diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,44688 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 7443,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4642857142857145e-08,
+      "loss": 0.8204,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.928571428571429e-08,
+      "loss": 0.7793,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3392857142857142e-07,
+      "loss": 0.7679,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7857142857142858e-07,
+      "loss": 0.7806,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2321428571428574e-07,
+      "loss": 0.7884,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6785714285714284e-07,
+      "loss": 0.7391,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.125e-07,
+      "loss": 0.798,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5714285714285716e-07,
+      "loss": 0.7675,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0178571428571434e-07,
+      "loss": 0.745,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4642857142857147e-07,
+      "loss": 0.7889,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.910714285714286e-07,
+      "loss": 0.794,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.357142857142857e-07,
+      "loss": 0.7309,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.80357142857143e-07,
+      "loss": 0.7461,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.25e-07,
+      "loss": 0.74,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.696428571428571e-07,
+      "loss": 0.7599,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.142857142857143e-07,
+      "loss": 0.8024,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.589285714285714e-07,
+      "loss": 0.7293,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.035714285714287e-07,
+      "loss": 0.7355,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.482142857142859e-07,
+      "loss": 0.7258,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.928571428571429e-07,
+      "loss": 0.7464,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.375000000000001e-07,
+      "loss": 0.7395,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.821428571428572e-07,
+      "loss": 0.7119,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0267857142857143e-06,
+      "loss": 0.7256,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0714285714285714e-06,
+      "loss": 0.7241,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1160714285714287e-06,
+      "loss": 0.6701,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.160714285714286e-06,
+      "loss": 0.675,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.205357142857143e-06,
+      "loss": 0.7088,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.25e-06,
+      "loss": 0.6507,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2946428571428574e-06,
+      "loss": 0.6665,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3392857142857143e-06,
+      "loss": 0.6215,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3839285714285715e-06,
+      "loss": 0.6604,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4285714285714286e-06,
+      "loss": 0.6653,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.473214285714286e-06,
+      "loss": 0.6836,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5178571428571428e-06,
+      "loss": 0.6433,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5625e-06,
+      "loss": 0.6355,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6071428571428574e-06,
+      "loss": 0.653,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6517857142857144e-06,
+      "loss": 0.6319,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6964285714285717e-06,
+      "loss": 0.6226,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7410714285714286e-06,
+      "loss": 0.6446,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7857142857142859e-06,
+      "loss": 0.6524,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.830357142857143e-06,
+      "loss": 0.6848,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8750000000000003e-06,
+      "loss": 0.6368,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9196428571428573e-06,
+      "loss": 0.6255,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9642857142857144e-06,
+      "loss": 0.6428,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0089285714285715e-06,
+      "loss": 0.6491,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0535714285714286e-06,
+      "loss": 0.6408,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.098214285714286e-06,
+      "loss": 0.6573,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1428571428571427e-06,
+      "loss": 0.6572,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1875000000000002e-06,
+      "loss": 0.6719,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2321428571428573e-06,
+      "loss": 0.6805,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2767857142857144e-06,
+      "loss": 0.6371,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.321428571428572e-06,
+      "loss": 0.6244,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3660714285714285e-06,
+      "loss": 0.6209,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.410714285714286e-06,
+      "loss": 0.6495,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.455357142857143e-06,
+      "loss": 0.6547,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5e-06,
+      "loss": 0.6159,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5446428571428573e-06,
+      "loss": 0.6511,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5892857142857148e-06,
+      "loss": 0.6225,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.633928571428572e-06,
+      "loss": 0.6084,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6785714285714285e-06,
+      "loss": 0.6242,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7232142857142856e-06,
+      "loss": 0.6061,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.767857142857143e-06,
+      "loss": 0.675,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8125e-06,
+      "loss": 0.6191,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 0.6324,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9017857142857148e-06,
+      "loss": 0.6425,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.946428571428572e-06,
+      "loss": 0.6411,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.991071428571429e-06,
+      "loss": 0.585,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0357142857142856e-06,
+      "loss": 0.6187,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.080357142857143e-06,
+      "loss": 0.6544,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.125e-06,
+      "loss": 0.5815,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1696428571428572e-06,
+      "loss": 0.6249,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2142857142857147e-06,
+      "loss": 0.5971,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.258928571428572e-06,
+      "loss": 0.5942,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.303571428571429e-06,
+      "loss": 0.6196,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3482142857142855e-06,
+      "loss": 0.6471,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3928571428571435e-06,
+      "loss": 0.6238,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4375e-06,
+      "loss": 0.6134,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.482142857142857e-06,
+      "loss": 0.5863,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5267857142857147e-06,
+      "loss": 0.6514,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5714285714285718e-06,
+      "loss": 0.6058,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.616071428571429e-06,
+      "loss": 0.6138,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.660714285714286e-06,
+      "loss": 0.57,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7053571428571434e-06,
+      "loss": 0.6429,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7500000000000005e-06,
+      "loss": 0.6476,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.794642857142857e-06,
+      "loss": 0.6098,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.839285714285715e-06,
+      "loss": 0.5443,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.883928571428572e-06,
+      "loss": 0.5932,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.928571428571429e-06,
+      "loss": 0.6126,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9732142857142855e-06,
+      "loss": 0.6173,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.017857142857143e-06,
+      "loss": 0.5957,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0625000000000005e-06,
+      "loss": 0.6224,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.107142857142857e-06,
+      "loss": 0.6218,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.151785714285715e-06,
+      "loss": 0.6219,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.196428571428572e-06,
+      "loss": 0.6359,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.241071428571429e-06,
+      "loss": 0.5867,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2857142857142855e-06,
+      "loss": 0.5896,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.330357142857143e-06,
+      "loss": 0.5976,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3750000000000005e-06,
+      "loss": 0.5983,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.419642857142857e-06,
+      "loss": 0.5824,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.464285714285715e-06,
+      "loss": 0.5838,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.508928571428572e-06,
+      "loss": 0.5653,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.553571428571429e-06,
+      "loss": 0.5493,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5982142857142854e-06,
+      "loss": 0.5768,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.642857142857144e-06,
+      "loss": 0.5643,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6875000000000004e-06,
+      "loss": 0.5798,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.732142857142857e-06,
+      "loss": 0.6051,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.776785714285715e-06,
+      "loss": 0.5454,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.821428571428572e-06,
+      "loss": 0.5933,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.866071428571429e-06,
+      "loss": 0.5633,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.910714285714286e-06,
+      "loss": 0.5907,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.955357142857144e-06,
+      "loss": 0.5521,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5e-06,
+      "loss": 0.5709,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.044642857142857e-06,
+      "loss": 0.5419,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.0892857142857146e-06,
+      "loss": 0.5927,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.133928571428571e-06,
+      "loss": 0.5579,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.1785714285714296e-06,
+      "loss": 0.5833,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.223214285714286e-06,
+      "loss": 0.5368,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.267857142857144e-06,
+      "loss": 0.528,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.3125e-06,
+      "loss": 0.5922,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.357142857142857e-06,
+      "loss": 0.6125,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.4017857142857145e-06,
+      "loss": 0.5547,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.446428571428571e-06,
+      "loss": 0.555,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.4910714285714295e-06,
+      "loss": 0.5522,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.535714285714286e-06,
+      "loss": 0.5435,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.580357142857144e-06,
+      "loss": 0.5315,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.625e-06,
+      "loss": 0.5583,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.669642857142858e-06,
+      "loss": 0.5276,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.7142857142857145e-06,
+      "loss": 0.5575,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.758928571428571e-06,
+      "loss": 0.5789,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.8035714285714295e-06,
+      "loss": 0.566,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.848214285714286e-06,
+      "loss": 0.5605,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.892857142857144e-06,
+      "loss": 0.5839,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9375e-06,
+      "loss": 0.55,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.982142857142858e-06,
+      "loss": 0.5959,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.0267857142857145e-06,
+      "loss": 0.5604,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.071428571428571e-06,
+      "loss": 0.5803,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.1160714285714295e-06,
+      "loss": 0.5211,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.160714285714286e-06,
+      "loss": 0.5008,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.205357142857144e-06,
+      "loss": 0.6093,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.25e-06,
+      "loss": 0.5309,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.294642857142858e-06,
+      "loss": 0.5883,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.3392857142857145e-06,
+      "loss": 0.5639,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.383928571428571e-06,
+      "loss": 0.5472,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.4285714285714295e-06,
+      "loss": 0.5342,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.473214285714287e-06,
+      "loss": 0.5617,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.517857142857144e-06,
+      "loss": 0.5503,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.5625e-06,
+      "loss": 0.5394,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.607142857142858e-06,
+      "loss": 0.5491,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.6517857142857144e-06,
+      "loss": 0.515,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.696428571428571e-06,
+      "loss": 0.5832,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.7410714285714294e-06,
+      "loss": 0.5492,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.785714285714287e-06,
+      "loss": 0.5173,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.830357142857144e-06,
+      "loss": 0.558,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.875e-06,
+      "loss": 0.564,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.919642857142858e-06,
+      "loss": 0.561,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.964285714285714e-06,
+      "loss": 0.5411,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.008928571428572e-06,
+      "loss": 0.5422,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.053571428571429e-06,
+      "loss": 0.5611,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.098214285714287e-06,
+      "loss": 0.5072,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.1428571428571436e-06,
+      "loss": 0.5524,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.1875e-06,
+      "loss": 0.5506,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.232142857142858e-06,
+      "loss": 0.5407,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.276785714285714e-06,
+      "loss": 0.5204,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.321428571428572e-06,
+      "loss": 0.5563,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.366071428571429e-06,
