss": 1.1232,
+      "step": 4553
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.944989166062031e-07,
+      "loss": 1.1429,
+      "step": 4554
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.92792652666159e-07,
+      "loss": 1.0401,
+      "step": 4555
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.91087692354133e-07,
+      "loss": 1.1426,
+      "step": 4556
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.893840362247809e-07,
+      "loss": 1.1897,
+      "step": 4557
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.87681684832341e-07,
+      "loss": 1.1229,
+      "step": 4558
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.859806387306204e-07,
+      "loss": 1.1533,
+      "step": 4559
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.84280898473008e-07,
+      "loss": 1.1289,
+      "step": 4560
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.825824646124616e-07,
+      "loss": 1.2257,
+      "step": 4561
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.808853377015197e-07,
+      "loss": 1.0643,
+      "step": 4562
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.791895182922911e-07,
+      "loss": 1.153,
+      "step": 4563
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.77495006936463e-07,
+      "loss": 1.1169,
+      "step": 4564
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.758018041852936e-07,
+      "loss": 0.8407,
+      "step": 4565
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.741099105896184e-07,
+      "loss": 1.142,
+      "step": 4566
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.724193266998449e-07,
+      "loss": 1.0975,
+      "step": 4567
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.707300530659542e-07,
+      "loss": 1.1519,
+      "step": 4568
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.690420902375031e-07,
+      "loss": 1.0759,
+      "step": 4569
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.673554387636224e-07,
+      "loss": 1.2355,
+      "step": 4570
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.65670099193014e-07,
+      "loss": 0.9794,
+      "step": 4571
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.639860720739524e-07,
+      "loss": 1.0755,
+      "step": 4572
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.623033579542884e-07,
+      "loss": 1.1617,
+      "step": 4573
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.606219573814447e-07,
+      "loss": 1.1535,
+      "step": 4574
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.589418709024146e-07,
+      "loss": 1.1484,
+      "step": 4575
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.572630990637644e-07,
+      "loss": 1.0883,
+      "step": 4576
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.555856424116345e-07,
+      "loss": 1.1544,
+      "step": 4577
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.539095014917381e-07,
+      "loss": 0.9428,
+      "step": 4578
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.522346768493574e-07,
+      "loss": 1.1708,
+      "step": 4579
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.505611690293459e-07,
+      "loss": 1.1643,
+      "step": 4580
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.488889785761324e-07,
+      "loss": 1.1929,
+      "step": 4581
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.47218106033716e-07,
+      "loss": 1.138,
+      "step": 4582
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.455485519456658e-07,
+      "loss": 1.2249,
+      "step": 4583
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.438803168551214e-07,
+      "loss": 1.0803,
+      "step": 4584
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.422134013047945e-07,
+      "loss": 1.1112,
+      "step": 4585
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.405478058369699e-07,
+      "loss": 1.1667,
+      "step": 4586
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.388835309934985e-07,
+      "loss": 1.1459,
+      "step": 4587
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.372205773158028e-07,
+      "loss": 1.1692,
+      "step": 4588
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.355589453448783e-07,
+      "loss": 1.1036,
+      "step": 4589
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.338986356212865e-07,
+      "loss": 1.1556,
+      "step": 4590
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.322396486851626e-07,
+      "loss": 0.9036,
+      "step": 4591
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.305819850762077e-07,
+      "loss": 1.0943,
+      "step": 4592
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.28925645333697e-07,
+      "loss": 1.3,
+      "step": 4593
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.27270629996469e-07,
+      "loss": 0.8069,
+      "step": 4594
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.256169396029374e-07,
+      "loss": 1.0949,
+      "step": 4595
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.239645746910803e-07,
+      "loss": 1.1521,
+      "step": 4596
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.223135357984481e-07,
+      "loss": 1.1336,
+      "step": 4597
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.206638234621574e-07,
+      "loss": 1.182,
+      "step": 4598
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.190154382188921e-07,
+      "loss": 1.0759,
+      "step": 4599
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.173683806049083e-07,
+      "loss": 1.1216,
+      "step": 4600
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.157226511560285e-07,
+      "loss": 1.0467,
+      "step": 4601
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.140782504076429e-07,
+      "loss": 1.1875,
+      "step": 4602
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.124351788947067e-07,
+      "loss": 1.0891,
+      "step": 4603
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.107934371517474e-07,
+      "loss": 1.2198,
+      "step": 4604
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.091530257128583e-07,
+      "loss": 1.0908,
+      "step": 4605
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.075139451116987e-07,
+      "loss": 1.1422,
+      "step": 4606
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.058761958814949e-07,
+      "loss": 0.8819,
+      "step": 4607
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.042397785550405e-07,
+      "loss": 1.1512,
+      "step": 4608
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.026046936646993e-07,
+      "loss": 1.1855,
+      "step": 4609
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.009709417423957e-07,
+      "loss": 1.1691,
+      "step": 4610
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.993385233196223e-07,
+      "loss": 1.1754,
+      "step": 4611
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.977074389274409e-07,
+      "loss": 1.1381,
+      "step": 4612
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.960776890964773e-07,
+      "loss": 1.0308,
+      "step": 4613
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.944492743569227e-07,
+      "loss": 1.1417,
+      "step": 4614
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.928221952385329e-07,
+      "loss": 1.1049,
+      "step": 4615
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.911964522706318e-07,
+      "loss": 1.2098,
+      "step": 4616
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.895720459821089e-07,
+      "loss": 1.1252,
+      "step": 4617
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.879489769014154e-07,
+      "loss": 1.1619,
+      "step": 4618
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.863272455565691e-07,
+      "loss": 1.1163,
+      "step": 4619
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.847068524751535e-07,
+      "loss": 0.8195,
+      "step": 4620
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.830877981843183e-07,
+      "loss": 1.1156,
+      "step": 4621
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.814700832107736e-07,
+      "loss": 1.1561,
+      "step": 4622
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.79853708080795e-07,
+      "loss": 1.093,
+      "step": 4623
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.782386733202253e-07,
+      "loss": 1.1586,
+      "step": 4624
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.766249794544662e-07,
+      "loss": 1.1172,
+      "step": 4625
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.750126270084891e-07,
+      "loss": 1.1986,
+      "step": 4626
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.734016165068226e-07,
+      "loss": 0.9836,
+      "step": 4627
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.717919484735649e-07,
+      "loss": 1.2432,
+      "step": 4628
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.701836234323718e-07,
+      "loss": 0.8446,
+      "step": 4629
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.685766419064679e-07,
+      "loss": 1.1941,
+      "step": 4630
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.669710044186347e-07,
+      "loss": 1.0707,
+      "step": 4631
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.653667114912223e-07,
+      "loss": 1.1919,
+      "step": 4632
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.637637636461393e-07,
+      "loss": 1.0823,
+      "step": 4633
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.62162161404857e-07,
+      "loss": 1.163,
+      "step": 4634
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.605619052884106e-07,
+      "loss": 1.1083,
+      "step": 4635
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.589629958173989e-07,
+      "loss": 0.9624,
+      "step": 4636
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.57365433511978e-07,
+      "loss": 1.175,
+      "step": 4637
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.557692188918682e-07,
+      "loss": 1.1433,
+      "step": 4638
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.541743524763518e-07,
+      "loss": 1.2079,
+      "step": 4639
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.52580834784274e-07,
+      "loss": 1.0806,
+      "step": 4640
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.509886663340372e-07,
+      "loss": 1.1614,
+      "step": 4641
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.49397847643606e-07,
+      "loss": 0.8408,
+      "step": 4642
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.478083792305087e-07,
+      "loss": 1.1778,
+      "step": 4643
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.462202616118331e-07,
+      "loss": 1.0895,
+      "step": 4644
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.446334953042256e-07,
+      "loss": 1.2065,
+      "step": 4645
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.430480808238939e-07,
+      "loss": 1.1347,
+      "step": 4646
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.414640186866063e-07,
+      "loss": 1.2396,
+      "step": 4647
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.398813094076935e-07,
+      "loss": 1.1051,
+      "step": 4648
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.382999535020425e-07,
+      "loss": 0.9345,
+      "step": 4649
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.367199514840996e-07,
+      "loss": 1.181,
+      "step": 4650
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.35141303867874e-07,
+      "loss": 1.0756,
+      "step": 4651
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.335640111669335e-07,
+      "loss": 1.1443,
+      "step": 4652
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.31988073894403e-07,
+      "loss": 1.1604,
+      "step": 4653
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.304134925629676e-07,
+      "loss": 1.1638,
+      "step": 4654
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.288402676848723e-07,
+      "loss": 1.1445,
+      "step": 4655
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.272683997719183e-07,
+      "loss": 0.9327,
+      "step": 4656
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.256978893354689e-07,
+      "loss": 1.188,
+      "step": 4657
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.241287368864426e-07,
+      "loss": 1.0811,
+      "step": 4658
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.225609429353187e-07,
+      "loss": 1.2273,
+      "step": 4659
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.209945079921306e-07,
+      "loss": 1.0994,
+      "step": 4660
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.194294325664759e-07,
+      "loss": 1.1148,
+      "step": 4661
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.178657171675025e-07,
+      "loss": 1.0662,
+      "step": 4662
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.16303362303923e-07,
+      "loss": 1.309,
+      "step": 4663
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.147423684840012e-07,
+      "loss": 1.0936,
+      "step": 4664
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.131827362155625e-07,
+      "loss": 0.8787,
+      "step": 4665
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.116244660059868e-07,
+      "loss": 1.2218,
+      "step": 4666
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.100675583622137e-07,
+      "loss": 1.1207,
+      "step": 4667
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.08512013790736e-07,
+      "loss": 1.1237,
+      "step": 4668
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.069578327976035e-07,
+      "loss": 1.1122,
+      "step": 4669
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.054050158884258e-07,
+      "loss": 1.1266,
+      "step": 4670
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.03853563568367e-07,
+      "loss": 1.1768,
+      "step": 4671
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.023034763421461e-07,
+      "loss": 0.8151,
+      "step": 4672
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.00754754714037e-07,
+      "loss": 1.1636,
+      "step": 4673
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.992073991878735e-07,
+      "loss": 1.0985,
+      "step": 4674
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.976614102670422e-07,
+      "loss": 1.1307,
+      "step": 4675
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.961167884544852e-07,
+      "loss": 1.1313,
+      "step": 4676
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.94573534252699e-07,
+      "loss": 1.2102,
+      "step": 4677
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.930316481637368e-07,
+      "loss": 1.118,
+      "step": 4678
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.914911306892081e-07,
+      "loss": 1.0987,
+      "step": 4679
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.899519823302743e-07,
+      "loss": 1.1323,
+      "step": 4680
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.884142035876507e-07,
+      "loss": 1.1626,
+      "step": 4681
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.868777949616107e-07,
+      "loss": 1.1665,
+      "step": 4682
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.8534275695198e-07,
+      "loss": 1.1536,
+      "step": 4683
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.838090900581385e-07,
+      "loss": 1.1629,
+      "step": 4684
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.822767947790178e-07,
+      "loss": 1.0075,
+      "step": 4685
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.807458716131072e-07,
+      "loss": 1.1321,
+      "step": 4686
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.792163210584491e-07,
+      "loss": 1.0558,
+      "step": 4687
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.776881436126371e-07,
+      "loss": 1.1438,
+      "step": 4688
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.761613397728174e-07,
+      "loss": 1.1099,
+      "step": 4689
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.746359100356937e-07,
+      "loss": 1.1381,
+      "step": 4690
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.731118548975181e-07,
+      "loss": 0.8867,
+      "step": 4691
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.715891748540994e-07,
+      "loss": 1.0945,
+      "step": 4692
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.700678704007947e-07,
+      "loss": 1.2202,
+      "step": 4693
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.685479420325187e-07,
+      "loss": 1.0957,
+      "step": 4694