+      "loss": 0.5685,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.410714285714287e-06,
+      "loss": 0.5519,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.4553571428571435e-06,
+      "loss": 0.5555,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 0.5298,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.544642857142858e-06,
+      "loss": 0.5154,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.589285714285714e-06,
+      "loss": 0.505,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.633928571428572e-06,
+      "loss": 0.511,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.67857142857143e-06,
+      "loss": 0.5271,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.723214285714287e-06,
+      "loss": 0.5428,
+      "step": 173
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.767857142857144e-06,
+      "loss": 0.5204,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.8125e-06,
+      "loss": 0.5461,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.857142857142858e-06,
+      "loss": 0.5131,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.901785714285715e-06,
+      "loss": 0.5456,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.946428571428571e-06,
+      "loss": 0.5532,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.99107142857143e-06,
+      "loss": 0.5339,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.035714285714286e-06,
+      "loss": 0.5163,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.080357142857143e-06,
+      "loss": 0.5709,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.125000000000001e-06,
+      "loss": 0.5288,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.169642857142858e-06,
+      "loss": 0.4962,
+      "step": 183
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.214285714285714e-06,
+      "loss": 0.5211,
+      "step": 184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.258928571428572e-06,
+      "loss": 0.4981,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.30357142857143e-06,
+      "loss": 0.4781,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.348214285714287e-06,
+      "loss": 0.5622,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.392857142857144e-06,
+      "loss": 0.5303,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.4375e-06,
+      "loss": 0.5355,
+      "step": 189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.482142857142858e-06,
+      "loss": 0.5287,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.526785714285715e-06,
+      "loss": 0.546,
+      "step": 191
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.571428571428571e-06,
+      "loss": 0.5642,
+      "step": 192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.61607142857143e-06,
+      "loss": 0.5316,
+      "step": 193
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.660714285714286e-06,
+      "loss": 0.5387,
+      "step": 194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.705357142857143e-06,
+      "loss": 0.5282,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.750000000000001e-06,
+      "loss": 0.5572,
+      "step": 196
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.794642857142858e-06,
+      "loss": 0.5696,
+      "step": 197
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.839285714285714e-06,
+      "loss": 0.5055,
+      "step": 198
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.883928571428572e-06,
+      "loss": 0.5287,
+      "step": 199
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.92857142857143e-06,
+      "loss": 0.5058,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.973214285714287e-06,
+      "loss": 0.5489,
+      "step": 201
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.017857142857144e-06,
+      "loss": 0.5192,
+      "step": 202
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.0625e-06,
+      "loss": 0.4852,
+      "step": 203
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.107142857142858e-06,
+      "loss": 0.5376,
+      "step": 204
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.151785714285715e-06,
+      "loss": 0.4967,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.196428571428571e-06,
+      "loss": 0.5439,
+      "step": 206
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.24107142857143e-06,
+      "loss": 0.5504,
+      "step": 207
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.285714285714288e-06,
+      "loss": 0.5283,
+      "step": 208
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.330357142857143e-06,
+      "loss": 0.5504,
+      "step": 209
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.375000000000001e-06,
+      "loss": 0.5196,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.419642857142858e-06,
+      "loss": 0.55,
+      "step": 211
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.464285714285714e-06,
+      "loss": 0.5194,
+      "step": 212
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.508928571428572e-06,
+      "loss": 0.562,
+      "step": 213
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.55357142857143e-06,
+      "loss": 0.5501,
+      "step": 214
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.598214285714287e-06,
+      "loss": 0.527,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.642857142857144e-06,
+      "loss": 0.5362,
+      "step": 216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.6875e-06,
+      "loss": 0.5389,
+      "step": 217
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.732142857142858e-06,
+      "loss": 0.5363,
+      "step": 218
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.776785714285715e-06,
+      "loss": 0.5403,
+      "step": 219
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.821428571428573e-06,
+      "loss": 0.5213,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.86607142857143e-06,
+      "loss": 0.5253,
+      "step": 221
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.910714285714288e-06,
+      "loss": 0.5313,
+      "step": 222
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.955357142857143e-06,
+      "loss": 0.5105,
+      "step": 223
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1e-05,
+      "loss": 0.5085,
+      "step": 224
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999999526537418e-06,
+      "loss": 0.5588,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999998106149756e-06,
+      "loss": 0.5523,
+      "step": 226
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999995738837287e-06,
+      "loss": 0.5198,
+      "step": 227
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999992424600458e-06,
+      "loss": 0.5052,
+      "step": 228
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999988163439896e-06,
+      "loss": 0.5336,
+      "step": 229
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99998295535641e-06,
+      "loss": 0.5344,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999976800350983e-06,
+      "loss": 0.5216,
+      "step": 231
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999969698424784e-06,
+      "loss": 0.5375,
+      "step": 232
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999961649579155e-06,
+      "loss": 0.5504,
+      "step": 233
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999952653815622e-06,
+      "loss": 0.5155,
+      "step": 234
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99994271113589e-06,
+      "loss": 0.5396,
+      "step": 235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99993182154184e-06,
+      "loss": 0.5297,
+      "step": 236
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999919985035534e-06,
+      "loss": 0.5501,
+      "step": 237
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999907201619217e-06,
+      "loss": 0.5451,
+      "step": 238
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999893471295305e-06,
+      "loss": 0.5374,
+      "step": 239
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999878794066402e-06,
+      "loss": 0.5353,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999863169935286e-06,
+      "loss": 0.5058,
+      "step": 241
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999846598904918e-06,
+      "loss": 0.5158,
+      "step": 242
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999829080978433e-06,
+      "loss": 0.5177,
+      "step": 243
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999810616159151e-06,
+      "loss": 0.5366,
+      "step": 244
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99979120445057e-06,
+      "loss": 0.539,
+      "step": 245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999770845856363e-06,
+      "loss": 0.5275,
+      "step": 246
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999749540380385e-06,
+      "loss": 0.5153,
+      "step": 247
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999727288026677e-06,
+      "loss": 0.5367,
+      "step": 248
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999704088799449e-06,
+      "loss": 0.5486,
+      "step": 249
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999679942703094e-06,
+      "loss": 0.5476,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999654849742187e-06,
+      "loss": 0.55,
+      "step": 251
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999628809921478e-06,
+      "loss": 0.5601,
+      "step": 252
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.9996018232459e-06,
+      "loss": 0.5112,
+      "step": 253
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999573889720565e-06,
+      "loss": 0.5459,
+      "step": 254
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999545009350759e-06,
+      "loss": 0.4988,
+      "step": 255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999515182141956e-06,
+      "loss": 0.5341,
+      "step": 256
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999484408099804e-06,
+      "loss": 0.544,
+      "step": 257
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99945268723013e-06,
+      "loss": 0.509,
+      "step": 258
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999420019538942e-06,
+      "loss": 0.5515,
+      "step": 259
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999386405032425e-06,
+      "loss": 0.5518,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999351843716948e-06,
+      "loss": 0.5239,
+      "step": 261
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999316335599055e-06,
+      "loss": 0.5329,
+      "step": 262
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999279880685472e-06,
+      "loss": 0.5482,
+      "step": 263
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999242478983097e-06,
+      "loss": 0.5085,
+      "step": 264
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999204130499022e-06,
+      "loss": 0.512,
+      "step": 265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999164835240504e-06,
+      "loss": 0.5247,
+      "step": 266
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999124593214989e-06,
+      "loss": 0.5314,
+      "step": 267
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999083404430095e-06,
+      "loss": 0.5248,
+      "step": 268
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999041268893622e-06,
+      "loss": 0.5316,
+      "step": 269
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998998186613552e-06,
+      "loss": 0.5149,
+      "step": 270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998954157598043e-06,
+      "loss": 0.4907,
+      "step": 271
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998909181855434e-06,
+      "loss": 0.5211,
+      "step": 272
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998863259394244e-06,
+      "loss": 0.4968,
+      "step": 273
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998816390223168e-06,
+      "loss": 0.5275,
+      "step": 274
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99876857435108e-06,
+      "loss": 0.51,
+      "step": 275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998719811787042e-06,
+      "loss": 0.5149,
+      "step": 276
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998670102540285e-06,
+      "loss": 0.5351,
+      "step": 277
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998619446620222e-06,
+      "loss": 0.5374,
+      "step": 278
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998567844036449e-06,
+      "loss": 0.5364,
+      "step": 279
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998515294798738e-06,
+      "loss": 0.534,
+      "step": 280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99846179891704e-06,
+      "loss": 0.5393,
+      "step": 281
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998407356401488e-06,
+      "loss": 0.5565,
+      "step": 282
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998351967262391e-06,
+      "loss": 0.5321,
+      "step": 283
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998295631510242e-06,
+      "loss": 0.5525,
+      "step": 284
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998238349155706e-06,
+      "loss": 0.5332,
+      "step": 285
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998180120209632e-06,
+      "loss": 0.5462,
+      "step": 286
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998120944683052e-06,
+      "loss": 0.4955,
+      "step": 287
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998060822587167e-06,
+      "loss": 0.4783,
+      "step": 288
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997999753933368e-06,
+      "loss": 0.5355,
+      "step": 289
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997937738733218e-06,
+      "loss": 0.5115,
+      "step": 290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997874776998463e-06,
+      "loss": 0.4971,
+      "step": 291
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997810868741024e-06,
+      "loss": 0.5348,
+      "step": 292
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997746013973008e-06,
+      "loss": 0.5623,
+      "step": 293
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997680212706698e-06,
+      "loss": 0.5572,
+      "step": 294
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997613464954551e-06,
+      "loss": 0.5273,
+      "step": 295
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997545770729213e-06,
+      "loss": 0.5472,
+      "step": 296
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.9974771300435e-06,
+      "loss": 0.5459,
+      "step": 297
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997407542910415e-06,
+      "loss": 0.5024,
+      "step": 298
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997337009343137e-06,
+      "loss": 0.504,
+      "step": 299
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99726552935502e-06,
+      "loss": 0.5348,
+      "step": 300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997193102959607e-06,
+      "loss": 0.5336,
+      "step": 301
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997119730170608e-06,
+      "loss": 0.5787,
+      "step": 302
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997045411001923e-06,
+      "loss": 0.5095,
+      "step": 303
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996970145467626e-06,
+      "loss": 0.5423,
+      "step": 304
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99689393358197e-06,
+      "loss": 0.5351,
+      "step": 305
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996816775359392e-06,
+      "loss": 0.5426,
+      "step": 306
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.9967386708145e-06,
+      "loss": 0.5169,
+      "step": 307
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996659619962088e-06,
+      "loss": 0.5169,
+      "step": 308
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996579622817127e-06,
+      "loss": 0.4865,
+      "step": 309
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996498679394769e-06,
+      "loss": 0.5336,
+      "step": 310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99641678971034e-06,
+      "loss": 0.5382,
+      "step": 311
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99633395377935e-06,
+      "loss": 0.5241,
+      "step": 312
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996250171617487e-06,
+      "loss": 0.5132,
+      "step": 313
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996165443240619e-06,
+      "loss": 0.5039,
+      "step": 314
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996079768664791e-06,
+      "loss": 0.5098,
+      "step": 315
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99599314790623e-06,
+      "loss": 0.5078,
+      "step": 316
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995905580981339e-06,
+      "loss": 0.5161,
+      "step": 317
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995817067906704e-06,
+      "loss": 0.5399,
+      "step": 318
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995727608699084e-06,
+      "loss": 0.5322,
+      "step": 319
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995637203375425e-06,
+      "loss": 0.5396,
+      "step": 320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995545851952848e-06,
+      "loss": 0.5015,
+      "step": 321
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995453554448652e-06,
+      "loss": 0.4874,
+      "step": 322
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995360310880319e-06,
+      "loss": 0.5631,
+      "step": 323
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995266121265507e-06,
+      "loss": 0.5373,
+      "step": 324
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995170985622052e-06,
+      "loss": 0.5442,
+      "step": 325
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995074903967974e-06,
+      "loss": 0.5285,
+      "step": 326
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994977876321468e-06,
+      "loss": 0.525,
+      "step": 327
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99487990270091e-06,
+      "loss": 0.5198,
+      "step": 328
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994780983124855e-06,
+      "loss": 0.5123,
+      "step": 329
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994681117612037e-06,
+      "loss": 0.5449,
+      "step": 330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994580306181368e-06,
+      "loss": 0.5338,
+      "step": 331
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994478548851941e-06,
+      "loss": 0.5246,
+      "step": 332
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994375845643026e-06,
+      "loss": 0.5315,
+      "step": 333
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994272196574077e-06,
+      "loss": 0.5535,
+      "step": 334
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.994167601664719e-06,
+      "loss": 0.5401,
+      "step": 335
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.994062060934763e-06,
+      "loss": 0.4832,
+      "step": 336
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993955574404197e-06,
+      "loss": 0.5694,
+      "step": 337
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99384814209319e-06,
+      "loss": 0.5409,
+      "step": 338
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993739764022084e-06,
+      "loss": 0.5088,
+      "step": 339
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993630440211404e-06,
+      "loss": 0.5416,
+      "step": 340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993520170681858e-06,
+      "loss": 0.5249,
+      "step": 341
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993408955454328e-06,
+      "loss": 0.5006,
+      "step": 342
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993296794549877e-06,
+      "loss": 0.5174,
+      "step": 343
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993183687989743e-06,
+      "loss": 0.5443,
+      "step": 344
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993069635795352e-06,
+      "loss": 0.5277,
+      "step": 345
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992954637988299e-06,
+      "loss": 0.5148,
+      "step": 346
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992838694590365e-06,
+      "loss": 0.5492,
+      "step": 347
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99272180562351e-06,
+      "loss": 0.5226,
+      "step": 348
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992603971109869e-06,
+      "loss": 0.5008,
+      "step": 349
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992485191071756e-06,
+      "loss": 0.5425,
+      "step": 350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99236546553167e-06,
+      "loss": 0.5176,
+      "step": 351
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992244794512282e-06,
+      "loss": 0.5045,
+      "step": 352
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992123178036447e-06,
+      "loss": 0.5346,
+      "step": 353
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.9920006161272e-06,
+      "loss": 0.5227,
+      "step": 354
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991877108807745e-06,
+      "loss": 0.5195,
+      "step": 355
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99175265610148e-06,
+      "loss": 0.5162,
+      "step": 356
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991627258031972e-06,
+      "loss": 0.5206,
+      "step": 357
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991500914622967e-06,
+      "loss": 0.5015,
+      "step": 358
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991373625898395e-06,
+      "loss": 0.5118,
+      "step": 359
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991245391882363e-06,
+      "loss": 0.4963,
+      "step": 360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991116212599155e-06,
+      "loss": 0.5138,
+      "step": 361
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990986088073235e-06,
+      "loss": 0.4976,
+      "step": 362
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99085501832925e-06,
+      "loss": 0.5029,
+      "step": 363
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99072300339202e-06,
+      "loss": 0.5145,
+      "step": 364
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990590043286546e-06,
+      "loss": 0.5072,
+      "step": 365
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990456138038012e-06,
+      "loss": 0.5301,
+      "step": 366
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990321287671773e-06,
+      "loss": 0.514,
+      "step": 367
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990185492213373e-06,
+      "loss": 0.5266,
+      "step": 368
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990048751688526e-06,
+      "loss": 0.5409,
+      "step": 369
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989911066123128e-06,
+      "loss": 0.4929,
+      "step": 370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989772435543256e-06,
+      "loss": 0.5312,
+      "step": 371
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989632859975166e-06,
+      "loss": 0.4995,
+      "step": 372
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98949233944529e-06,
+      "loss": 0.5145,
+      "step": 373
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98935087398024e-06,
+      "loss": 0.5155,
+      "step": 374
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989208463606808e-06,
+      "loss": 0.4963,
+      "step": 375
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989065108351965e-06,
+      "loss": 0.4975,
+      "step": 376
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988920808242858e-06,
+      "loss": 0.5197,
+      "step": 377
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988775563306819e-06,
+      "loss": 0.5269,
+      "step": 378
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988629373571353e-06,
+      "loss": 0.4847,
+      "step": 379
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988482239064146e-06,
+      "loss": 0.4989,
+      "step": 380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988334159813064e-06,
+      "loss": 0.5296,
+      "step": 381
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988185135846149e-06,
+      "loss": 0.5055,
+      "step": 382
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988035167191626e-06,
+      "loss": 0.5413,
+      "step": 383
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987884253877898e-06,
+      "loss": 0.4835,
+      "step": 384
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98773239593354e-06,
+      "loss": 0.5429,
+      "step": 385
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98757959338732e-06,
+      "loss": 0.5337,
+      "step": 386
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987425846268168e-06,
+      "loss": 0.5427,
+      "step": 387
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987271154605208e-06,
+      "loss": 0.518,
+      "step": 388
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987115518427734e-06,
+      "loss": 0.5389,
+      "step": 389
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98695893776522e-06,
+      "loss": 0.5378,
+      "step": 390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98680141264732e-06,
+      "loss": 0.5416,
+      "step": 391
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98664294310387e-06,
+      "loss": 0.5363,
+      "step": 392
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986483529164877e-06,
+      "loss": 0.5106,
+      "step": 393
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986323170860538e-06,
+      "loss": 0.5226,
+      "step": 394
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986161868221215e-06,
+      "loss": 0.5194,
+      "step": 395
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985999621277462e-06,
+      "loss": 0.5597,
+      "step": 396
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985836430060003e-06,
+      "loss": 0.4986,
+      "step": 397
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985672294599745e-06,
+      "loss": 0.5282,
+      "step": 398
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985507214927773e-06,
+      "loss": 0.5359,
+      "step": 399
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98534119107535e-06,
+      "loss": 0.4943,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98517422307392e-06,
+      "loss": 0.5558,
+      "step": 401
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985006310955104e-06,
+      "loss": 0.5015,
+      "step": 402
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.9848374547507e-06,
+      "loss": 0.532,
+      "step": 403
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984667654492688e-06,
+      "loss": 0.542,
+      "step": 404
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984496910213226e-06,
+      "loss": 0.5173,
+      "step": 405
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98432522194465e-06,
+      "loss": 0.5063,
+      "step": 406
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984152589719476e-06,
+      "loss": 0.5206,
+      "step": 407
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983979013570397e-06,
+      "loss": 0.522,
+      "step": 408
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983804493530286e-06,
+      "loss": 0.5007,
+      "step": 409
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983629029632194e-06,
+      "loss": 0.5446,
+      "step": 410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983452621909353e-06,
+      "loss": 0.5225,
+      "step": 411
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983275270395168e-06,
+      "loss": 0.5257,
+      "step": 412
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983096975123231e-06,
+      "loss": 0.5203,
+      "step": 413
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982917736127307e-06,
+      "loss": 0.4946,
+      "step": 414
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982737553441341e-06,
+      "loss": 0.5251,
+      "step": 415
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982556427099456e-06,
+      "loss": 0.5351,
+      "step": 416
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982374357135957e-06,
+      "loss": 0.5391,
+      "step": 417
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982191343585323e-06,
+      "loss": 0.5252,
+      "step": 418
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982007386482216e-06,
+      "loss": 0.5475,
+      "step": 419
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981822485861471e-06,
+      "loss": 0.5242,
+      "step": 420
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98163664175811e-06,
+      "loss": 0.5209,
+      "step": 421
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981449854207328e-06,
+      "loss": 0.5103,
+      "step": 422
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981262123244496e-06,
+      "loss": 0.5363,
+      "step": 423
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981073448905174e-06,
+      "loss": 0.5197,
+      "step": 424
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98088383122509e-06,
+      "loss": 0.514,
+      "step": 425
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980693270240153e-06,
+      "loss": 0.5526,
+      "step": 426
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980501765986455e-06,
+      "loss": 0.501,
+      "step": 427
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980309318500265e-06,
+      "loss": 0.5352,
+      "step": 428
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980115927818026e-06,
+      "loss": 0.4911,
+      "step": 429
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979921593976367e-06,
+      "loss": 0.5083,
+      "step": 430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97972631701209e-06,
+      "loss": 0.5179,
+      "step": 431
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97953009696218e-06,
+      "loss": 0.5276,
+      "step": 432
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979332933863794e-06,
+      "loss": 0.5007,
+      "step": 433
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979134827754276e-06,
+      "loss": 0.5159,
+      "step": 434
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97893577867114e-06,
+      "loss": 0.504,
+      "step": 435
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978735786652086e-06,
+      "loss": 0.4944,
+      "step": 436
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978534851734989e-06,
+      "loss": 0.5247,
+      "step": 437
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978332973957904e-06,
+      "loss": 0.5176,
+      "step": 438
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97813015335906e-06,
+      "loss": 0.4897,
+      "step": 439
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97792638997687e-06,
+      "loss": 0.5034,
+      "step": 440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977721683849927e-06,
+      "loss": 0.4855,
+      "step": 441
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977516035016997e-06,
+      "loss": 0.4879,
+      "step": 442
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977309443517024e-06,
+      "loss": 0.5283,
+      "step": 443
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977101909389137e-06,
+      "loss": 0.5392,
+      "step": 444