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.670293902437331e-07,
+      "loss": 1.1485,
+      "step": 4695
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.655122155284573e-07,
+      "loss": 1.1319,
+      "step": 4696
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.639964183802573e-07,
+      "loss": 1.1806,
+      "step": 4697
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.62481999292255e-07,
+      "loss": 1.1276,
+      "step": 4698
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.609689587571212e-07,
+      "loss": 1.0676,
+      "step": 4699
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.594572972670817e-07,
+      "loss": 1.1568,
+      "step": 4700
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.579470153139085e-07,
+      "loss": 0.8967,
+      "step": 4701
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.564381133889309e-07,
+      "loss": 1.1379,
+      "step": 4702
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.549305919830241e-07,
+      "loss": 1.1825,
+      "step": 4703
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.534244515866157e-07,
+      "loss": 1.1173,
+      "step": 4704
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.519196926896855e-07,
+      "loss": 1.1811,
+      "step": 4705
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.50416315781764e-07,
+      "loss": 1.1174,
+      "step": 4706
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.489143213519301e-07,
+      "loss": 0.9649,
+      "step": 4707
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.474137098888123e-07,
+      "loss": 1.0633,
+      "step": 4708
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.459144818805924e-07,
+      "loss": 1.2772,
+      "step": 4709
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.444166378150014e-07,
+      "loss": 1.1128,
+      "step": 4710
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.429201781793183e-07,
+      "loss": 1.2005,
+      "step": 4711
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.414251034603709e-07,
+      "loss": 1.1541,
+      "step": 4712
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.399314141445396e-07,
+      "loss": 1.1506,
+      "step": 4713
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.384391107177541e-07,
+      "loss": 0.8969,
+      "step": 4714
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.369481936654899e-07,
+      "loss": 1.1198,
+      "step": 4715
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.354586634727729e-07,
+      "loss": 1.1827,
+      "step": 4716
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.339705206241793e-07,
+      "loss": 1.1655,
+      "step": 4717
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.324837656038336e-07,
+      "loss": 1.1718,
+      "step": 4718
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.309983988954078e-07,
+      "loss": 1.1667,
+      "step": 4719
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.295144209821209e-07,
+      "loss": 0.7971,
+      "step": 4720
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.280318323467433e-07,
+      "loss": 1.1699,
+      "step": 4721
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.265506334715927e-07,
+      "loss": 1.0852,
+      "step": 4722
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.250708248385335e-07,
+      "loss": 1.1586,
+      "step": 4723
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.235924069289768e-07,
+      "loss": 1.2461,
+      "step": 4724
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.221153802238845e-07,
+      "loss": 1.2591,
+      "step": 4725
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.206397452037628e-07,
+      "loss": 1.1018,
+      "step": 4726
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.191655023486682e-07,
+      "loss": 0.7792,
+      "step": 4727
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.176926521382005e-07,
+      "loss": 1.114,
+      "step": 4728
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.162211950515102e-07,
+      "loss": 1.1541,
+      "step": 4729
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.147511315672917e-07,
+      "loss": 1.0975,
+      "step": 4730
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.132824621637891e-07,
+      "loss": 1.2042,
+      "step": 4731
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.118151873187889e-07,
+      "loss": 1.1015,
+      "step": 4732
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.103493075096285e-07,
+      "loss": 1.1932,
+      "step": 4733
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.088848232131862e-07,
+      "loss": 1.145,
+      "step": 4734
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.074217349058926e-07,
+      "loss": 1.0577,
+      "step": 4735
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.059600430637175e-07,
+      "loss": 0.785,
+      "step": 4736
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.044997481621829e-07,
+      "loss": 1.1752,
+      "step": 4737
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.030408506763514e-07,
+      "loss": 1.0797,
+      "step": 4738
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.015833510808323e-07,
+      "loss": 1.2386,
+      "step": 4739
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.001272498497808e-07,
+      "loss": 1.0989,
+      "step": 4740
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.986725474568984e-07,
+      "loss": 1.2201,
+      "step": 4741
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.972192443754288e-07,
+      "loss": 1.1771,
+      "step": 4742
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.957673410781617e-07,
+      "loss": 0.9026,
+      "step": 4743
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.94316838037431e-07,
+      "loss": 1.1825,
+      "step": 4744
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.928677357251179e-07,
+      "loss": 1.1111,
+      "step": 4745
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.914200346126443e-07,
+      "loss": 1.1552,
+      "step": 4746
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.899737351709757e-07,
+      "loss": 1.1475,
+      "step": 4747
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.885288378706251e-07,
+      "loss": 1.1307,
+      "step": 4748
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.870853431816488e-07,
+      "loss": 0.8556,
+      "step": 4749
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.856432515736439e-07,
+      "loss": 1.172,
+      "step": 4750
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.842025635157529e-07,
+      "loss": 1.1937,
+      "step": 4751
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.82763279476662e-07,
+      "loss": 1.193,
+      "step": 4752
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.813253999246016e-07,
+      "loss": 1.0926,
+      "step": 4753
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.798889253273422e-07,
+      "loss": 1.1783,
+      "step": 4754
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.784538561521986e-07,
+      "loss": 1.0723,
+      "step": 4755
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.770201928660297e-07,
+      "loss": 1.0795,
+      "step": 4756
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.755879359352368e-07,
+      "loss": 1.1764,
+      "step": 4757
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.741570858257617e-07,
+      "loss": 1.0989,
+      "step": 4758
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.727276430030887e-07,
+      "loss": 1.1568,
+      "step": 4759
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.712996079322482e-07,
+      "loss": 1.0922,
+      "step": 4760
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.698729810778065e-07,
+      "loss": 1.176,
+      "step": 4761
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.684477629038782e-07,
+      "loss": 1.0963,
+      "step": 4762
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.670239538741135e-07,
+      "loss": 0.8456,
+      "step": 4763
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.656015544517102e-07,
+      "loss": 1.162,
+      "step": 4764
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.64180565099401e-07,
+      "loss": 1.133,
+      "step": 4765
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.62760986279467e-07,
+      "loss": 1.1162,
+      "step": 4766
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.613428184537235e-07,
+      "loss": 1.0983,
+      "step": 4767
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.599260620835324e-07,
+      "loss": 1.1012,
+      "step": 4768
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.585107176297928e-07,
+      "loss": 1.1934,
+      "step": 4769
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.570967855529453e-07,
+      "loss": 1.1874,
+      "step": 4770
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.556842663129726e-07,
+      "loss": 1.142,
+      "step": 4771
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.54273160369397e-07,
+      "loss": 0.8807,
+      "step": 4772
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.528634681812795e-07,
+      "loss": 1.1568,
+      "step": 4773
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.51455190207222e-07,
+      "loss": 1.0807,
+      "step": 4774
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.50048326905367e-07,
+      "loss": 1.1997,
+      "step": 4775
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.486428787333987e-07,
+      "loss": 1.146,
+      "step": 4776
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.472388461485357e-07,
+      "loss": 1.133,
+      "step": 4777
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.458362296075399e-07,
+      "loss": 1.1406,
+      "step": 4778
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.444350295667112e-07,
+      "loss": 0.878,
+      "step": 4779
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.430352464818917e-07,
+      "loss": 1.1459,
+      "step": 4780
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.416368808084578e-07,
+      "loss": 1.1498,
+      "step": 4781
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.40239933001327e-07,
+      "loss": 1.1606,
+      "step": 4782
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.388444035149538e-07,
+      "loss": 1.1535,
+      "step": 4783
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.374502928033371e-07,
+      "loss": 1.1629,
+      "step": 4784
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.360576013200087e-07,
+      "loss": 1.1258,
+      "step": 4785
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.346663295180377e-07,
+      "loss": 1.1675,
+      "step": 4786
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.33276477850036e-07,
+      "loss": 1.1447,
+      "step": 4787
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.318880467681527e-07,
+      "loss": 1.0878,
+      "step": 4788
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.305010367240704e-07,
+      "loss": 1.2254,
+      "step": 4789
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.291154481690131e-07,
+      "loss": 1.149,
+      "step": 4790
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.277312815537423e-07,
+      "loss": 1.164,
+      "step": 4791
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.263485373285566e-07,
+      "loss": 0.8654,
+      "step": 4792
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.249672159432913e-07,
+      "loss": 1.1572,
+      "step": 4793
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.235873178473167e-07,
+      "loss": 1.092,
+      "step": 4794
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.222088434895462e-07,
+      "loss": 1.1507,
+      "step": 4795
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.208317933184233e-07,
+      "loss": 1.1159,
+      "step": 4796
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.194561677819327e-07,
+      "loss": 1.1589,
+      "step": 4797
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.180819673275928e-07,
+      "loss": 0.9287,
+      "step": 4798
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.16709192402461e-07,
+      "loss": 1.0992,
+      "step": 4799
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.153378434531282e-07,
+      "loss": 1.1378,
+      "step": 4800
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.139679209257243e-07,
+      "loss": 1.0719,
+      "step": 4801
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.125994252659118e-07,
+      "loss": 1.1475,
+      "step": 4802
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.112323569188927e-07,
+      "loss": 1.1361,
+      "step": 4803
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.09866716329402e-07,
+      "loss": 1.1126,
+      "step": 4804
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.085025039417097e-07,
+      "loss": 1.1189,
+      "step": 4805
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.071397201996243e-07,
+      "loss": 1.1155,
+      "step": 4806
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.057783655464883e-07,
+      "loss": 1.1869,
+      "step": 4807
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.044184404251768e-07,
+      "loss": 0.9291,
+      "step": 4808
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.030599452781022e-07,
+      "loss": 1.1361,
+      "step": 4809
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.017028805472119e-07,
+      "loss": 1.1779,
+      "step": 4810
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.003472466739879e-07,
+      "loss": 1.1647,
+      "step": 4811
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.989930440994451e-07,
+      "loss": 1.1326,
+      "step": 4812
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.976402732641329e-07,
+      "loss": 1.1774,
+      "step": 4813
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.962889346081375e-07,
+      "loss": 0.8132,
+      "step": 4814
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.949390285710777e-07,
+      "loss": 1.1015,
+      "step": 4815
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.935905555921051e-07,
+      "loss": 1.2363,
+      "step": 4816
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.922435161099066e-07,
+      "loss": 1.0567,
+      "step": 4817
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.908979105626999e-07,
+      "loss": 1.2081,
+      "step": 4818
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.895537393882423e-07,
+      "loss": 1.0307,
+      "step": 4819
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.882110030238191e-07,
+      "loss": 1.122,
+      "step": 4820
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.868697019062497e-07,
+      "loss": 0.8841,
+      "step": 4821
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.855298364718858e-07,
+      "loss": 1.1064,
+      "step": 4822
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.841914071566179e-07,
+      "loss": 1.1863,
+      "step": 4823
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.828544143958625e-07,
+      "loss": 1.1481,
+      "step": 4824
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.815188586245701e-07,
+      "loss": 1.1722,
+      "step": 4825
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.801847402772271e-07,
+      "loss": 1.1307,
+      "step": 4826
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.788520597878477e-07,
+      "loss": 1.0593,
+      "step": 4827
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.775208175899827e-07,
+      "loss": 1.1188,
+      "step": 4828
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.76191014116711e-07,
+      "loss": 1.0896,
+      "step": 4829
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.748626498006477e-07,
+      "loss": 1.1486,
+      "step": 4830
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.735357250739343e-07,
+      "loss": 1.1477,
+      "step": 4831
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.7221024036825e-07,
+      "loss": 1.1819,
+      "step": 4832
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.708861961148004e-07,
+      "loss": 1.2538,
+      "step": 4833
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.695635927443271e-07,
+      "loss": 0.824,
+      "step": 4834
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.682424306870977e-07,
+      "loss": 1.1306,
+      "step": 4835