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976893432672637e-06,
+      "loss": 0.5016,
+      "step": 445
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97668401340701e-06,
+      "loss": 0.5171,
+      "step": 446
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976473651631912e-06,
+      "loss": 0.5217,
+      "step": 447
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97626234738719e-06,
+      "loss": 0.5378,
+      "step": 448
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976050100712854e-06,
+      "loss": 0.4999,
+      "step": 449
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975836911649102e-06,
+      "loss": 0.5166,
+      "step": 450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975622780236312e-06,
+      "loss": 0.54,
+      "step": 451
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975407706515035e-06,
+      "loss": 0.5352,
+      "step": 452
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975191690526005e-06,
+      "loss": 0.5589,
+      "step": 453
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974974732310128e-06,
+      "loss": 0.4924,
+      "step": 454
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974756831908496e-06,
+      "loss": 0.5133,
+      "step": 455
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974537989362376e-06,
+      "loss": 0.5241,
+      "step": 456
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974318204713211e-06,
+      "loss": 0.5013,
+      "step": 457
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974097478002629e-06,
+      "loss": 0.5011,
+      "step": 458
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973875809272428e-06,
+      "loss": 0.5132,
+      "step": 459
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97365319856459e-06,
+      "loss": 0.5002,
+      "step": 460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973429645921277e-06,
+      "loss": 0.498,
+      "step": 461
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973205151384823e-06,
+      "loss": 0.5146,
+      "step": 462
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972979714997746e-06,
+      "loss": 0.5104,
+      "step": 463
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972753336802738e-06,
+      "loss": 0.523,
+      "step": 464
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972526016842673e-06,
+      "loss": 0.5135,
+      "step": 465
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972297755160604e-06,
+      "loss": 0.5162,
+      "step": 466
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972068551799759e-06,
+      "loss": 0.5077,
+      "step": 467
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971838406803543e-06,
+      "loss": 0.5423,
+      "step": 468
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971607320215546e-06,
+      "loss": 0.4948,
+      "step": 469
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97137529207953e-06,
+      "loss": 0.556,
+      "step": 470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971142322439437e-06,
+      "loss": 0.5019,
+      "step": 471
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97090841133939e-06,
+      "loss": 0.5163,
+      "step": 472
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970673558823689e-06,
+      "loss": 0.527,
+      "step": 473
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970437764936806e-06,
+      "loss": 0.5215,
+      "step": 474
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970201029723405e-06,
+      "loss": 0.5273,
+      "step": 475
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969963353228314e-06,
+      "loss": 0.5026,
+      "step": 476
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969724735496546e-06,
+      "loss": 0.5166,
+      "step": 477
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969485176573295e-06,
+      "loss": 0.5264,
+      "step": 478
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969244676503929e-06,
+      "loss": 0.5045,
+      "step": 479
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.96900323533399e-06,
+      "loss": 0.5258,
+      "step": 480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.96876085310921e-06,
+      "loss": 0.5178,
+      "step": 481
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.96851752987549e-06,
+      "loss": 0.5044,
+      "step": 482
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.96827326567891e-06,
+      "loss": 0.5375,
+      "step": 483
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.968028060565732e-06,
+      "loss": 0.5407,
+      "step": 484
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967781914582394e-06,
+      "loss": 0.499,
+      "step": 485
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967534827775513e-06,
+      "loss": 0.5443,
+      "step": 486
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967286800191882e-06,
+      "loss": 0.5327,
+      "step": 487
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967037831878473e-06,
+      "loss": 0.5057,
+      "step": 488
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966787922882439e-06,
+      "loss": 0.5325,
+      "step": 489
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96653707325111e-06,
+      "loss": 0.5131,
+      "step": 490
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966285283031989e-06,
+      "loss": 0.5123,
+      "step": 491
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966032552272764e-06,
+      "loss": 0.5341,
+      "step": 492
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.9657788810213e-06,
+      "loss": 0.4847,
+      "step": 493
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965524269325635e-06,
+      "loss": 0.4996,
+      "step": 494
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965268717233991e-06,
+      "loss": 0.5178,
+      "step": 495
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965012224794765e-06,
+      "loss": 0.5019,
+      "step": 496
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964754792056534e-06,
+      "loss": 0.5488,
+      "step": 497
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96449641906805e-06,
+      "loss": 0.4908,
+      "step": 498
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964237105878246e-06,
+      "loss": 0.5032,
+      "step": 499
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96397685253623e-06,
+      "loss": 0.5092,
+      "step": 500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963715659091293e-06,
+      "loss": 0.5459,
+      "step": 501
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.9634535255929e-06,
+      "loss": 0.5058,
+      "step": 502
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963190452090697e-06,
+      "loss": 0.5317,
+      "step": 503
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962926438634501e-06,
+      "loss": 0.5479,
+      "step": 504
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962661485274319e-06,
+      "loss": 0.4985,
+      "step": 505
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962395592060324e-06,
+      "loss": 0.5065,
+      "step": 506
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962128759042876e-06,
+      "loss": 0.538,
+      "step": 507
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961860986272505e-06,
+      "loss": 0.5285,
+      "step": 508
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961592273799925e-06,
+      "loss": 0.5098,
+      "step": 509
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961322621676026e-06,
+      "loss": 0.5357,
+      "step": 510
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961052029951878e-06,
+      "loss": 0.5131,
+      "step": 511
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960780498678726e-06,
+      "loss": 0.5482,
+      "step": 512
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960508027907993e-06,
+      "loss": 0.4984,
+      "step": 513
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96023461769128e-06,
+      "loss": 0.4993,
+      "step": 514
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959960268080371e-06,
+      "loss": 0.493,
+      "step": 515
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959684979127217e-06,
+      "loss": 0.5378,
+      "step": 516
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95940875088396e-06,
+      "loss": 0.5046,
+      "step": 517
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959131583402912e-06,
+      "loss": 0.5376,
+      "step": 518
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958853476736562e-06,
+      "loss": 0.5235,
+      "step": 519
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95857443093758e-06,
+      "loss": 0.5329,
+      "step": 520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958294446058815e-06,
+      "loss": 0.4949,
+      "step": 521
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95801352215329e-06,
+      "loss": 0.5334,
+      "step": 522
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957731659274208e-06,
+      "loss": 0.531,
+      "step": 523
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957448857474951e-06,
+      "loss": 0.5397,
+      "step": 524
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957165116809077e-06,
+      "loss": 0.5148,
+      "step": 525
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95688043733032e-06,
+      "loss": 0.5291,
+      "step": 526
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956594819092597e-06,
+      "loss": 0.5058,
+      "step": 527
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956308262149999e-06,
+      "loss": 0.5094,
+      "step": 528
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956020766556793e-06,
+      "loss": 0.5283,
+      "step": 529
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955732332367429e-06,
+      "loss": 0.5169,
+      "step": 530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955442959636533e-06,
+      "loss": 0.5472,
+      "step": 531
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955152648418906e-06,
+      "loss": 0.4867,
+      "step": 532
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954861398769529e-06,
+      "loss": 0.4893,
+      "step": 533
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95456921074356e-06,
+      "loss": 0.5062,
+      "step": 534
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954276084396337e-06,
+      "loss": 0.5159,
+      "step": 535
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95398201978337e-06,
+      "loss": 0.5357,
+      "step": 536
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953687016960355e-06,
+      "loss": 0.4952,
+      "step": 537
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953391075983157e-06,
+      "loss": 0.5074,
+      "step": 538
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953094196907825e-06,
+      "loss": 0.4949,
+      "step": 539
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952796379790583e-06,
+      "loss": 0.4908,
+      "step": 540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952497624687834e-06,
+      "loss": 0.4993,
+      "step": 541
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952197931656156e-06,
+      "loss": 0.5058,
+      "step": 542
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951897300752306e-06,
+      "loss": 0.4999,
+      "step": 543
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951595732033223e-06,
+      "loss": 0.5341,
+      "step": 544
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951293225556016e-06,
+      "loss": 0.5486,
+      "step": 545
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950989781377975e-06,
+      "loss": 0.5262,
+      "step": 546
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95068539955657e-06,
+      "loss": 0.5025,
+      "step": 547
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950380080149443e-06,
+      "loss": 0.488,
+      "step": 548
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95007382321442e-06,
+      "loss": 0.5621,
+      "step": 549
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949766628809502e-06,
+      "loss": 0.547,
+      "step": 550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949458496992865e-06,
+      "loss": 0.5227,
+      "step": 551
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949149427822865e-06,
+      "loss": 0.4902,
+      "step": 552
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948839421358036e-06,
+      "loss": 0.5035,
+      "step": 553
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948528477657086e-06,
+      "loss": 0.5229,
+      "step": 554
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948216596778906e-06,
+      "loss": 0.4994,
+      "step": 555
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947903778782561e-06,
+      "loss": 0.5129,
+      "step": 556
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947590023727292e-06,
+      "loss": 0.4664,
+      "step": 557
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947275331672522e-06,
+      "loss": 0.5145,
+      "step": 558
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946959702677848e-06,
+      "loss": 0.5224,
+      "step": 559
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946643136803046e-06,
+      "loss": 0.513,
+      "step": 560
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946325634108067e-06,
+      "loss": 0.5024,
+      "step": 561
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946007194653044e-06,
+      "loss": 0.4979,
+      "step": 562
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945687818498282e-06,
+      "loss": 0.5015,
+      "step": 563
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945367505704268e-06,
+      "loss": 0.4787,
+      "step": 564
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945046256331663e-06,
+      "loss": 0.5374,
+      "step": 565
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944724070441307e-06,
+      "loss": 0.5029,
+      "step": 566
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94440094809422e-06,
+      "loss": 0.4981,
+      "step": 567
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944076889351592e-06,
+      "loss": 0.5222,
+      "step": 568
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943751894274798e-06,
+      "loss": 0.5125,
+      "step": 569
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943425962925386e-06,
+      "loss": 0.5227,
+      "step": 570
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943099095365083e-06,
+      "loss": 0.5363,
+      "step": 571
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942771291655792e-06,
+      "loss": 0.5024,
+      "step": 572
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942442551859595e-06,
+      "loss": 0.5137,
+      "step": 573
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94211287603875e-06,
+      "loss": 0.5181,
+      "step": 574
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941782264255695e-06,
+      "loss": 0.5342,
+      "step": 575
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941450716573037e-06,
+      "loss": 0.4624,
+      "step": 576
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941118233053573e-06,
+      "loss": 0.5152,
+      "step": 577
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940784813760268e-06,
+      "loss": 0.4913,
+      "step": 578
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940450458756264e-06,
+      "loss": 0.5198,
+      "step": 579
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940115168104885e-06,
+      "loss": 0.5244,
+      "step": 580
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939778941869628e-06,
+      "loss": 0.5152,
+      "step": 581
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939441780114174e-06,
+      "loss": 0.5114,
+      "step": 582
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939103682902373e-06,
+      "loss": 0.4818,
+      "step": 583
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938764650298255e-06,
+      "loss": 0.5125,
+      "step": 584
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938424682366027e-06,
+      "loss": 0.494,
+      "step": 585
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93808377917008e-06,
+      "loss": 0.5464,
+      "step": 586
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937741940774968e-06,
+      "loss": 0.4941,
+      "step": 587
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937399167245435e-06,
+      "loss": 0.5267,
+      "step": 588
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937055458646395e-06,
+      "loss": 0.5047,
+      "step": 589
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936710815042942e-06,
+      "loss": 0.5058,
+      "step": 590
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936365236500347e-06,
+      "loss": 0.4923,
+      "step": 591
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936018723084057e-06,
+      "loss": 0.5195,
+      "step": 592
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935671274859695e-06,
+      "loss": 0.5079,
+      "step": 593
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935322891893064e-06,
+      "loss": 0.5195,
+      "step": 594
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934973574250142e-06,
+      "loss": 0.5361,
+      "step": 595
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934623321997087e-06,
+      "loss": 0.5023,
+      "step": 596
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934272135200227e-06,
+      "loss": 0.5375,
+      "step": 597
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933920013926075e-06,
+      "loss": 0.4938,
+      "step": 598
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933566958241315e-06,
+      "loss": 0.5347,
+      "step": 599
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933212968212812e-06,
+      "loss": 0.5231,
+      "step": 600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932858043907607e-06,
+      "loss": 0.5527,
+      "step": 601
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932502185392916e-06,
+      "loss": 0.4958,
+      "step": 602
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932145392736134e-06,
+      "loss": 0.5171,
+      "step": 603
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931787666004832e-06,
+      "loss": 0.5085,
+      "step": 604
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931429005266756e-06,
+      "loss": 0.5289,
+      "step": 605
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931069410589837e-06,
+      "loss": 0.5378,
+      "step": 606
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93070888204217e-06,
+      "loss": 0.5425,
+      "step": 607
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93034741969204e-06,
+      "loss": 0.5164,
+      "step": 608
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929985023607895e-06,
+      "loss": 0.5171,
+      "step": 609
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929621693858373e-06,
+      "loss": 0.4822,
+      "step": 610
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929257430512282e-06,
+      "loss": 0.5146,
+      "step": 611
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928892233638608e-06,
+      "loss": 0.5573,
+      "step": 612
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928526103306514e-06,
+      "loss": 0.4846,
+      "step": 613
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928159039585338e-06,
+      "loss": 0.5056,
+      "step": 614
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927791042544599e-06,
+      "loss": 0.5118,
+      "step": 615
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927422112253988e-06,
+      "loss": 0.5032,
+      "step": 616
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927052248783376e-06,
+      "loss": 0.4779,
+      "step": 617
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926681452202809e-06,
+      "loss": 0.4934,
+      "step": 618
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92630972258251e-06,
+      "loss": 0.515,
+      "step": 619
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925937059992881e-06,
+      "loss": 0.4918,
+      "step": 620
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925563464504497e-06,
+      "loss": 0.4747,
+      "step": 621
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92518893618811e-06,
+      "loss": 0.4986,
+      "step": 622
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924813475114654e-06,
+      "loss": 0.5245,
+      "step": 623
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924437081355232e-06,
+      "loss": 0.4706,
+      "step": 624
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924059754981128e-06,
+      "loss": 0.5001,
+      "step": 625
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923681496063805e-06,
+      "loss": 0.5214,
+      "step": 626
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923302304674896e-06,
+      "loss": 0.5429,
+      "step": 627
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.922922180886217e-06,
+      "loss": 0.5182,
+      "step": 628
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.922541124769753e-06,
+      "loss": 0.5447,
+      "step": 629
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.922159136397677e-06,
+      "loss": 0.5015,
+      "step": 630
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.921776215842326e-06,
+      "loss": 0.5031,
+      "step": 631
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.921392363176223e-06,
+      "loss": 0.4941,
+      "step": 632
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.921007578472064e-06,
+      "loss": 0.5202,
+      "step": 633
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920621861802719e-06,
+      "loss": 0.5064,
+      "step": 634
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920235213241238e-06,
+      "loss": 0.5197,
+      "step": 635
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91984763286085e-06,
+      "loss": 0.4909,
+      "step": 636
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919459120734948e-06,
+      "loss": 0.5184,
+      "step": 637
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91906967693712e-06,
+      "loss": 0.5188,
+      "step": 638
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.918679301541115e-06,
+      "loss": 0.483,
+      "step": 639
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.918287994620868e-06,
+      "loss": 0.4883,
+      "step": 640
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917895756250484e-06,
+      "loss": 0.5051,
+      "step": 641
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917502586504246e-06,
+      "loss": 0.4994,
+      "step": 642
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917108485456619e-06,
+      "loss": 0.4958,
+      "step": 643
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.916713453182236e-06,
+      "loss": 0.5155,
+      "step": 644
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91631748975591e-06,
+      "loss": 0.4985,
+      "step": 645
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915920595252635e-06,
+      "loss": 0.5344,
+      "step": 646
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915522769747573e-06,
+      "loss": 0.4986,
+      "step": 647
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915124013316066e-06,
+      "loss": 0.5052,
+      "step": 648
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.914724326033633e-06,
+      "loss": 0.5259,
+      "step": 649
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91432370797597e-06,
+      "loss": 0.4855,
+      "step": 650
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913922159218947e-06,
+      "loss": 0.5064,
+      "step": 651
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913519679838613e-06,
+      "loss": 0.5272,
+      "step": 652
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91311626991119e-06,
+      "loss": 0.5228,
+      "step": 653
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.912711929513076e-06,
+      "loss": 0.5068,
+      "step": 654
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.912306658720851e-06,
+      "loss": 0.5504,
+      "step": 655
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.911900457611265e-06,
+      "loss": 0.5058,
+      "step": 656
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.911493326261246e-06,
+      "loss": 0.523,
+      "step": 657
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.9110852647479e-06,
+      "loss": 0.5336,
+      "step": 658
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910676273148507e-06,
+      "loss": 0.5218,
+      "step": 659
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910266351540524e-06,
+      "loss": 0.5028,
+      "step": 660
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.909855500001584e-06,
+      "loss": 0.5057,
+      "step": 661
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.909443718609494e-06,
+      "loss": 0.512,
+      "step": 662
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.909031007442244e-06,
+      "loss": 0.5261,
+      "step": 663
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908617366577992e-06,
+      "loss": 0.5041,
+      "step": 664
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908202796095076e-06,
+      "loss": 0.4613,
+      "step": 665
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90778729607201e-06,
+      "loss": 0.5138,
+      "step": 666
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90737086658748e-06,
+      "loss": 0.5189,
+      "step": 667
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.906953507720358e-06,
+      "loss": 0.4945,
+      "step": 668
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90653521954968e-06,
+      "loss": 0.5356,
+      "step": 669
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.906116002154666e-06,
+      "loss": 0.4662,
+      "step": 670
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90569585561471e-06,
+      "loss": 0.4747,
+      "step": 671
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905274780009379e-06,
+      "loss": 0.5253,
+      "step": 672
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90485277541842e-06,
+      "loss": 0.5274,
+      "step": 673
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.904429841921755e-06,
+      "loss": 0.4961,
+      "step": 674
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.904005979599481e-06,
+      "loss": 0.5098,
+      "step": 675
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.903581188531868e-06,
+      "loss": 0.5377,
+      "step": 676
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90315546879937e-06,
+      "loss": 0.5238,
+      "step": 677
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90272882048261e-06,
+      "loss": 0.51,
+      "step": 678
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.902301243662386e-06,
+      "loss": 0.5004,
+      "step": 679
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90187273841968e-06,
+      "loss": 0.4935,
+      "step": 680
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90144330483564e-06,
+      "loss": 0.501,
+      "step": 681
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.901012942991597e-06,
+      "loss": 0.5263,
+      "step": 682
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.900581652969055e-06,
+      "loss": 0.5395,
+      "step": 683
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90014943484969e-06,
+      "loss": 0.5232,
+      "step": 684
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899716288715363e-06,
+      "loss": 0.5417,
+      "step": 685
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.8992822146481e-06,
+      "loss": 0.5335,
+      "step": 686
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.898847212730113e-06,
+      "loss": 0.5264,
+      "step": 687
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.898411283043784e-06,
+      "loss": 0.5083,
+      "step": 688
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89797442567167e-06,
+      "loss": 0.5087,
+      "step": 689
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.897536640696505e-06,
+      "loss": 0.5747,
+      "step": 690
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.8970979282012e-06,
+      "loss": 0.4871,
+      "step": 691
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89665828826884e-06,
+      "loss": 0.5121,
+      "step": 692
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.896217720982686e-06,
+      "loss": 0.5111,
+      "step": 693
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.895776226426176e-06,
+      "loss": 0.4903,
+      "step": 694
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.895333804682923e-06,
+      "loss": 0.53,
+      "step": 695
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.894890455836712e-06,
+      "loss": 0.4994,
+      "step": 696
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.894446179971508e-06,
+      "loss": 0.488,
+      "step": 697
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.894000977171452e-06,
+      "loss": 0.4932,
+      "step": 698
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.893554847520857e-06,
+      "loss": 0.5332,
+      "step": 699
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.893107791104213e-06,
+      "loss": 0.525,
+      "step": 700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.892659808006188e-06,
+      "loss": 0.5148,
+      "step": 701
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.892210898311621e-06,
+      "loss": 0.4877,
+      "step": 702
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.891761062105529e-06,
+      "loss": 0.5439,
+      "step": 703
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.891310299473105e-06,
+      "loss": 0.5252,
+      "step": 704
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890858610499717e-06,
+      "loss": 0.4776,
+      "step": 705
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890405995270907e-06,
+      "loss": 0.5307,
+      "step": 706
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.889952453872395e-06,
+      "loss": 0.4785,
+      "step": 707
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.889497986390073e-06,
+      "loss": 0.4928,
+      "step": 708
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.889042592910013e-06,
+      "loss": 0.5162,
+      "step": 709
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.888586273518457e-06,
+      "loss": 0.4819,
+      "step": 710
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.888129028301828e-06,
+      "loss": 0.4947,
+      "step": 711
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.887670857346717e-06,
+      "loss": 0.5082,
+      "step": 712
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.887211760739898e-06,
+      "loss": 0.5169,
+      "step": 713
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.886751738568318e-06,
+      "loss": 0.523,
+      "step": 714
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.886290790919095e-06,
+      "loss": 0.4946,
+      "step": 715
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88582891787953e-06,
+      "loss": 0.5083,
+      "step": 716
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.885366119537088e-06,
+      "loss": 0.5025,
+      "step": 717
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.884902395979424e-06,