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.669227103729158e-07,
+      "loss": 1.1737,
+      "step": 4836
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.656044322311122e-07,
+      "loss": 1.1229,
+      "step": 4837
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.642875966905526e-07,
+      "loss": 1.1203,
+      "step": 4838
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.629722041796292e-07,
+      "loss": 1.1978,
+      "step": 4839
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.616582551262667e-07,
+      "loss": 1.0698,
+      "step": 4840
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.603457499579201e-07,
+      "loss": 1.1994,
+      "step": 4841
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.590346891015758e-07,
+      "loss": 1.1213,
+      "step": 4842
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.577250729837486e-07,
+      "loss": 0.8887,
+      "step": 4843
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.564169020304827e-07,
+      "loss": 1.1164,
+      "step": 4844
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.551101766673544e-07,
+      "loss": 1.1433,
+      "step": 4845
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.538048973194699e-07,
+      "loss": 1.171,
+      "step": 4846
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.52501064411462e-07,
+      "loss": 1.1373,
+      "step": 4847
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.511986783674944e-07,
+      "loss": 1.214,
+      "step": 4848
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.498977396112615e-07,
+      "loss": 1.1111,
+      "step": 4849
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.485982485659857e-07,
+      "loss": 0.7952,
+      "step": 4850
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.473002056544191e-07,
+      "loss": 1.2075,
+      "step": 4851
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.46003611298841e-07,
+      "loss": 1.1962,
+      "step": 4852
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.447084659210588e-07,
+      "loss": 1.1372,
+      "step": 4853
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.434147699424147e-07,
+      "loss": 1.1816,
+      "step": 4854
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.421225237837724e-07,
+      "loss": 1.1468,
+      "step": 4855
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.408317278655262e-07,
+      "loss": 0.8747,
+      "step": 4856
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.39542382607598e-07,
+      "loss": 1.1605,
+      "step": 4857
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.382544884294416e-07,
+      "loss": 1.0978,
+      "step": 4858
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.369680457500343e-07,
+      "loss": 1.1381,
+      "step": 4859
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.356830549878817e-07,
+      "loss": 1.1249,
+      "step": 4860
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.343995165610194e-07,
+      "loss": 1.1443,
+      "step": 4861
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.331174308870085e-07,
+      "loss": 1.1361,
+      "step": 4862
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.318367983829393e-07,
+      "loss": 1.1229,
+      "step": 4863
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.30557619465426e-07,
+      "loss": 1.1531,
+      "step": 4864
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.29279894550615e-07,
+      "loss": 1.1112,
+      "step": 4865
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.280036240541742e-07,
+      "loss": 1.1731,
+      "step": 4866
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.26728808391303e-07,
+      "loss": 1.1457,
+      "step": 4867
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.254554479767238e-07,
+      "loss": 1.2218,
+      "step": 4868
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.241835432246888e-07,
+      "loss": 1.1005,
+      "step": 4869
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.229130945489736e-07,
+      "loss": 0.8391,
+      "step": 4870
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.216441023628832e-07,
+      "loss": 1.1666,
+      "step": 4871
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.203765670792455e-07,
+      "loss": 1.0835,
+      "step": 4872
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.191104891104181e-07,
+      "loss": 1.1441,
+      "step": 4873
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.178458688682814e-07,
+      "loss": 1.1592,
+      "step": 4874
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.165827067642415e-07,
+      "loss": 1.1397,
+      "step": 4875
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.153210032092326e-07,
+      "loss": 1.1603,
+      "step": 4876
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.140607586137147e-07,
+      "loss": 1.1595,
+      "step": 4877
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.128019733876699e-07,
+      "loss": 1.0546,
+      "step": 4878
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.115446479406061e-07,
+      "loss": 1.1482,
+      "step": 4879
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.102887826815589e-07,
+      "loss": 1.1481,
+      "step": 4880
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.090343780190882e-07,
+      "loss": 1.1358,
+      "step": 4881
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.077814343612775e-07,
+      "loss": 1.1549,
+      "step": 4882
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.065299521157346e-07,
+      "loss": 1.0853,
+      "step": 4883
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.052799316895912e-07,
+      "loss": 1.1662,
+      "step": 4884
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.040313734895092e-07,
+      "loss": 1.1063,
+      "step": 4885
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.02784277921668e-07,
+      "loss": 0.889,
+      "step": 4886
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.015386453917742e-07,
+      "loss": 1.155,
+      "step": 4887
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.002944763050566e-07,
+      "loss": 1.1029,
+      "step": 4888
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.990517710662718e-07,
+      "loss": 1.1371,
+      "step": 4889
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.97810530079697e-07,
+      "loss": 1.1051,
+      "step": 4890
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.965707537491327e-07,
+      "loss": 1.1637,
+      "step": 4891
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.953324424779033e-07,
+      "loss": 0.8387,
+      "step": 4892
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.940955966688599e-07,
+      "loss": 1.1355,
+      "step": 4893
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.928602167243729e-07,
+      "loss": 1.2215,
+      "step": 4894
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.916263030463375e-07,
+      "loss": 1.1027,
+      "step": 4895
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.903938560361698e-07,
+      "loss": 1.1819,
+      "step": 4896
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.891628760948114e-07,
+      "loss": 1.1143,
+      "step": 4897
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.879333636227273e-07,
+      "loss": 1.1142,
+      "step": 4898
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.867053190199011e-07,
+      "loss": 0.9152,
+      "step": 4899
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.854787426858438e-07,
+      "loss": 1.177,
+      "step": 4900
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.842536350195836e-07,
+      "loss": 1.0946,
+      "step": 4901
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.830299964196761e-07,
+      "loss": 1.1228,
+      "step": 4902
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.818078272841942e-07,
+      "loss": 1.0942,
+      "step": 4903
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.805871280107366e-07,
+      "loss": 1.1429,
+      "step": 4904
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.793678989964207e-07,
+      "loss": 0.9239,
+      "step": 4905
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.781501406378886e-07,
+      "loss": 1.085,
+      "step": 4906
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.769338533313012e-07,
+      "loss": 1.1348,
+      "step": 4907
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.757190374723436e-07,
+      "loss": 1.1608,
+      "step": 4908
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.745056934562192e-07,
+      "loss": 1.1098,
+      "step": 4909
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.732938216776545e-07,
+      "loss": 1.135,
+      "step": 4910
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.720834225308962e-07,
+      "loss": 1.1496,
+      "step": 4911
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.708744964097145e-07,
+      "loss": 1.1243,
+      "step": 4912
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.696670437073958e-07,
+      "loss": 1.0662,
+      "step": 4913
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6846106481675035e-07,
+      "loss": 1.1274,
+      "step": 4914
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.672565601301082e-07,
+      "loss": 0.8897,
+      "step": 4915
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.660535300393204e-07,
+      "loss": 1.1351,
+      "step": 4916
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6485197493575774e-07,
+      "loss": 1.1367,
+      "step": 4917
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6365189521031086e-07,
+      "loss": 1.1484,
+      "step": 4918
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.62453291253388e-07,
+      "loss": 1.1154,
+      "step": 4919
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.612561634549251e-07,
+      "loss": 1.1679,
+      "step": 4920
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.600605122043694e-07,
+      "loss": 0.8335,
+      "step": 4921
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5886633789069213e-07,
+      "loss": 1.1281,
+      "step": 4922
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.576736409023813e-07,
+      "loss": 1.1458,
+      "step": 4923
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5648242162744883e-07,
+      "loss": 1.0944,
+      "step": 4924
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5529268045342243e-07,
+      "loss": 1.198,
+      "step": 4925
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.541044177673493e-07,
+      "loss": 1.0895,
+      "step": 4926
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5291763395579393e-07,
+      "loss": 1.1658,
+      "step": 4927
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5173232940484534e-07,
+      "loss": 0.8956,
+      "step": 4928
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.505485045001068e-07,
+      "loss": 1.0522,
+      "step": 4929
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.493661596267007e-07,
+      "loss": 1.1878,
+      "step": 4930
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.481852951692672e-07,
+      "loss": 1.1658,
+      "step": 4931
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.470059115119674e-07,
+      "loss": 1.1882,
+      "step": 4932
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4582800903848033e-07,
+      "loss": 1.081,
+      "step": 4933
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.446515881320013e-07,
+      "loss": 0.8967,
+      "step": 4934
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4347664917524293e-07,
+      "loss": 1.1514,
+      "step": 4935
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.423031925504395e-07,
+      "loss": 1.1255,
+      "step": 4936
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4113121863934074e-07,
+      "loss": 1.1927,
+      "step": 4937
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.399607278232132e-07,
+      "loss": 1.0571,
+      "step": 4938
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.387917204828429e-07,
+      "loss": 1.1484,
+      "step": 4939
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.376241969985318e-07,
+      "loss": 1.2216,
+      "step": 4940
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.364581577500987e-07,
+      "loss": 0.8569,
+      "step": 4941
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3529360311688083e-07,
+      "loss": 1.1049,
+      "step": 4942
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3413053347773406e-07,
+      "loss": 1.2841,
+      "step": 4943
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.329689492110278e-07,
+      "loss": 0.8878,
+      "step": 4944
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.318088506946483e-07,
+      "loss": 1.0997,
+      "step": 4945
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3065023830600074e-07,
+      "loss": 1.1516,
+      "step": 4946
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.29493112422007e-07,
+      "loss": 1.0589,
+      "step": 4947
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.283374734191037e-07,
+      "loss": 1.1628,
+      "step": 4948
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2718332167324293e-07,
+      "loss": 1.1264,
+      "step": 4949
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2603065755989493e-07,
+      "loss": 0.9303,
+      "step": 4950
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2487948145404635e-07,
+      "loss": 1.1767,
+      "step": 4951
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.237297937301987e-07,
+      "loss": 1.0852,
+      "step": 4952
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.225815947623685e-07,
+      "loss": 1.1441,
+      "step": 4953
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.214348849240868e-07,
+      "loss": 1.1272,
+      "step": 4954
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.202896645884064e-07,
+      "loss": 1.166,
+      "step": 4955
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.191459341278897e-07,
+      "loss": 1.0973,
+      "step": 4956
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.180036939146154e-07,
+      "loss": 1.1446,
+      "step": 4957
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1686294432017605e-07,
+      "loss": 1.1799,
+      "step": 4958
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.15723685715686e-07,
+      "loss": 1.1324,
+      "step": 4959
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1458591847176657e-07,
+      "loss": 1.1681,
+      "step": 4960
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.134496429585583e-07,
+      "loss": 1.1498,
+      "step": 4961
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1231485954571327e-07,
+      "loss": 1.1269,
+      "step": 4962
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1118156860240346e-07,
+      "loss": 0.8395,
+      "step": 4963
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1004977049731e-07,
+      "loss": 1.1756,
+      "step": 4964
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.089194655986306e-07,
+      "loss": 1.1119,
+      "step": 4965
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.077906542740767e-07,
+      "loss": 1.1848,
+      "step": 4966
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.06663336890874e-07,
+      "loss": 1.1229,
+      "step": 4967
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0553751381576414e-07,
+      "loss": 1.1415,
+      "step": 4968
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.044131854149991e-07,
+      "loss": 1.1437,
+      "step": 4969
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.032903520543463e-07,
+      "loss": 0.8781,
+      "step": 4970
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0216901409908695e-07,
+      "loss": 1.2509,
+      "step": 4971
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.010491719140169e-07,
+      "loss": 1.1457,
+      "step": 4972
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9993082586344266e-07,
+      "loss": 1.1434,
+      "step": 4973
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9881397631118644e-07,
+      "loss": 1.1218,
+      "step": 4974
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9769862362058266e-07,
+      "loss": 1.1716,
+      "step": 4975
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.965847681544771e-07,
+      "loss": 1.0869,
+      "step": 4976
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9547241027523164e-07,