+      "loss": 0.5152,
+      "step": 718
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.884437747294353e-06,
+      "loss": 0.5119,
+      "step": 719
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.883972173569877e-06,
+      "loss": 0.4882,
+      "step": 720
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88350567489417e-06,
+      "loss": 0.5281,
+      "step": 721
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.883038251355576e-06,
+      "loss": 0.4761,
+      "step": 722
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88256990304262e-06,
+      "loss": 0.5102,
+      "step": 723
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.882100630044e-06,
+      "loss": 0.4869,
+      "step": 724
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.881630432448588e-06,
+      "loss": 0.4885,
+      "step": 725
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.881159310345434e-06,
+      "loss": 0.5403,
+      "step": 726
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.880687263823761e-06,
+      "loss": 0.5083,
+      "step": 727
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.880214292972968e-06,
+      "loss": 0.4703,
+      "step": 728
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.879740397882626e-06,
+      "loss": 0.4891,
+      "step": 729
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.879265578642489e-06,
+      "loss": 0.5203,
+      "step": 730
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.878789835342476e-06,
+      "loss": 0.5182,
+      "step": 731
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.878313168072686e-06,
+      "loss": 0.524,
+      "step": 732
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.877835576923393e-06,
+      "loss": 0.5008,
+      "step": 733
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.877357061985048e-06,
+      "loss": 0.52,
+      "step": 734
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.876877623348272e-06,
+      "loss": 0.5059,
+      "step": 735
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.876397261103865e-06,
+      "loss": 0.5186,
+      "step": 736
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.8759159753428e-06,
+      "loss": 0.5283,
+      "step": 737
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.875433766156224e-06,
+      "loss": 0.5195,
+      "step": 738
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87495063363546e-06,
+      "loss": 0.5105,
+      "step": 739
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.874466577872011e-06,
+      "loss": 0.5398,
+      "step": 740
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.873981598957543e-06,
+      "loss": 0.483,
+      "step": 741
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87349569698391e-06,
+      "loss": 0.4986,
+      "step": 742
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87300887204313e-06,
+      "loss": 0.4893,
+      "step": 743
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.8725211242274e-06,
+      "loss": 0.5135,
+      "step": 744
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.872032453629097e-06,
+      "loss": 0.4967,
+      "step": 745
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.871542860340763e-06,
+      "loss": 0.4964,
+      "step": 746
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.871052344455123e-06,
+      "loss": 0.4996,
+      "step": 747
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.870560906065073e-06,
+      "loss": 0.5086,
+      "step": 748
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.870068545263682e-06,
+      "loss": 0.5079,
+      "step": 749
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.869575262144196e-06,
+      "loss": 0.5213,
+      "step": 750
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.869081056800035e-06,
+      "loss": 0.5293,
+      "step": 751
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.868585929324798e-06,
+      "loss": 0.5352,
+      "step": 752
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.868089879812251e-06,
+      "loss": 0.5459,
+      "step": 753
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86759290835634e-06,
+      "loss": 0.5417,
+      "step": 754
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.867095015051183e-06,
+      "loss": 0.5246,
+      "step": 755
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.866596199991074e-06,
+      "loss": 0.5201,
+      "step": 756
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.866096463270482e-06,
+      "loss": 0.5278,
+      "step": 757
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.865595804984047e-06,
+      "loss": 0.539,
+      "step": 758
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.865094225226588e-06,
+      "loss": 0.5061,
+      "step": 759
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.864591724093097e-06,
+      "loss": 0.5177,
+      "step": 760
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.864088301678741e-06,
+      "loss": 0.5272,
+      "step": 761
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.863583958078858e-06,
+      "loss": 0.5178,
+      "step": 762
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.863078693388965e-06,
+      "loss": 0.5187,
+      "step": 763
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.862572507704751e-06,
+      "loss": 0.5095,
+      "step": 764
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86206540112208e-06,
+      "loss": 0.5172,
+      "step": 765
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.861557373736992e-06,
+      "loss": 0.4867,
+      "step": 766
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.861048425645697e-06,
+      "loss": 0.5014,
+      "step": 767
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.860538556944584e-06,
+      "loss": 0.502,
+      "step": 768
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.860027767730214e-06,
+      "loss": 0.5399,
+      "step": 769
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.859516058099323e-06,
+      "loss": 0.534,
+      "step": 770
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85900342814882e-06,
+      "loss": 0.5193,
+      "step": 771
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.858489877975791e-06,
+      "loss": 0.5313,
+      "step": 772
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.857975407677494e-06,
+      "loss": 0.5189,
+      "step": 773
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85746001735136e-06,
+      "loss": 0.4946,
+      "step": 774
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.856943707095003e-06,
+      "loss": 0.5142,
+      "step": 775
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.856426477006196e-06,
+      "loss": 0.496,
+      "step": 776
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.8559083271829e-06,
+      "loss": 0.4958,
+      "step": 777
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.855389257723242e-06,
+      "loss": 0.545,
+      "step": 778
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.854869268725529e-06,
+      "loss": 0.4867,
+      "step": 779
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.854348360288237e-06,
+      "loss": 0.5289,
+      "step": 780
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.853826532510017e-06,
+      "loss": 0.4768,
+      "step": 781
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.8533037854897e-06,
+      "loss": 0.4774,
+      "step": 782
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.85278011932628e-06,
+      "loss": 0.5121,
+      "step": 783
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.852255534118935e-06,
+      "loss": 0.4759,
+      "step": 784
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.851730029967016e-06,
+      "loss": 0.4902,
+      "step": 785
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.851203606970042e-06,
+      "loss": 0.5193,
+      "step": 786
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.85067626522771e-06,
+      "loss": 0.5132,
+      "step": 787
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.85014800483989e-06,
+      "loss": 0.5383,
+      "step": 788
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.849618825906632e-06,
+      "loss": 0.4957,
+      "step": 789
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.84908872852815e-06,
+      "loss": 0.5023,
+      "step": 790
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.848557712804834e-06,
+      "loss": 0.5024,
+      "step": 791
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.848025778837255e-06,
+      "loss": 0.5074,
+      "step": 792
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.847492926726152e-06,
+      "loss": 0.5119,
+      "step": 793
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.84695915657244e-06,
+      "loss": 0.5058,
+      "step": 794
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.846424468477204e-06,
+      "loss": 0.4931,
+      "step": 795
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.845888862541711e-06,
+      "loss": 0.5192,
+      "step": 796
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.845352338867393e-06,
+      "loss": 0.4897,
+      "step": 797
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.844814897555858e-06,
+      "loss": 0.5172,
+      "step": 798
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.844276538708895e-06,
+      "loss": 0.5129,
+      "step": 799
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.843737262428457e-06,
+      "loss": 0.5017,
+      "step": 800
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.843197068816674e-06,
+      "loss": 0.4863,
+      "step": 801
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.842655957975856e-06,
+      "loss": 0.5389,
+      "step": 802
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.842113930008473e-06,
+      "loss": 0.5013,
+      "step": 803
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.841570985017183e-06,
+      "loss": 0.5081,
+      "step": 804
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.841027123104812e-06,
+      "loss": 0.544,
+      "step": 805
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.840482344374356e-06,
+      "loss": 0.4725,
+      "step": 806
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83993664892899e-06,
+      "loss": 0.5026,
+      "step": 807
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.839390036872061e-06,
+      "loss": 0.5332,
+      "step": 808
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.838842508307085e-06,
+      "loss": 0.5197,
+      "step": 809
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.838294063337762e-06,
+      "loss": 0.5183,
+      "step": 810
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.837744702067955e-06,
+      "loss": 0.51,
+      "step": 811
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.837194424601706e-06,
+      "loss": 0.5071,
+      "step": 812
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83664323104323e-06,
+      "loss": 0.5231,
+      "step": 813
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.836091121496913e-06,
+      "loss": 0.5199,
+      "step": 814
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.835538096067317e-06,
+      "loss": 0.5043,
+      "step": 815
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83498415485918e-06,
+      "loss": 0.4934,
+      "step": 816
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.834429297977403e-06,
+      "loss": 0.4682,
+      "step": 817
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.833873525527074e-06,
+      "loss": 0.5061,
+      "step": 818
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.833316837613446e-06,
+      "loss": 0.5207,
+      "step": 819
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.832759234341946e-06,
+      "loss": 0.5156,
+      "step": 820
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83220071581818e-06,
+      "loss": 0.4773,
+      "step": 821
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.831641282147916e-06,
+      "loss": 0.5097,
+      "step": 822
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83108093343711e-06,
+      "loss": 0.5232,
+      "step": 823
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.830519669791878e-06,
+      "loss": 0.4836,
+      "step": 824
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.829957491318517e-06,
+      "loss": 0.5132,
+      "step": 825
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.829394398123495e-06,
+      "loss": 0.4819,
+      "step": 826
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.828830390313455e-06,
+      "loss": 0.4931,
+      "step": 827
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.828265467995208e-06,
+      "loss": 0.4806,
+      "step": 828
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.827699631275744e-06,
+      "loss": 0.4678,
+      "step": 829
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.827132880262226e-06,
+      "loss": 0.5295,
+      "step": 830
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.826565215061985e-06,
+      "loss": 0.5298,
+      "step": 831
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82599663578253e-06,
+      "loss": 0.5174,
+      "step": 832
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82542714253154e-06,
+      "loss": 0.5134,
+      "step": 833
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82485673541687e-06,
+      "loss": 0.4966,
+      "step": 834
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.824285414546546e-06,
+      "loss": 0.465,
+      "step": 835
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.823713180028768e-06,
+      "loss": 0.4909,
+      "step": 836
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.823140031971908e-06,
+      "loss": 0.4905,
+      "step": 837
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.822565970484512e-06,
+      "loss": 0.5268,
+      "step": 838
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.821990995675296e-06,
+      "loss": 0.5169,
+      "step": 839
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.821415107653158e-06,
+      "loss": 0.5031,
+      "step": 840
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.820838306527156e-06,
+      "loss": 0.5081,
+      "step": 841
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.820260592406533e-06,
+      "loss": 0.4645,
+      "step": 842
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.819681965400694e-06,
+      "loss": 0.5078,
+      "step": 843
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.819102425619225e-06,
+      "loss": 0.5093,
+      "step": 844
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.818521973171884e-06,
+      "loss": 0.5132,
+      "step": 845
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.817940608168596e-06,
+      "loss": 0.4692,
+      "step": 846
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.817358330719466e-06,
+      "loss": 0.5099,
+      "step": 847
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.816775140934768e-06,
+      "loss": 0.4818,
+      "step": 848
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.816191038924948e-06,
+      "loss": 0.5028,
+      "step": 849
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.815606024800628e-06,
+      "loss": 0.4981,
+      "step": 850
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.815020098672599e-06,
+      "loss": 0.4918,
+      "step": 851
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.814433260651829e-06,
+      "loss": 0.4952,
+      "step": 852
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.813845510849454e-06,
+      "loss": 0.4947,
+      "step": 853
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.813256849376787e-06,
+      "loss": 0.521,
+      "step": 854
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.812667276345312e-06,
+      "loss": 0.5001,
+      "step": 855
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.81207679186668e-06,
+      "loss": 0.5326,
+      "step": 856
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.811485396052726e-06,
+      "loss": 0.4849,
+      "step": 857
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.810893089015452e-06,
+      "loss": 0.5002,
+      "step": 858
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.810299870867027e-06,
+      "loss": 0.5168,
+      "step": 859
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.8097057417198e-06,
+      "loss": 0.5214,
+      "step": 860
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.809110701686291e-06,
+      "loss": 0.52,
+      "step": 861
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.808514750879193e-06,
+      "loss": 0.5332,
+      "step": 862
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.807917889411366e-06,
+      "loss": 0.4703,
+      "step": 863
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.807320117395849e-06,
+      "loss": 0.4909,
+      "step": 864
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.806721434945851e-06,
+      "loss": 0.5002,
+      "step": 865
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.806121842174754e-06,
+      "loss": 0.5084,
+      "step": 866
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.805521339196111e-06,
+      "loss": 0.5092,
+      "step": 867
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.804919926123648e-06,
+      "loss": 0.5324,
+      "step": 868
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.804317603071265e-06,
+      "loss": 0.5018,
+      "step": 869
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.803714370153033e-06,
+      "loss": 0.4894,
+      "step": 870
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.803110227483192e-06,
+      "loss": 0.5147,
+      "step": 871
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.802505175176163e-06,
+      "loss": 0.4956,
+      "step": 872
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.801899213346528e-06,
+      "loss": 0.4927,
+      "step": 873
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.801292342109052e-06,
+      "loss": 0.5188,
+      "step": 874
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.800684561578664e-06,
+      "loss": 0.4949,
+      "step": 875
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.800075871870472e-06,
+      "loss": 0.5392,
+      "step": 876
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79946627309975e-06,
+      "loss": 0.499,
+      "step": 877
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.798855765381945e-06,
+      "loss": 0.5106,
+      "step": 878
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.798244348832683e-06,
+      "loss": 0.4977,
+      "step": 879
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.797632023567753e-06,
+      "loss": 0.47,
+      "step": 880
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.797018789703124e-06,
+      "loss": 0.5118,
+      "step": 881
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.796404647354931e-06,
+      "loss": 0.5013,
+      "step": 882
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.795789596639483e-06,
+      "loss": 0.5094,
+      "step": 883
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79517363767326e-06,
+      "loss": 0.4718,
+      "step": 884
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79455677057292e-06,
+      "loss": 0.5021,
+      "step": 885
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.793938995455285e-06,
+      "loss": 0.5211,
+      "step": 886
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.793320312437353e-06,
+      "loss": 0.4925,
+      "step": 887
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.792700721636294e-06,
+      "loss": 0.5244,
+      "step": 888
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.792080223169449e-06,
+      "loss": 0.4804,
+      "step": 889
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.791458817154329e-06,
+      "loss": 0.5367,
+      "step": 890
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.790836503708623e-06,
+      "loss": 0.5387,
+      "step": 891
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.790213282950184e-06,
+      "loss": 0.476,
+      "step": 892
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.789589154997043e-06,
+      "loss": 0.5321,
+      "step": 893
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.7889641199674e-06,
+      "loss": 0.5061,
+      "step": 894
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.788338177979626e-06,
+      "loss": 0.4997,
+      "step": 895
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.787711329152268e-06,
+      "loss": 0.5257,
+      "step": 896
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.787083573604039e-06,
+      "loss": 0.503,
+      "step": 897
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.786454911453827e-06,
+      "loss": 0.516,
+      "step": 898
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.785825342820692e-06,
+      "loss": 0.4924,
+      "step": 899
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.785194867823867e-06,
+      "loss": 0.4903,
+      "step": 900
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.784563486582747e-06,
+      "loss": 0.4989,
+      "step": 901
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.783931199216916e-06,
+      "loss": 0.5289,
+      "step": 902
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.78329800584611e-06,
+      "loss": 0.5052,
+      "step": 903
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.782663906590255e-06,
+      "loss": 0.4881,
+      "step": 904
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.782028901569437e-06,
+      "loss": 0.4898,
+      "step": 905
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.781392990903915e-06,
+      "loss": 0.5265,
+      "step": 906
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.780756174714119e-06,
+      "loss": 0.5119,
+      "step": 907
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.780118453120659e-06,
+      "loss": 0.5107,
+      "step": 908
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.779479826244302e-06,
+      "loss": 0.471,
+      "step": 909
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.778840294206003e-06,
+      "loss": 0.4877,
+      "step": 910
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77819985712687e-06,
+      "loss": 0.506,
+      "step": 911
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.7775585151282e-06,
+      "loss": 0.5197,
+      "step": 912
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.776916268331452e-06,
+      "loss": 0.4622,
+      "step": 913
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.776273116858255e-06,
+      "loss": 0.5049,
+      "step": 914
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.775629060830416e-06,
+      "loss": 0.5275,
+      "step": 915
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.774984100369906e-06,
+      "loss": 0.5267,
+      "step": 916
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.774338235598875e-06,
+      "loss": 0.4952,
+      "step": 917
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.773691466639636e-06,
+      "loss": 0.4894,
+      "step": 918
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.773043793614678e-06,
+      "loss": 0.5356,
+      "step": 919
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.772395216646663e-06,
+      "loss": 0.5209,
+      "step": 920
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77174573585842e-06,
+      "loss": 0.4937,
+      "step": 921
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.771095351372953e-06,
+      "loss": 0.4759,
+      "step": 922
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.770444063313432e-06,
+      "loss": 0.5115,
+      "step": 923
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.769791871803202e-06,
+      "loss": 0.5202,
+      "step": 924
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.76913877696578e-06,
+      "loss": 0.4996,
+      "step": 925
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.768484778924852e-06,
+      "loss": 0.4991,
+      "step": 926
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.767829877804273e-06,
+      "loss": 0.4957,
+      "step": 927
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.767174073728075e-06,
+      "loss": 0.4943,
+      "step": 928
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.766517366820456e-06,
+      "loss": 0.5363,
+      "step": 929
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.765859757205785e-06,
+      "loss": 0.5222,
+      "step": 930
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.765201245008604e-06,
+      "loss": 0.5389,
+      "step": 931
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.764541830353627e-06,
+      "loss": 0.4983,
+      "step": 932
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.763881513365735e-06,
+      "loss": 0.5098,
+      "step": 933
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.763220294169987e-06,
+      "loss": 0.4775,
+      "step": 934
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.7625581728916e-06,
+      "loss": 0.5262,
+      "step": 935
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.761895149655977e-06,
+      "loss": 0.4805,
+      "step": 936
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.76123122458868e-06,
+      "loss": 0.4856,
+      "step": 937
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.760566397815449e-06,
+      "loss": 0.5082,
+      "step": 938
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.759900669462193e-06,
+      "loss": 0.4937,
+      "step": 939
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.75923403965499e-06,
+      "loss": 0.5165,
+      "step": 940
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.758566508520088e-06,
+      "loss": 0.465,
+      "step": 941
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.75789807618391e-06,
+      "loss": 0.4891,
+      "step": 942
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.757228742773044e-06,
+      "loss": 0.5345,
+      "step": 943
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.756558508414255e-06,
+      "loss": 0.4804,
+      "step": 944
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.755887373234474e-06,
+      "loss": 0.5094,
+      "step": 945
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.755215337360802e-06,
+      "loss": 0.5136,
+      "step": 946
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.754542400920517e-06,
+      "loss": 0.5049,
+      "step": 947
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.75386856404106e-06,
+      "loss": 0.4644,
+      "step": 948
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.753193826850045e-06,
+      "loss": 0.5157,
+      "step": 949
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.752518189475258e-06,
+      "loss": 0.499,
+      "step": 950
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.751841652044656e-06,
+      "loss": 0.486,
+      "step": 951
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.751164214686364e-06,
+      "loss": 0.5306,
+      "step": 952
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.750485877528678e-06,
+      "loss": 0.4898,
+      "step": 953
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.749806640700066e-06,
+      "loss": 0.4857,
+      "step": 954
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.749126504329162e-06,
+      "loss": 0.4848,
+      "step": 955
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.74844546854478e-06,
+      "loss": 0.4786,
+      "step": 956
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.747763533475893e-06,
+      "loss": 0.5206,
+      "step": 957
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.74708069925165e-06,
+      "loss": 0.5126,
+      "step": 958
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.74639696600137e-06,
+      "loss": 0.4925,
+      "step": 959
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.745712333854543e-06,
+      "loss": 0.5298,
+      "step": 960
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.745026802940827e-06,
+      "loss": 0.5193,
+      "step": 961
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.74434037339005e-06,
+      "loss": 0.4935,
+      "step": 962
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.743653045332213e-06,
+      "loss": 0.4738,
+      "step": 963
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.742964818897487e-06,
+      "loss": 0.4941,
+      "step": 964
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.74227569421621e-06,
+      "loss": 0.5154,
+      "step": 965
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.741585671418891e-06,
+      "loss": 0.5076,
+      "step": 966
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.740894750636212e-06,
+      "loss": 0.4724,
+      "step": 967
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.740202931999023e-06,
+      "loss": 0.4892,
+      "step": 968
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.739510215638342e-06,
+      "loss": 0.5385,
+      "step": 969
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.73881660168536e-06,
+      "loss": 0.5084,
+      "step": 970
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.73812209027144e-06,
+      "loss": 0.5136,
+      "step": 971
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.737426681528107e-06,
+      "loss": 0.5058,
+      "step": 972
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.736730375587065e-06,
+      "loss": 0.5047,
+      "step": 973
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.736033172580181e-06,
+      "loss": 0.5391,
+      "step": 974
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.735335072639498e-06,
+      "loss": 0.5045,
+      "step": 975
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.734636075897223e-06,
+      "loss": 0.483,
+      "step": 976
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.733936182485739e-06,
+      "loss": 0.5059,
+      "step": 977
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.73323539253759e-06,
+      "loss": 0.4734,
+      "step": 978
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.732533706185499e-06,
+      "loss": 0.5199,
+      "step": 979
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.731831123562355e-06,
+      "loss": 0.5051,
+      "step": 980
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.731127644801215e-06,
+      "loss": 0.5112,
+      "step": 981
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.730423270035307e-06,
+      "loss": 0.481,
+      "step": 982
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.72971799939803e-06,
+      "loss": 0.5187,
+      "step": 983
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.729011833022952e-06,
+      "loss": 0.4979,
+      "step": 984
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.728304771043811e-06,
+      "loss": 0.5054,
+      "step": 985
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.727596813594511e-06,
+      "loss": 0.486,
+      "step": 986
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.726887960809131e-06,
+      "loss": 0.5009,
+      "step": 987
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.726178212821917e-06,
+      "loss": 0.5139,
+      "step": 988
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.725467569767285e-06,
+      "loss": 0.5117,
+      "step": 989