+      "loss": 0.908,
+      "step": 4977
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.943615503447201e-07,
+      "loss": 1.2708,
+      "step": 4978
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9325218872432736e-07,
+      "loss": 0.8317,
+      "step": 4979
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.921443257749513e-07,
+      "loss": 1.1652,
+      "step": 4980
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.910379618570037e-07,
+      "loss": 1.0786,
+      "step": 4981
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.899330973304083e-07,
+      "loss": 1.1537,
+      "step": 4982
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8882973255459975e-07,
+      "loss": 1.1069,
+      "step": 4983
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8772786788852436e-07,
+      "loss": 1.1439,
+      "step": 4984
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.866275036906436e-07,
+      "loss": 1.088,
+      "step": 4985
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.85528640318929e-07,
+      "loss": 1.076,
+      "step": 4986
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8443127813086287e-07,
+      "loss": 1.1702,
+      "step": 4987
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.833354174834403e-07,
+      "loss": 1.1207,
+      "step": 4988
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.822410587331654e-07,
+      "loss": 1.1659,
+      "step": 4989
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8114820223605965e-07,
+      "loss": 1.1223,
+      "step": 4990
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.800568483476502e-07,
+      "loss": 1.192,
+      "step": 4991
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7896699742297704e-07,
+      "loss": 1.1128,
+      "step": 4992
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7787864981658986e-07,
+      "loss": 0.8331,
+      "step": 4993
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7679180588255424e-07,
+      "loss": 1.1235,
+      "step": 4994
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7570646597444196e-07,
+      "loss": 1.1457,
+      "step": 4995
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.746226304453354e-07,
+      "loss": 1.1824,
+      "step": 4996
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7354029964782877e-07,
+      "loss": 1.1351,
+      "step": 4997
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7245947393402814e-07,
+      "loss": 1.1194,
+      "step": 4998
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7138015365554834e-07,
+      "loss": 0.8797,
+      "step": 4999
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7030233916351467e-07,
+      "loss": 1.0782,
+      "step": 5000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.692260308085616e-07,
+      "loss": 1.1659,
+      "step": 5001
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.681512289408357e-07,
+      "loss": 1.2115,
+      "step": 5002
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.670779339099934e-07,
+      "loss": 1.1413,
+      "step": 5003
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.660061460651981e-07,
+      "loss": 1.1558,
+      "step": 5004
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6493586575512497e-07,
+      "loss": 1.1642,
+      "step": 5005
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6386709332795864e-07,
+      "loss": 0.8763,
+      "step": 5006
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.627998291313939e-07,
+      "loss": 1.1473,
+      "step": 5007
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6173407351263403e-07,
+      "loss": 1.0763,
+      "step": 5008
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.606698268183895e-07,
+      "loss": 1.1546,
+      "step": 5009
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5960708939488443e-07,
+      "loss": 1.0581,
+      "step": 5010
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5854586158784787e-07,
+      "loss": 1.143,
+      "step": 5011
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5748614374251924e-07,
+      "loss": 0.8462,
+      "step": 5012
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.564279362036488e-07,
+      "loss": 1.2446,
+      "step": 5013
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.553712393154912e-07,
+      "loss": 1.1344,
+      "step": 5014
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5431605342181266e-07,
+      "loss": 0.812,
+      "step": 5015
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5326237886588734e-07,
+      "loss": 1.0893,
+      "step": 5016
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.522102159904983e-07,
+      "loss": 1.2225,
+      "step": 5017
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.511595651379351e-07,
+      "loss": 1.0867,
+      "step": 5018
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5011042664999663e-07,
+      "loss": 1.1323,
+      "step": 5019
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.490628008679875e-07,
+      "loss": 1.0882,
+      "step": 5020
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.480166881327257e-07,
+      "loss": 1.1113,
+      "step": 5021
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.469720887845324e-07,
+      "loss": 1.0028,
+      "step": 5022
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.459290031632362e-07,
+      "loss": 1.1605,
+      "step": 5023
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4488743160817406e-07,
+      "loss": 1.1017,
+      "step": 5024
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4384737445819385e-07,
+      "loss": 1.1633,
+      "step": 5025
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4280883205164596e-07,
+      "loss": 1.103,
+      "step": 5026
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4177180472639015e-07,
+      "loss": 1.1375,
+      "step": 5027
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.407362928197916e-07,
+      "loss": 0.9066,
+      "step": 5028
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3970229666872634e-07,
+      "loss": 1.0791,
+      "step": 5029
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.386698166095742e-07,
+      "loss": 1.217,
+      "step": 5030
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3763885297822153e-07,
+      "loss": 1.1283,
+      "step": 5031
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.366094061100622e-07,
+      "loss": 1.1322,
+      "step": 5032
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.355814763399973e-07,
+      "loss": 1.0657,
+      "step": 5033
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.345550640024348e-07,
+      "loss": 1.1852,
+      "step": 5034
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3353016943128644e-07,
+      "loss": 0.9165,
+      "step": 5035
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.325067929599718e-07,
+      "loss": 1.0904,
+      "step": 5036
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3148493492141663e-07,
+      "loss": 1.1864,
+      "step": 5037
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3046459564805453e-07,
+      "loss": 1.0851,
+      "step": 5038
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.294457754718211e-07,
+      "loss": 1.1542,
+      "step": 5039
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.284284747241601e-07,
+      "loss": 1.1664,
+      "step": 5040
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2741269373602003e-07,
+      "loss": 0.8913,
+      "step": 5041
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2639843283785777e-07,
+      "loss": 1.1393,
+      "step": 5042
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2538569235963216e-07,
+      "loss": 1.0789,
+      "step": 5043
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.243744726308079e-07,
+      "loss": 1.1707,
+      "step": 5044
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.233647739803575e-07,
+      "loss": 1.1629,
+      "step": 5045
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.223565967367559e-07,
+      "loss": 1.1289,
+      "step": 5046
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.213499412279836e-07,
+      "loss": 1.171,
+      "step": 5047
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.203448077815291e-07,
+      "loss": 1.2289,
+      "step": 5048
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1934119672438093e-07,
+      "loss": 1.133,
+      "step": 5049
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.183391083830345e-07,
+      "loss": 1.1656,
+      "step": 5050
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.173385430834913e-07,
+      "loss": 0.8993,
+      "step": 5051
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1633950115125655e-07,
+      "loss": 1.1325,
+      "step": 5052
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.15341982911338e-07,
+      "loss": 1.1329,
+      "step": 5053
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.143459886882494e-07,
+      "loss": 1.1214,
+      "step": 5054
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.133515188060077e-07,
+      "loss": 1.1885,
+      "step": 5055
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1235857358813683e-07,
+      "loss": 1.1331,
+      "step": 5056
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.113671533576612e-07,
+      "loss": 0.8017,
+      "step": 5057
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.103772584371106e-07,
+      "loss": 1.1756,
+      "step": 5058
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0938888914851694e-07,
+      "loss": 1.1064,
+      "step": 5059
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0840204581342013e-07,
+      "loss": 1.1751,
+      "step": 5060
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0741672875285953e-07,
+      "loss": 1.0877,
+      "step": 5061
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0643293828737984e-07,
+      "loss": 1.132,
+      "step": 5062
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0545067473702617e-07,
+      "loss": 1.0678,
+      "step": 5063
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.044699384213523e-07,
+      "loss": 1.1293,
+      "step": 5064
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0349072965941194e-07,
+      "loss": 1.186,
+      "step": 5065
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0251304876976085e-07,
+      "loss": 1.1945,
+      "step": 5066
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.015368960704584e-07,
+      "loss": 1.0719,
+      "step": 5067
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.005622718790685e-07,
+      "loss": 1.1914,
+      "step": 5068
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.995891765126574e-07,
+      "loss": 1.1445,
+      "step": 5069
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9861761028779244e-07,
+      "loss": 0.9667,
+      "step": 5070
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9764757352054363e-07,
+      "loss": 1.1252,
+      "step": 5071
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9667906652648517e-07,
+      "loss": 1.2015,
+      "step": 5072
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9571208962069277e-07,
+      "loss": 1.1933,
+      "step": 5073
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.947466431177437e-07,
+      "loss": 1.126,
+      "step": 5074
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9378272733171676e-07,
+      "loss": 1.1385,
+      "step": 5075
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.928203425761961e-07,
+      "loss": 1.0761,
+      "step": 5076
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.918594891642629e-07,
+      "loss": 0.8548,
+      "step": 5077
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.909001674085055e-07,
+      "loss": 1.1607,
+      "step": 5078
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.899423776210092e-07,
+      "loss": 1.0906,
+      "step": 5079
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8898612011336366e-07,
+      "loss": 1.2471,
+      "step": 5080
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8803139519665955e-07,
+      "loss": 1.0562,
+      "step": 5081
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.870782031814889e-07,
+      "loss": 1.1593,
+      "step": 5082
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8612654437794376e-07,
+      "loss": 1.0565,
+      "step": 5083
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8517641909562075e-07,
+      "loss": 1.1191,
+      "step": 5084
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.842278276436128e-07,
+      "loss": 1.1774,
+      "step": 5085
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8328077033051794e-07,
+      "loss": 0.8164,
+      "step": 5086
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.823352474644342e-07,
+      "loss": 1.1925,
+      "step": 5087
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.813912593529594e-07,
+      "loss": 1.0739,
+      "step": 5088
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.804488063031918e-07,
+      "loss": 1.1839,
+      "step": 5089
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.795078886217301e-07,
+      "loss": 1.0805,
+      "step": 5090
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.785685066146776e-07,
+      "loss": 1.1716,
+      "step": 5091
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7763066058763275e-07,
+      "loss": 1.1404,
+      "step": 5092
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.766943508456965e-07,
+      "loss": 0.8652,
+      "step": 5093
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.757595776934691e-07,
+      "loss": 1.2391,
+      "step": 5094
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.748263414350538e-07,
+      "loss": 1.1363,
+      "step": 5095
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7389464237405126e-07,
+      "loss": 1.1992,
+      "step": 5096
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.729644808135612e-07,
+      "loss": 1.0954,
+      "step": 5097
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7203585705618486e-07,
+      "loss": 1.198,
+      "step": 5098
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.711087714040239e-07,
+      "loss": 1.0773,
+      "step": 5099
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.701832241586788e-07,
+      "loss": 0.8101,
+      "step": 5100
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6925921562124867e-07,
+      "loss": 1.1612,
+      "step": 5101
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.683367460923314e-07,
+      "loss": 1.0958,
+      "step": 5102
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6741581587202747e-07,
+      "loss": 1.1947,
+      "step": 5103
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6649642525993447e-07,
+      "loss": 1.0799,
+      "step": 5104
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6557857455514925e-07,
+      "loss": 1.1253,
+      "step": 5105
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.646622640562663e-07,
+      "loss": 0.9656,
+      "step": 5106
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.637474940613821e-07,
+      "loss": 1.0427,
+      "step": 5107
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.628342648680904e-07,
+      "loss": 1.1737,
+      "step": 5108
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.619225767734829e-07,
+      "loss": 1.2224,
+      "step": 5109
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6101243007415065e-07,
+      "loss": 1.0672,
+      "step": 5110
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.601038250661847e-07,
+      "loss": 1.2045,
+      "step": 5111
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5919676204517073e-07,
+      "loss": 1.1772,
+      "step": 5112
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5829124130619777e-07,
+      "loss": 0.8873,
+      "step": 5113
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.57387263143849e-07,
+      "loss": 1.1179,
+      "step": 5114
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5648482785220865e-07,
+      "loss": 1.0961,
+      "step": 5115
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5558393572485575e-07,
+      "loss": 1.1601,
+      "step": 5116
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5468458705487145e-07,
+      "loss": 1.0979,
+      "step": 5117
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5378678213483057e-07,
+      "loss": 1.1637,
+      "step": 5118
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5289052125680945e-07,
+      "loss": 1.1057,
+      "step": 5119
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.519958047123794e-07,
+      "loss": 1.1557,
+      "step": 5120
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.511026327926114e-07,
+      "loss": 1.1835,
+      "step": 5121
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5021100578807136e-07,
+      "loss": 0.8177,
+      "step": 5122
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4932092398882555e-07,
+      "loss": 1.0701,
+      "step": 5123
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.484323876844352e-07,
+      "loss": 1.1687,
+      "step": 5124
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.475453971639591e-07,
+      "loss": 1.1237,
+      "step": 5125
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4665995271595576e-07,
+      "loss": 1.2174,
+      "step": 5126
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4577605462847764e-07,
+      "loss": 1.1028,
+      "step": 5127
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4489370318907523e-07,
+      "loss": 1.1676,
+      "step": 5128
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.440128986847945e-07,
+      "loss": 0.8988,
+      "step": 5129
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.431336414021818e-07,
+      "loss": 1.2007,
+      "step": 5130
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4225593162727656e-07,
+      "loss": 1.0504,
+      "step": 5131
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.413797696456166e-07,
+      "loss": 1.1866,
+      "step": 5132
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.405051557422344e-07,
+      "loss": 1.1026,
+      "step": 5133
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3963209020166133e-07,
+      "loss": 1.1303,
+      "step": 5134
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3876057330792344e-07,
+      "loss": 0.9212,
+      "step": 5135
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.37890605344544e-07,
+      "loss": 1.0818,
+      "step": 5136
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.370221865945399e-07,
+      "loss": 1.1623,
+      "step": 5137
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.361553173404274e-07,
+      "loss": 1.0714,
+      "step": 5138
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3528999786421758e-07,
+      "loss": 1.1253,
+      "step": 5139
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3442622844741582e-07,
+      "loss": 1.1975,
+      "step": 5140
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3356400937102343e-07,
+      "loss": 1.2378,
+      "step": 5141
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3270334091553992e-07,
+      "loss": 0.966,
+      "step": 5142
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3184422336095857e-07,
+      "loss": 1.1818,
+      "step": 5143
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3098665698676747e-07,
+      "loss": 1.1421,
+      "step": 5144
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.301306420719507e-07,
+      "loss": 1.1186,
+      "step": 5145
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.292761788949882e-07,
+      "loss": 1.1668,
+      "step": 5146
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2842326773385436e-07,
+      "loss": 1.1174,
+      "step": 5147
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2757190886602e-07,
+      "loss": 0.8638,
+      "step": 5148
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2672210256844795e-07,
+      "loss": 1.1324,
+      "step": 5149
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2587384911759992e-07,
+      "loss": 1.2152,
+      "step": 5150
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.25027148789429e-07,
+      "loss": 1.0652,
+      "step": 5151
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2418200185938488e-07,
+      "loss": 1.1827,
+      "step": 5152
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.233384086024115e-07,
+      "loss": 1.205,
+      "step": 5153
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2249636929294772e-07,
+      "loss": 1.155,
+      "step": 5154
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2165588420492545e-07,
+      "loss": 1.1753,
+      "step": 5155
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2081695361177325e-07,
+      "loss": 1.1223,
+      "step": 5156
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1997957778641166e-07,
+      "loss": 1.2523,
+      "step": 5157
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1914375700125722e-07,
+      "loss": 0.8872,
+      "step": 5158
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1830949152821968e-07,
+      "loss": 1.1852,
+      "step": 5159
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1747678163870134e-07,
+      "loss": 1.1965,
+      "step": 5160
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.166456276036033e-07,
+      "loss": 1.146,
+      "step": 5161
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1581602969331484e-07,
+      "loss": 1.1852,
+      "step": 5162
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1498798817772281e-07,
+      "loss": 1.1437,
+      "step": 5163
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1416150332620401e-07,
+      "loss": 0.868,
+      "step": 5164
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1333657540763441e-07,
+      "loss": 1.122,
+      "step": 5165
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1251320469037827e-07,
+      "loss": 1.0485,
+      "step": 5166
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1169139144229578e-07,
+      "loss": 1.1372,
+      "step": 5167
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1087113593073915e-07,
+      "loss": 1.111,
+      "step": 5168
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1005243842255552e-07,
+      "loss": 1.1859,
+      "step": 5169
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.092352991840846e-07,
+      "loss": 1.0909,
+      "step": 5170
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0841971848115761e-07,
+      "loss": 1.1094,
+      "step": 5171
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0760569657910012e-07,
+      "loss": 1.0594,
+      "step": 5172
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0679323374273086e-07,
+      "loss": 1.1783,
+      "step": 5173
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0598233023636116e-07,
+      "loss": 1.1183,
+      "step": 5174
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0517298632379445e-07,
+      "loss": 1.1367,
+      "step": 5175
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0436520226832679e-07,
+      "loss": 1.0907,
+      "step": 5176
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.035589783327474e-07,
+      "loss": 0.925,
+      "step": 5177
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.027543147793387e-07,
+      "loss": 1.1975,
+      "step": 5178
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0195121186987354e-07,
+      "loss": 1.1312,
+      "step": 5179
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0114966986561734e-07,
+      "loss": 1.1838,
+      "step": 5180
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0034968902733042e-07,
+      "loss": 1.1547,
+      "step": 5181
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9955126961526073e-07,
+      "loss": 1.183,
+      "step": 5182
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9875441188915324e-07,
+      "loss": 1.0945,
+      "step": 5183
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9795911610824004e-07,
+      "loss": 0.9094,
+      "step": 5184
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9716538253124918e-07,
+      "loss": 1.1761,
+      "step": 5185
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9637321141639743e-07,
+      "loss": 1.1351,
+      "step": 5186
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9558260302139642e-07,
+      "loss": 1.2484,
+      "step": 5187
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9479355760344488e-07,
+      "loss": 1.1525,
+      "step": 5188
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9400607541923798e-07,
+      "loss": 1.1936,
+      "step": 5189
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9322015672495974e-07,
+      "loss": 1.1467,
+      "step": 5190
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9243580177628395e-07,
+      "loss": 1.0886,
+      "step": 5191
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9165301082837984e-07,
+      "loss": 1.1641,
+      "step": 5192
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.908717841359048e-07,
+      "loss": 0.9215,
+      "step": 5193
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.900921219530083e-07,
+      "loss": 1.1312,
+      "step": 5194
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8931402453333025e-07,
+      "loss": 1.1043,
+      "step": 5195
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8853749213000206e-07,
+      "loss": 1.1644,
+      "step": 5196
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8776252499564663e-07,
+      "loss": 1.1012,
+      "step": 5197
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.869891233823762e-07,
+      "loss": 1.14,
+      "step": 5198
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8621728754179392e-07,
+      "loss": 1.0619,
+      "step": 5199
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8544701772499563e-07,
+      "loss": 0.8794,
+      "step": 5200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8467831418256478e-07,
+      "loss": 1.1723,
+      "step": 5201
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.839111771645774e-07,
+      "loss": 1.1729,
+      "step": 5202
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8314560692059836e-07,
+      "loss": 1.1936,
+      "step": 5203
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8238160369968449e-07,
+      "loss": 1.0715,
+      "step": 5204
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8161916775038146e-07,
+      "loss": 1.1836,
+      "step": 5205
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8085829932072584e-07,
+      "loss": 1.0014,
+      "step": 5206
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8009899865824354e-07,
+      "loss": 1.1026,
+      "step": 5207
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7934126600995084e-07,
+      "loss": 1.1188,
+      "step": 5208
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7858510162235497e-07,
+      "loss": 1.1014,
+      "step": 5209
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7783050574145088e-07,
+      "loss": 1.1715,
+      "step": 5210
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.770774786127244e-07,
+      "loss": 1.1186,
+      "step": 5211
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7632602048115122e-07,
+      "loss": 1.1816,
+      "step": 5212
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.755761315911969e-07,
+      "loss": 0.986,
+      "step": 5213
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7482781218681522e-07,
+      "loss": 1.189,
+      "step": 5214
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7408106251144974e-07,
+      "loss": 1.0474,
+      "step": 5215
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7333588280803503e-07,
+      "loss": 1.1731,
+      "step": 5216
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7259227331899165e-07,
+      "loss": 1.0866,
+      "step": 5217
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.718502342862327e-07,
+      "loss": 1.2076,
+      "step": 5218
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7110976595115847e-07,
+      "loss": 1.1262,
+      "step": 5219
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7037086855465902e-07,
+      "loss": 0.7751,
+      "step": 5220
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6963354233711205e-07,
+      "loss": 1.1738,
+      "step": 5221
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6889778753838682e-07,
+      "loss": 1.1472,
+      "step": 5222
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6816360439783797e-07,
+      "loss": 1.1596,
+      "step": 5223
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.674309931543122e-07,
+      "loss": 1.1325,
+      "step": 5224
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6669995404614225e-07,
+      "loss": 1.0683,
+      "step": 5225
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6597048731115005e-07,
+      "loss": 1.1338,
+      "step": 5226
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6524259318664694e-07,
+      "loss": 1.1284,
+      "step": 5227
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6451627190943177e-07,
+      "loss": 1.1707,
+      "step": 5228
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6379152371579277e-07,
+      "loss": 0.7869,
+      "step": 5229
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6306834884150413e-07,
+      "loss": 1.1044,
+      "step": 5230
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6234674752183045e-07,
+      "loss": 1.145,
+      "step": 5231
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.616267199915239e-07,
+      "loss": 1.0801,
+      "step": 5232
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6090826648482494e-07,
+      "loss": 1.2575,
+      "step": 5233
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.601913872354599e-07,
+      "loss": 1.0914,
+      "step": 5234
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5947608247664558e-07,
+      "loss": 1.1758,
+      "step": 5235
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.587623524410853e-07,
+      "loss": 0.814,
+      "step": 5236
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5805019736097105e-07,
+      "loss": 1.185,
+      "step": 5237
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5733961746798032e-07,
+      "loss": 1.1031,
+      "step": 5238
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5663061299327987e-07,
+      "loss": 1.1696,
+      "step": 5239
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5592318416752516e-07,
+      "loss": 1.0988,
+      "step": 5240
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5521733122085602e-07,
+      "loss": 1.1703,
+      "step": 5241
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.54513054382901e-07,
+      "loss": 0.9705,
+      "step": 5242
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.538103538827762e-07,
+      "loss": 1.0649,
+      "step": 5243
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5310922994908605e-07,
+      "loss": 1.1696,
+      "step": 5244
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.524096828099192e-07,
+      "loss": 1.1109,
+      "step": 5245
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.51711712692853e-07,
+      "loss": 1.1284,
+      "step": 5246
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.510153198249531e-07,
+      "loss": 1.0224,
+      "step": 5247
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.503205044327688e-07,
+      "loss": 1.1648,
+      "step": 5248
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4962726674233928e-07,
+      "loss": 1.0797,
+      "step": 5249
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4893560697918806e-07,
+      "loss": 1.1685,
+      "step": 5250
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4824552536832792e-07,
+      "loss": 1.2172,
+      "step": 5251
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4755702213425539e-07,
+      "loss": 1.1096,
+      "step": 5252
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4687009750095627e-07,
+      "loss": 1.1307,
+      "step": 5253
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4618475169190017e-07,
+      "loss": 1.1776,
+      "step": 5254
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4550098493004537e-07,
+      "loss": 0.8857,
+      "step": 5255
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.44818797437834e-07,
+      "loss": 1.136,
+      "step": 5256
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4413818943719792e-07,
+      "loss": 1.1335,
+      "step": 5257
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4345916114955173e-07,
+      "loss": 1.1475,
+      "step": 5258
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4278171279579757e-07,
+      "loss": 1.0878,
+      "step": 5259
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4210584459632414e-07,
+      "loss": 1.1583,
+      "step": 5260
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4143155677100395e-07,
+      "loss": 1.1612,
+      "step": 5261
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.407588495391987e-07,
+      "loss": 1.1666,
+      "step": 5262
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.400877231197534e-07,
+      "loss": 1.1319,
+      "step": 5263
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3941817773099885e-07,
+      "loss": 1.2137,
+      "step": 5264
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3875021359075257e-07,
+      "loss": 0.8012,
+      "step": 5265
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3808383091631738e-07,
+      "loss": 1.1074,
+      "step": 5266
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3741902992448153e-07,
+      "loss": 1.2177,
+      "step": 5267
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3675581083151867e-07,
+      "loss": 1.1381,
+      "step": 5268
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.360941738531868e-07,
+      "loss": 1.1882,
+      "step": 5269
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3543411920473094e-07,
+      "loss": 1.1065,
+      "step": 5270
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3477564710088097e-07,
+      "loss": 0.8157,
+      "step": 5271
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3411875775585115e-07,
+      "loss": 1.1741,
+      "step": 5272
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3346345138334104e-07,
+      "loss": 1.0862,
+      "step": 5273
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.328097281965357e-07,
+      "loss": 1.1169,
+      "step": 5274
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3215758840810556e-07,
+      "loss": 1.1398,
+      "step": 5275
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3150703223020423e-07,
+      "loss": 1.1851,
+      "step": 5276
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.308580598744713e-07,
+      "loss": 1.1111,
+      "step": 5277
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3021067155203183e-07,
+      "loss": 0.8518,
+      "step": 5278
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2956486747349396e-07,
+      "loss": 1.1269,
+      "step": 5279
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2892064784895186e-07,
+      "loss": 1.2483,
+      "step": 5280
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2827801288798235e-07,
+      "loss": 1.1288,
+      "step": 5281
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2763696279964933e-07,
+      "loss": 1.1719,
+      "step": 5282
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2699749779249926e-07,
+      "loss": 1.0801,
+      "step": 5283
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2635961807456298e-07,
+      "loss": 0.8728,
+      "step": 5284
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2572332385335674e-07,
+      "loss": 1.1805,
+      "step": 5285
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.250886153358799e-07,
+      "loss": 1.0863,
+      "step": 5286
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2445549272861623e-07,
+      "loss": 1.1541,
+      "step": 5287
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2382395623753484e-07,
+      "loss": 1.0765,
+      "step": 5288
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2319400606808584e-07,
+      "loss": 1.1667,
+      "step": 5289
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2256564242520642e-07,
+      "loss": 1.1686,
+      "step": 5290
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.219388655133158e-07,
+      "loss": 0.9335,
+      "step": 5291
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2131367553631867e-07,
+      "loss": 1.1158,
+      "step": 5292
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.206900726976007e-07,
+      "loss": 1.1246,
+      "step": 5293
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.20068057200034e-07,
+      "loss": 1.1116,
+      "step": 5294
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1944762924597286e-07,
+      "loss": 1.2108,
+      "step": 5295
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1882878903725525e-07,
+      "loss": 1.0892,
+      "step": 5296
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1821153677520237e-07,
+      "loss": 1.1413,
+      "step": 5297
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1759587266062078e-07,
+      "loss": 1.0595,
+      "step": 5298
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1698179689379751e-07,
+      "loss": 1.1557,
+      "step": 5299
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1636930967450444e-07,
+      "loss": 0.9087,
+      "step": 5300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.157584112019966e-07,
+      "loss": 1.188,
+      "step": 5301
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.151491016750117e-07,
+      "loss": 1.1184,
+      "step": 5302
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1454138129177172e-07,
+      "loss": 1.1813,
+      "step": 5303
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1393525024997909e-07,
+      "loss": 1.0879,
+      "step": 5304
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1333070874682217e-07,
+      "loss": 1.2073,
+      "step": 5305
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1272775697897087e-07,
+      "loss": 1.1047,
+      "step": 5306
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1212639514257829e-07,
+      "loss": 0.9139,
+      "step": 5307
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1152662343327792e-07,
+      "loss": 1.1717,
+      "step": 5308
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1092844204618979e-07,
+      "loss": 1.0549,
+      "step": 5309
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1033185117591494e-07,
+      "loss": 1.2179,
+      "step": 5310
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0973685101653641e-07,
+      "loss": 1.1261,
+      "step": 5311
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.091434417616194e-07,
+      "loss": 1.1358,
+      "step": 5312
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0855162360421278e-07,
+      "loss": 0.8546,
+      "step": 5313
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0796139673684758e-07,
+      "loss": 1.1183,
+      "step": 5314
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.073727613515374e-07,
+      "loss": 1.0939,
+      "step": 5315
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.067857176397763e-07,
+      "loss": 1.0604,
+      "step": 5316
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0620026579254261e-07,
+      "loss": 1.1551,
+      "step": 5317
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0561640600029566e-07,
+      "loss": 1.1282,
+      "step": 5318
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0503413845297739e-07,
+      "loss": 1.1302,
+      "step": 5319
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0445346334001182e-07,
+      "loss": 0.8304,
+      "step": 5320
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0387438085030454e-07,
+      "loss": 1.164,
+      "step": 5321
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0329689117224262e-07,
+      "loss": 1.0526,
+      "step": 5322
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0272099449369632e-07,
+      "loss": 1.2012,
+      "step": 5323
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0214669100201579e-07,
+      "loss": 1.1215,
+      "step": 5324
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0157398088403492e-07,
+      "loss": 1.1582,
+      "step": 5325
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0100286432606743e-07,
+      "loss": 1.0926,
+      "step": 5326
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0043334151391083e-07,
+      "loss": 1.021,
+      "step": 5327
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.986541263284077e-08,
+      "loss": 1.1752,
+      "step": 5328
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.929907786761783e-08,
+      "loss": 1.1748,
+      "step": 5329
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.873433740248183e-08,
+      "loss": 1.1758,
+      "step": 5330
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.817119142115472e-08,
+      "loss": 1.13,
+      "step": 5331
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.760964010683993e-08,
+      "loss": 1.1123,
+      "step": 5332
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.704968364222133e-08,
+      "loss": 1.1361,
+      "step": 5333
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.649132220946489e-08,
+      "loss": 1.1322,
+      "step": 5334
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.59345559902164e-08,
+      "loss": 1.1572,
+      "step": 5335
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.53793851656043e-08,
+      "loss": 0.879,
+      "step": 5336
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.482580991623747e-08,
+      "loss": 1.1647,
+      "step": 5337
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.427383042220517e-08,
+      "loss": 1.1696,
+      "step": 5338
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.372344686307655e-08,
+      "loss": 1.1203,
+      "step": 5339
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.317465941790449e-08,
+      "loss": 1.1346,
+      "step": 5340
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.262746826522062e-08,
+      "loss": 1.1945,
+      "step": 5341
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.208187358303755e-08,
+      "loss": 1.1712,
+      "step": 5342
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.15378755488483e-08,
+      "loss": 0.8744,
+      "step": 5343
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.099547433962686e-08,
+      "loss": 1.1613,
+      "step": 5344
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.045467013182762e-08,
+      "loss": 1.124,
+      "step": 5345
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.991546310138599e-08,
+      "loss": 1.1517,
+      "step": 5346
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.937785342371663e-08,
+      "loss": 1.0966,
+      "step": 5347
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.884184127371465e-08,
+      "loss": 1.14,
+      "step": 5348
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.830742682575776e-08,
+      "loss": 0.9692,
+      "step": 5349
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.777461025370083e-08,
+      "loss": 1.1902,
+      "step": 5350
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.72433917308796e-08,
+      "loss": 1.1538,
+      "step": 5351
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.671377143011194e-08,
+      "loss": 1.1174,
+      "step": 5352
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.618574952369341e-08,
+      "loss": 1.1338,
+      "step": 5353
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.565932618340156e-08,
+      "loss": 1.0769,
+      "step": 5354
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.513450158049109e-08,
+      "loss": 1.1592,
+      "step": 5355
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.461127588570039e-08,
+      "loss": 0.8884,
+      "step": 5356
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.40896492692439e-08,
+      "loss": 1.1609,
+      "step": 5357
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.356962190081918e-08,
+      "loss": 1.2198,
+      "step": 5358
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.305119394960037e-08,
+      "loss": 1.102,
+      "step": 5359
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.253436558424476e-08,
+      "loss": 1.1201,
+      "step": 5360
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.201913697288622e-08,
+      "loss": 1.0124,
+      "step": 5361
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.150550828313896e-08,
+      "loss": 0.8041,
+      "step": 5362
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.099347968209769e-08,
+      "loss": 1.0929,
+      "step": 5363
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.048305133633583e-08,
+      "loss": 1.3142,
+      "step": 5364
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.99742234119072e-08,
+      "loss": 0.8791,
+      "step": 5365
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.946699607434272e-08,
+      "loss": 1.117,
+      "step": 5366
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.896136948865429e-08,
+      "loss": 1.1461,
+      "step": 5367
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.845734381933367e-08,
+      "loss": 1.112,
+      "step": 5368
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.795491923035082e-08,
+      "loss": 1.1628,
+      "step": 5369
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.745409588515385e-08,
+      "loss": 1.1258,
+      "step": 5370
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.695487394667134e-08,
+      "loss": 1.1667,
+      "step": 5371
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.64572535773117e-08,
+      "loss": 1.0971,
+      "step": 5372
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.59612349389599e-08,
+      "loss": 1.1475,
+      "step": 5373
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.546681819298184e-08,
+      "loss": 1.1588,
+      "step": 5374
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.497400350022111e-08,
+      "loss": 1.1399,
+      "step": 5375
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.448279102100109e-08,
+      "loss": 1.1582,
+      "step": 5376
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.399318091512342e-08,
+      "loss": 1.1096,
+      "step": 5377
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.350517334186791e-08,
+      "loss": 0.8354,
+      "step": 5378
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.301876845999368e-08,
+      "loss": 1.1245,
+      "step": 5379
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.253396642773913e-08,
+      "loss": 1.0471,
+      "step": 5380
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.20507674028198e-08,
+      "loss": 1.2482,
+      "step": 5381
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.156917154243048e-08,
+      "loss": 1.1261,
+      "step": 5382
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.108917900324419e-08,
+      "loss": 1.1765,
+      "step": 5383
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.061078994141268e-08,
+      "loss": 1.1524,
+      "step": 5384
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.013400451256646e-08,
+      "loss": 0.8244,
+      "step": 5385
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.965882287181258e-08,
+      "loss": 1.1288,
+      "step": 5386
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.918524517373904e-08,
+      "loss": 1.216,
+      "step": 5387
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.87132715724087e-08,
+      "loss": 1.0624,
+      "step": 5388
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.824290222136599e-08,
+      "loss": 1.194,
+      "step": 5389
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.777413727363069e-08,
+      "loss": 1.089,
+      "step": 5390
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.730697688170251e-08,
+      "loss": 0.9515,
+      "step": 5391
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.684142119755821e-08,
+      "loss": 1.1704,
+      "step": 5392
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.637747037265274e-08,
+      "loss": 1.1789,
+      "step": 5393
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.591512455791926e-08,
+      "loss": 1.1334,
+      "step": 5394
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.545438390376858e-08,
+      "loss": 1.1509,
+      "step": 5395
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.499524856008854e-08,
+      "loss": 1.1622,
+      "step": 5396
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.453771867624581e-08,
+      "loss": 1.1417,
+      "step": 5397
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.408179440108409e-08,
+      "loss": 0.8574,
+      "step": 5398
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.362747588292584e-08,
+      "loss": 1.3106,
+      "step": 5399
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.31747632695695e-08,
+      "loss": 1.137,
+      "step": 5400
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.272365670829228e-08,
+      "loss": 0.8767,
+      "step": 5401
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.227415634584844e-08,
+      "loss": 1.1328,
+      "step": 5402
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.182626232847044e-08,
+      "loss": 1.1775,
+      "step": 5403
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.137997480186675e-08,
+      "loss": 1.255,
+      "step": 5404
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.093529391122399e-08,
+      "loss": 1.195,
+      "step": 5405
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.049221980120645e-08,
+      "loss": 1.1294,
+      "step": 5406
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.005075261595495e-08,
+      "loss": 0.9006,
+      "step": 5407
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.961089249908903e-08,
+      "loss": 1.1895,
+      "step": 5408
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.917263959370312e-08,
+      "loss": 1.1101,
+      "step": 5409
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.87359940423704e-08,
+      "loss": 1.1597,
+      "step": 5410
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.830095598714114e-08,
+      "loss": 1.1317,
+      "step": 5411
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.786752556954212e-08,
+      "loss": 1.1657,
+      "step": 5412
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.743570293057665e-08,
+      "loss": 1.139,
+      "step": 5413
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.700548821072627e-08,
+      "loss": 0.8258,
+      "step": 5414
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.6576881549949e-08,
+      "loss": 1.1194,
+      "step": 5415
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.614988308767888e-08,
+      "loss": 1.0815,
+      "step": 5416
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.5724492962828116e-08,
+      "loss": 1.15,
+      "step": 5417
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.530071131378434e-08,
+      "loss": 1.1274,
+      "step": 5418
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.4878538278412275e-08,
+      "loss": 1.2285,
+      "step": 5419
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.4457973994054837e-08,
+      "loss": 0.8399,
+      "step": 5420
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.4039018597529246e-08,
+      "loss": 1.1024,
+      "step": 5421
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.3621672225130925e-08,
+      "loss": 1.1513,
+      "step": 5422
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.320593501263071e-08,
+      "loss": 1.1447,
+      "step": 5423
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.279180709527765e-08,
+      "loss": 1.1812,
+      "step": 5424
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.2379288607795064e-08,
+      "loss": 1.1171,
+      "step": 5425
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.196837968438506e-08,
+      "loss": 1.1752,
+      "step": 5426
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.155908045872349e-08,
+      "loss": 0.8514,
+      "step": 5427
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.115139106396494e-08,
+      "loss": 1.1513,
+      "step": 5428
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.074531163273888e-08,
+      "loss": 1.1196,
+      "step": 5429
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.0340842297151304e-08,
+      "loss": 1.152,
+      "step": 5430
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.9937983188784736e-08,
+      "loss": 1.152,
+      "step": 5431
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.953673443869711e-08,
+      "loss": 1.1676,
+      "step": 5432
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.913709617742402e-08,
+      "loss": 1.1167,
+      "step": 5433
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.873906853497534e-08,
+      "loss": 1.1239,
+      "step": 5434
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8342651640838065e-08,
+      "loss": 1.171,
+      "step": 5435
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.794784562397459e-08,
+      "loss": 0.9786,
+      "step": 5436
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.7554650612823274e-08,
+      "loss": 1.0991,
+      "step": 5437
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.716306673529958e-08,
+      "loss": 1.1431,
+      "step": 5438
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.677309411879327e-08,
+      "loss": 1.1176,
+      "step": 5439
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.638473289017065e-08,
+      "loss": 1.1605,
+      "step": 5440
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.599798317577342e-08,
+      "loss": 1.0582,
+      "step": 5441
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5612845101419834e-08,
+      "loss": 1.1368,
+      "step": 5442
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5229318792403554e-08,
+      "loss": 0.8229,
+      "step": 5443
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4847404373492555e-08,
+      "loss": 1.0917,
+      "step": 5444
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.446710196893245e-08,
+      "loss": 1.1728,
+      "step": 5445
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4088411702443715e-08,
+      "loss": 1.0993,
+      "step": 5446
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3711333697221694e-08,
+      "loss": 1.1861,
+      "step": 5447
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.333586807593715e-08,
+      "loss": 1.1675,
+      "step": 5448
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.296201496073793e-08,
+      "loss": 1.0422,
+      "step": 5449
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2589774473246195e-08,
+      "loss": 1.1086,
+      "step": 5450
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.221914673455896e-08,
+      "loss": 1.1516,
+      "step": 5451
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1850131865249224e-08,
+      "loss": 1.1833,
+      "step": 5452
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.148272998536595e-08,
+      "loss": 1.1663,
+      "step": 5453
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.111694121443133e-08,
+      "loss": 1.1367,
+      "step": 5454
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.075276567144515e-08,
+      "loss": 1.1569,
+      "step": 5455
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.039020347488043e-08,
+      "loss": 0.8629,
+      "step": 5456
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0029254742687265e-08,
+      "loss": 1.0559,
+      "step": 5457
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9669919592288385e-08,
+      "loss": 1.1502,
+      "step": 5458
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.931219814058474e-08,
+      "loss": 1.1172,
+      "step": 5459
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.895609050394933e-08,
+      "loss": 1.1779,
+      "step": 5460
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.860159679823172e-08,
+      "loss": 1.1368,
+      "step": 5461
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8248717138755776e-08,
+      "loss": 1.1629,
+      "step": 5462
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7897451640321326e-08,
+      "loss": 0.8184,
+      "step": 5463
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7547800417201964e-08,
+      "loss": 1.1345,
+      "step": 5464
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.719976358314614e-08,
+      "loss": 1.231,
+      "step": 5465
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6853341251378294e-08,
+      "loss": 1.1096,
+      "step": 5466
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.650853353459605e-08,
+      "loss": 1.1636,
+      "step": 5467
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.616534054497301e-08,
+      "loss": 1.1065,
+      "step": 5468
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5823762394156526e-08,
+      "loss": 1.1811,
+      "step": 5469
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5483799193269944e-08,
+      "loss": 1.0769,
+      "step": 5470
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5145451052909784e-08,
+      "loss": 1.1679,
+      "step": 5471
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4808718083147454e-08,
+      "loss": 0.8119,
+      "step": 5472
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.447360039352976e-08,
+      "loss": 1.0892,
+      "step": 5473
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.414009809307728e-08,
+      "loss": 1.2071,
+      "step": 5474
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3808211290284886e-08,
+      "loss": 1.0576,
+      "step": 5475
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3477940093122886e-08,
+      "loss": 1.1308,
+      "step": 5476
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.314928460903532e-08,
+      "loss": 1.1181,
+      "step": 5477
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.282224494494057e-08,
+      "loss": 1.1468,
+      "step": 5478
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.249682120723074e-08,
+      "loss": 0.9325,
+      "step": 5479
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2173013501773934e-08,
+      "loss": 1.0951,
+      "step": 5480
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.185082193391143e-08,
+      "loss": 1.1928,
+      "step": 5481
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1530246608458273e-08,
+      "loss": 1.0561,
+      "step": 5482
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.121128762970438e-08,
+      "loss": 1.1533,
+      "step": 5483
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0893945101414524e-08,
+      "loss": 1.1017,
+      "step": 5484
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.057821912682612e-08,
+      "loss": 0.8731,
+      "step": 5485
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0264109808651466e-08,
+      "loss": 1.1819,
+      "step": 5486
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.995161724907658e-08,
+      "loss": 1.0821,
+      "step": 5487
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.964074154976293e-08,
+      "loss": 1.1727,
+      "step": 5488
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9331482811843504e-08,
+      "loss": 1.0863,
+      "step": 5489
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9023841135927822e-08,
+      "loss": 1.1815,
+      "step": 5490
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.871781662209694e-08,
+      "loss": 1.1344,
+      "step": 5491
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8413409369907887e-08,
+      "loss": 0.7898,
+      "step": 5492
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8110619478390332e-08,
+      "loss": 1.0814,
+      "step": 5493
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7809447046048266e-08,
+      "loss": 1.1796,
+      "step": 5494
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.750989217085942e-08,
+      "loss": 1.1172,
+      "step": 5495
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7211954950274734e-08,
+      "loss": 1.2114,
+      "step": 5496
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6915635481220005e-08,
+      "loss": 1.1382,
+      "step": 5497
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6620933860094234e-08,
+      "loss": 0.9435,
+      "step": 5498
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6327850182769065e-08,
+      "loss": 1.1209,
+      "step": 5499
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6036384544591564e-08,
+      "loss": 1.0768,
+      "step": 5500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5746537040381436e-08,
+      "loss": 1.1225,
+      "step": 5501
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.545830776443159e-08,
+      "loss": 1.1878,
+      "step": 5502
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5171696810509794e-08,
+      "loss": 1.1598,
+      "step": 5503
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.488670427185591e-08,
+      "loss": 1.1023,
+      "step": 5504
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4603330241184666e-08,
+      "loss": 1.1112,
+      "step": 5505
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4321574810682314e-08,
+      "loss": 1.1528,
+      "step": 5506
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4041438072011092e-08,
+      "loss": 1.1188,
+      "step": 5507
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3762920116304767e-08,
+      "loss": 0.916,
+      "step": 5508
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3486021034170857e-08,
+      "loss": 1.1226,
+      "step": 5509
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3210740915690644e-08,
+      "loss": 1.1816,
+      "step": 5510
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.29370798504186e-08,
+      "loss": 1.1825,
+      "step": 5511
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2665037927381838e-08,
+      "loss": 1.1257,
+      "step": 5512
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2394615235081795e-08,
+      "loss": 1.136,
+      "step": 5513
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2125811861492543e-08,
+      "loss": 0.891,
+      "step": 5514
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.185862789406079e-08,
+      "loss": 1.1344,
+      "step": 5515
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1593063419708128e-08,
+      "loss": 1.1178,
+      "step": 5516
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1329118524827662e-08,
+      "loss": 1.1907,
+      "step": 5517
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1066793295285714e-08,
+      "loss": 1.1079,
+      "step": 5518
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.080608781642235e-08,
+      "loss": 1.158,
+      "step": 5519
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0547002173051388e-08,
+      "loss": 1.0994,
+      "step": 5520
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0289536449457638e-08,
+      "loss": 0.8916,
+      "step": 5521
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0033690729400757e-08,
+      "loss": 1.1126,
+      "step": 5522
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9779465096112505e-08,
+      "loss": 1.1293,
+      "step": 5523
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.952685963229728e-08,
+      "loss": 1.163,
+      "step": 5524
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9275874420133788e-08,
+      "loss": 1.1008,
+      "step": 5525
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9026509541272276e-08,
+      "loss": 1.1582,
+      "step": 5526
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8778765076836182e-08,
+      "loss": 1.0223,
+      "step": 5527
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8532641107422144e-08,
+      "loss": 1.1154,
+      "step": 5528
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8288137713100007e-08,
+      "loss": 1.1145,
+      "step": 5529
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8045254973410586e-08,
+      "loss": 1.1257,
+      "step": 5530
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7803992967369564e-08,
+      "loss": 1.1509,
+      "step": 5531
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7564351773464162e-08,
+      "loss": 1.1984,
+      "step": 5532
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7326331469654235e-08,
+      "loss": 1.173,
+      "step": 5533
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7089932133373954e-08,
+      "loss": 0.9182,
+      "step": 5534
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6855153841527915e-08,
+      "loss": 1.1653,
+      "step": 5535
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6621996670495022e-08,
+      "loss": 1.131,
+      "step": 5536
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6390460696125155e-08,
+      "loss": 1.1763,
+      "step": 5537
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.616054599374306e-08,
+      "loss": 1.1026,
+      "step": 5538
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5932252638143906e-08,
+      "loss": 1.1559,
+      "step": 5539
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.570558070359718e-08,
+      "loss": 1.1515,
+      "step": 5540
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5480530263842773e-08,
+      "loss": 1.0719,
+      "step": 5541
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5257101392094908e-08,
+      "loss": 1.2032,
+      "step": 5542
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5035294161039882e-08,
+      "loss": 0.8314,
+      "step": 5543
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.481510864283553e-08,
+      "loss": 1.1463,
+      "step": 5544
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4596544909113441e-08,
+      "loss": 1.2069,
+      "step": 5545
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.437960303097674e-08,
+      "loss": 1.0968,
+      "step": 5546
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4164283079001196e-08,
+      "loss": 1.1507,
+      "step": 5547
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3950585123235216e-08,
+      "loss": 1.0866,
+      "step": 5548
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3738509233198195e-08,
+      "loss": 1.1779,
+      "step": 5549
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3528055477883273e-08,
+      "loss": 0.9067,
+      "step": 5550
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3319223925756242e-08,
+      "loss": 1.0945,
+      "step": 5551
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3112014644753313e-08,
+      "loss": 1.1398,
+      "step": 5552
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2906427702284452e-08,
+      "loss": 1.1673,
+      "step": 5553
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2702463165231161e-08,
+      "loss": 1.0973,
+      "step": 5554
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2500121099947582e-08,
+      "loss": 1.1423,
+      "step": 5555
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2299401572259395e-08,
+      "loss": 1.0905,
+      "step": 5556
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2100304647465477e-08,
+      "loss": 1.0473,
+      "step": 5557
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.190283039033624e-08,
+      "loss": 1.1647,
+      "step": 5558
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1706978865113072e-08,
+      "loss": 1.0988,
+      "step": 5559
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1512750135511674e-08,
+      "loss": 1.1728,
+      "step": 5560
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1320144264718169e-08,
+      "loss": 1.2137,
+      "step": 5561
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1129161315391879e-08,
+      "loss": 1.1874,
+      "step": 5562
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.093980134966255e-08,
+      "loss": 0.896,
+      "step": 5563
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.075206442913368e-08,
+      "loss": 1.126,
+      "step": 5564
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0565950614879195e-08,
+      "loss": 1.1459,
+      "step": 5565
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0381459967446772e-08,
+      "loss": 1.042,
+      "step": 5566
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0198592546855068e-08,
+      "loss": 1.1556,
+      "step": 5567
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0017348412593719e-08,
+      "loss": 1.1655,
+      "step": 5568
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.83772762362556e-09,
+      "loss": 1.141,
+      "step": 5569
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.659730238385512e-09,
+      "loss": 0.7632,
+      "step": 5570
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.48335631477948e-09,
+      "loss": 1.0413,
+      "step": 5571
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.308605910185453e-09,
+      "loss": 1.1365,
+      "step": 5572
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.135479081453513e-09,
+      "loss": 1.073,
+      "step": 5573
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.963975884905829e-09,
+      "loss": 1.2134,
+      "step": 5574
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.794096376334993e-09,
+      "loss": 1.1143,
+      "step": 5575
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.625840611007353e-09,
+      "loss": 1.1859,
+      "step": 5576
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.459208643659122e-09,
+      "loss": 1.1311,
+      "step": 5577
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.294200528499718e-09,
+      "loss": 1.1946,
+      "step": 5578
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.130816319209534e-09,
+      "loss": 0.9871,
+      "step": 5579
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.96905606894105e-09,
+      "loss": 1.1,
+      "step": 5580
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.80891983031773e-09,
+      "loss": 1.1349,
+      "step": 5581
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.650407655435676e-09,
+      "loss": 1.1736,
+      "step": 5582
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.49351959586253e-09,
+      "loss": 1.1771,
+      "step": 5583
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.338255702636354e-09,
+      "loss": 1.0919,
+      "step": 5584
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.1846160262684094e-09,
+      "loss": 1.1611,
+      "step": 5585
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.032600616740936e-09,
+      "loss": 0.9955,
+      "step": 5586
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.882209523507155e-09,
+      "loss": 1.1338,
+      "step": 5587
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.733442795492928e-09,
+      "loss": 1.214,
+      "step": 5588
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.586300481095098e-09,
+      "loss": 1.1138,
+      "step": 5589
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.440782628181486e-09,
+      "loss": 1.1715,
+      "step": 5590
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.29688928409311e-09,
+      "loss": 1.0923,
+      "step": 5591
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.154620495640862e-09,
+      "loss": 0.8871,
+      "step": 5592
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.013976309107716e-09,
+      "loss": 1.1655,
+      "step": 5593
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.874956770248186e-09,
+      "loss": 1.0984,
+      "step": 5594
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.737561924288315e-09,
+      "loss": 1.1585,
+      "step": 5595
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.601791815925129e-09,
+      "loss": 1.1716,
+      "step": 5596
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.4676464893282935e-09,
+      "loss": 1.0723,
+      "step": 5597
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.3351259881379016e-09,
+      "loss": 1.1277,
+      "step": 5598
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.204230355465023e-09,
+      "loss": 0.9629,
+      "step": 5599
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.074959633892818e-09,
+      "loss": 1.1268,
+      "step": 5600
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.947313865476533e-09,
+      "loss": 1.2186,
+      "step": 5601
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.821293091741286e-09,
+      "loss": 1.0554,
+      "step": 5602
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6968973536842825e-09,
+      "loss": 1.1378,
+      "step": 5603
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.574126691774261e-09,
+      "loss": 1.0628,
+      "step": 5604
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.452981145950941e-09,
+      "loss": 1.098,
+      "step": 5605
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.333460755625574e-09,
+      "loss": 1.1331,
+      "step": 5606
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.2155655596809455e-09,
+      "loss": 1.1408,
+      "step": 5607
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0992955964702656e-09,
+      "loss": 1.1989,
+      "step": 5608
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.984650903818277e-09,
+      "loss": 1.1032,
+      "step": 5609
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.871631519022367e-09,
+      "loss": 1.1049,
+      "step": 5610
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.760237478849793e-09,
+      "loss": 1.1789,
+      "step": 5611
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6504688195387883e-09,
+      "loss": 1.1037,
+      "step": 5612
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5423255767996765e-09,
+      "loss": 1.1954,
+      "step": 5613
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4358077858137607e-09,
+      "loss": 1.1381,
+      "step": 5614
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.330915481233321e-09,
+      "loss": 0.8595,
+      "step": 5615
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.227648697182173e-09,
+      "loss": 1.1676,
+      "step": 5616
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.126007467255665e-09,
+      "loss": 1.2389,
+      "step": 5617
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0259918245184593e-09,
+      "loss": 1.1151,
+      "step": 5618
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9276018015089725e-09,
+      "loss": 1.2101,
+      "step": 5619
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.830837430234934e-09,
+      "loss": 1.1403,
+      "step": 5620
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7356987421761626e-09,
+      "loss": 0.8845,
+      "step": 5621
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6421857682834562e-09,
+      "loss": 1.1676,
+      "step": 5622
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5502985389774806e-09,
+      "loss": 1.0668,
+      "step": 5623
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4600370841521004e-09,
+      "loss": 1.1513,
+      "step": 5624
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.371401433170495e-09,
+      "loss": 1.0593,
+      "step": 5625
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2843916148684862e-09,
+      "loss": 1.2112,
+      "step": 5626
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.199007657551766e-09,
+      "loss": 1.1126,
+      "step": 5627
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1152495889970035e-09,
+      "loss": 0.8775,
+      "step": 5628
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0331174364535135e-09,
+      "loss": 1.2034,
+      "step": 5629
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9526112266399223e-09,
+      "loss": 1.0922,
+      "step": 5630
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8737309857463916e-09,
+      "loss": 1.1385,
+      "step": 5631
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.796476739434616e-09,
+      "loss": 1.1695,
+      "step": 5632
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.72084851283727e-09,
+      "loss": 1.1551,
+      "step": 5633
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.64684633055745e-09,
+      "loss": 1.1228,
+      "step": 5634
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5744702166697878e-09,
+      "loss": 1.1284,
+      "step": 5635
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.503720194719338e-09,
+      "loss": 1.1684,
+      "step": 5636
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4345962877226893e-09,
+      "loss": 1.1192,
+      "step": 5637
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.367098518167964e-09,
+      "loss": 1.1667,
+      "step": 5638
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3012269080125982e-09,
+      "loss": 1.1278,
+      "step": 5639
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.236981478687227e-09,
+      "loss": 1.1232,
+      "step": 5640
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1743622510906883e-09,
+      "loss": 0.7961,
+      "step": 5641
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1133692455955747e-09,
+      "loss": 1.1812,
+      "step": 5642
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.054002482043237e-09,
+      "loss": 1.2078,
+      "step": 5643
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.962619797476702e-10,
+      "loss": 1.1266,
+      "step": 5644
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.401477574932927e-10,
+      "loss": 1.1654,
+      "step": 5645
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.856598335338362e-10,
+      "loss": 1.0864,
+      "step": 5646
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.32798225596787e-10,
+      "loss": 1.1594,
+      "step": 5647
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.815629508783896e-10,
+      "loss": 1.1086,
+      "step": 5648
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.319540260469771e-10,
+      "loss": 1.1792,
+      "step": 5649
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.839714672407516e-10,
+      "loss": 1.0446,
+      "step": 5650
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.376152900700039e-10,
+      "loss": 1.0647,
+      "step": 5651
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.928855096154485e-10,
+      "loss": 1.1852,
+      "step": 5652
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.497821404282233e-10,
+      "loss": 1.1378,
+      "step": 5653
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.083051965304453e-10,
+      "loss": 1.1293,
+      "step": 5654
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.684546914163201e-10,
+      "loss": 1.087,
+      "step": 5655
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3023063804936706e-10,
+      "loss": 1.1365,
+      "step": 5656
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.93633048865194e-10,
+      "loss": 0.9358,
+      "step": 5657
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.586619357692778e-10,
+      "loss": 1.0825,
+      "step": 5658
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.253173101386287e-10,
+      "loss": 1.1903,
+      "step": 5659
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.935991828206808e-10,
+      "loss": 1.1782,
+      "step": 5660
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6350756413440203e-10,
+      "loss": 1.0854,
+      "step": 5661
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3504246386918394e-10,
+      "loss": 1.167,
+      "step": 5662
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0820389128539674e-10,
+      "loss": 1.125,
+      "step": 5663
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8299185511383432e-10,
+      "loss": 0.8326,
+      "step": 5664
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5940636355626926e-10,
+      "loss": 1.2173,
+      "step": 5665
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3744742428656311e-10,
+      "loss": 1.0757,
+      "step": 5666
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1711504444733567e-10,
+      "loss": 1.2223,
+      "step": 5667
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.840923065385088e-11,
+      "loss": 1.0833,
+      "step": 5668
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.132998899124112e-11,
+      "loss": 1.1416,
+      "step": 5669
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.58773250156175e-11,
+      "loss": 0.9681,
+      "step": 5670
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.2051243754625006e-11,
+      "loss": 1.1438,
+      "step": 5671
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9851749705221987e-11,
+      "loss": 1.1783,
+      "step": 5672
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.927884683701088e-11,
+      "loss": 1.0713,
+      "step": 5673
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.033253858890749e-11,
+      "loss": 1.1296,
+      "step": 5674
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.30128278719166e-11,
+      "loss": 1.1053,
+      "step": 5675
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.31971706691148e-12,
+      "loss": 1.1847,
+      "step": 5676
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2532080262992484e-12,
+      "loss": 0.8682,
+      "step": 5677
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.133020729106378e-13,
+      "loss": 1.0672,
+      "step": 5678
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 1.2433,
+      "step": 5679
+    },
+    {
+      "epoch": 1.0,
+      "step": 5679,
+      "total_flos": 0.0,
+      "train_loss": 1.19415502449349,
+      "train_runtime": 39517.253,
+      "train_samples_per_second": 18.402,
+      "train_steps_per_second": 0.144
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 5679,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 0.0,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}