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.724756031779819e-06,
+      "loss": 0.5394,
+      "step": 990
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.724043598994273e-06,
+      "loss": 0.5045,
+      "step": 991
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.723330271545574e-06,
+      "loss": 0.4903,
+      "step": 992
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.722616049568811e-06,
+      "loss": 0.5077,
+      "step": 993
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.721900933199253e-06,
+      "loss": 0.5096,
+      "step": 994
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.721184922572327e-06,
+      "loss": 0.5151,
+      "step": 995
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.720468017823637e-06,
+      "loss": 0.4868,
+      "step": 996
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.719750219088954e-06,
+      "loss": 0.5165,
+      "step": 997
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.719031526504217e-06,
+      "loss": 0.5134,
+      "step": 998
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.718311940205538e-06,
+      "loss": 0.5131,
+      "step": 999
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.717591460329195e-06,
+      "loss": 0.5028,
+      "step": 1000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.716870087011634e-06,
+      "loss": 0.5095,
+      "step": 1001
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.716147820389474e-06,
+      "loss": 0.5202,
+      "step": 1002
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.715424660599502e-06,
+      "loss": 0.5545,
+      "step": 1003
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.714700607778673e-06,
+      "loss": 0.5113,
+      "step": 1004
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.713975662064112e-06,
+      "loss": 0.5039,
+      "step": 1005
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.713249823593112e-06,
+      "loss": 0.4981,
+      "step": 1006
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.712523092503139e-06,
+      "loss": 0.5224,
+      "step": 1007
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.71179546893182e-06,
+      "loss": 0.4967,
+      "step": 1008
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.71106695301696e-06,
+      "loss": 0.5295,
+      "step": 1009
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.710337544896526e-06,
+      "loss": 0.5221,
+      "step": 1010
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.70960724470866e-06,
+      "loss": 0.4918,
+      "step": 1011
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.70887605259167e-06,
+      "loss": 0.4811,
+      "step": 1012
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.70814396868403e-06,
+      "loss": 0.5152,
+      "step": 1013
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.707410993124386e-06,
+      "loss": 0.5016,
+      "step": 1014
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.706677126051555e-06,
+      "loss": 0.4777,
+      "step": 1015
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.705942367604521e-06,
+      "loss": 0.509,
+      "step": 1016
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.705206717922434e-06,
+      "loss": 0.4862,
+      "step": 1017
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.704470177144611e-06,
+      "loss": 0.5098,
+      "step": 1018
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.70373274541055e-06,
+      "loss": 0.4833,
+      "step": 1019
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.702994422859907e-06,
+      "loss": 0.4992,
+      "step": 1020
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.702255209632505e-06,
+      "loss": 0.5006,
+      "step": 1021
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.701515105868344e-06,
+      "loss": 0.4777,
+      "step": 1022
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.700774111707589e-06,
+      "loss": 0.5308,
+      "step": 1023
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.70003222729057e-06,
+      "loss": 0.5101,
+      "step": 1024
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.699289452757791e-06,
+      "loss": 0.4759,
+      "step": 1025
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.698545788249923e-06,
+      "loss": 0.5391,
+      "step": 1026
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.697801233907804e-06,
+      "loss": 0.4971,
+      "step": 1027
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.697055789872441e-06,
+      "loss": 0.5465,
+      "step": 1028
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.696309456285009e-06,
+      "loss": 0.4717,
+      "step": 1029
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.695562233286855e-06,
+      "loss": 0.5239,
+      "step": 1030
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.694814121019491e-06,
+      "loss": 0.4923,
+      "step": 1031
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.694065119624596e-06,
+      "loss": 0.5221,
+      "step": 1032
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.693315229244023e-06,
+      "loss": 0.5326,
+      "step": 1033
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.692564450019788e-06,
+      "loss": 0.4689,
+      "step": 1034
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.691812782094076e-06,
+      "loss": 0.5322,
+      "step": 1035
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.691060225609245e-06,
+      "loss": 0.5083,
+      "step": 1036
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.690306780707817e-06,
+      "loss": 0.5027,
+      "step": 1037
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.689552447532483e-06,
+      "loss": 0.5269,
+      "step": 1038
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.688797226226101e-06,
+      "loss": 0.4936,
+      "step": 1039
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.6880411169317e-06,
+      "loss": 0.5065,
+      "step": 1040
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.687284119792476e-06,
+      "loss": 0.5175,
+      "step": 1041
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.686526234951792e-06,
+      "loss": 0.4625,
+      "step": 1042
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.685767462553179e-06,
+      "loss": 0.4647,
+      "step": 1043
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.685007802740341e-06,
+      "loss": 0.4923,
+      "step": 1044
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.684247255657142e-06,
+      "loss": 0.524,
+      "step": 1045
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.68348582144762e-06,
+      "loss": 0.5085,
+      "step": 1046
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.682723500255979e-06,
+      "loss": 0.5306,
+      "step": 1047
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.681960292226593e-06,
+      "loss": 0.4568,
+      "step": 1048
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.681196197504e-06,
+      "loss": 0.509,
+      "step": 1049
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.680431216232908e-06,
+      "loss": 0.4866,
+      "step": 1050
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.679665348558194e-06,
+      "loss": 0.4988,
+      "step": 1051
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.6788985946249e-06,
+      "loss": 0.4907,
+      "step": 1052
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.678130954578242e-06,
+      "loss": 0.4939,
+      "step": 1053
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.677362428563597e-06,
+      "loss": 0.5128,
+      "step": 1054
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.67659301672651e-06,
+      "loss": 0.4866,
+      "step": 1055
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.675822719212699e-06,
+      "loss": 0.488,
+      "step": 1056
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.675051536168047e-06,
+      "loss": 0.5219,
+      "step": 1057
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.674279467738603e-06,
+      "loss": 0.4477,
+      "step": 1058
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.673506514070585e-06,
+      "loss": 0.5046,
+      "step": 1059
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.67273267531038e-06,
+      "loss": 0.5061,
+      "step": 1060
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.671957951604542e-06,
+      "loss": 0.5013,
+      "step": 1061
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.671182343099791e-06,
+      "loss": 0.4775,
+      "step": 1062
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.670405849943017e-06,
+      "loss": 0.518,
+      "step": 1063
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.669628472281273e-06,
+      "loss": 0.4943,
+      "step": 1064
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.668850210261783e-06,
+      "loss": 0.5264,
+      "step": 1065
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.668071064031944e-06,
+      "loss": 0.5013,
+      "step": 1066
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.66729103373931e-06,
+      "loss": 0.5316,
+      "step": 1067
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.666510119531606e-06,
+      "loss": 0.4786,
+      "step": 1068
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.665728321556727e-06,
+      "loss": 0.4875,
+      "step": 1069
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.664945639962736e-06,
+      "loss": 0.5309,
+      "step": 1070
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.664162074897857e-06,
+      "loss": 0.4922,
+      "step": 1071
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.663377626510488e-06,
+      "loss": 0.5126,
+      "step": 1072
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.662592294949191e-06,
+      "loss": 0.5181,
+      "step": 1073
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.661806080362697e-06,
+      "loss": 0.5352,
+      "step": 1074
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.661018982899904e-06,
+      "loss": 0.5192,
+      "step": 1075
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.660231002709873e-06,
+      "loss": 0.4761,
+      "step": 1076
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.65944213994184e-06,
+      "loss": 0.5253,
+      "step": 1077
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.6586523947452e-06,
+      "loss": 0.5046,
+      "step": 1078
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.657861767269521e-06,
+      "loss": 0.4769,
+      "step": 1079
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.657070257664537e-06,
+      "loss": 0.5299,
+      "step": 1080
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.656277866080143e-06,
+      "loss": 0.4725,
+      "step": 1081
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.655484592666413e-06,
+      "loss": 0.5289,
+      "step": 1082
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.654690437573577e-06,
+      "loss": 0.5085,
+      "step": 1083
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.653895400952037e-06,
+      "loss": 0.4848,
+      "step": 1084
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.65309948295236e-06,
+      "loss": 0.4979,
+      "step": 1085
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.652302683725283e-06,
+      "loss": 0.5104,
+      "step": 1086
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.651505003421706e-06,
+      "loss": 0.4939,
+      "step": 1087
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.650706442192698e-06,
+      "loss": 0.4838,
+      "step": 1088
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.649907000189496e-06,
+      "loss": 0.4943,
+      "step": 1089
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.649106677563503e-06,
+      "loss": 0.5,
+      "step": 1090
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.648305474466284e-06,
+      "loss": 0.5002,
+      "step": 1091
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.647503391049579e-06,
+      "loss": 0.5043,
+      "step": 1092
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.646700427465287e-06,
+      "loss": 0.492,
+      "step": 1093
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.645896583865482e-06,
+      "loss": 0.4885,
+      "step": 1094
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.645091860402396e-06,
+      "loss": 0.4974,
+      "step": 1095
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.644286257228433e-06,
+      "loss": 0.5237,
+      "step": 1096
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.64347977449616e-06,
+      "loss": 0.4764,
+      "step": 1097
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.642672412358317e-06,
+      "loss": 0.4559,
+      "step": 1098
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.641864170967803e-06,
+      "loss": 0.5137,
+      "step": 1099
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.641055050477689e-06,
+      "loss": 0.5179,
+      "step": 1100
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.640245051041209e-06,
+      "loss": 0.5217,
+      "step": 1101
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.639434172811763e-06,
+      "loss": 0.5108,
+      "step": 1102
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.638622415942923e-06,
+      "loss": 0.5069,
+      "step": 1103
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.637809780588423e-06,
+      "loss": 0.5033,
+      "step": 1104
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.63699626690216e-06,
+      "loss": 0.5172,
+      "step": 1105
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.636181875038205e-06,
+      "loss": 0.5139,
+      "step": 1106
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.635366605150791e-06,
+      "loss": 0.6141,
+      "step": 1107
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.634550457394318e-06,
+      "loss": 0.5143,
+      "step": 1108
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.633733431923351e-06,
+      "loss": 0.5175,
+      "step": 1109
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.632915528892624e-06,
+      "loss": 0.4938,
+      "step": 1110
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.632096748457035e-06,
+      "loss": 0.5134,
+      "step": 1111
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.631277090771646e-06,
+      "loss": 0.5282,
+      "step": 1112
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.630456555991691e-06,
+      "loss": 0.5235,
+      "step": 1113
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.629635144272568e-06,
+      "loss": 0.5126,
+      "step": 1114
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.628812855769838e-06,
+      "loss": 0.4916,
+      "step": 1115
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.62798969063923e-06,
+      "loss": 0.4795,
+      "step": 1116
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.62716564903664e-06,
+      "loss": 0.4999,
+      "step": 1117
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.62634073111813e-06,
+      "loss": 0.5093,
+      "step": 1118
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.625514937039926e-06,
+      "loss": 0.4946,
+      "step": 1119
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.62468826695842e-06,
+      "loss": 0.4981,
+      "step": 1120
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.623860721030174e-06,
+      "loss": 0.4814,
+      "step": 1121
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.623032299411908e-06,
+      "loss": 0.512,
+      "step": 1122
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.622203002260516e-06,
+      "loss": 0.4897,
+      "step": 1123
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.621372829733056e-06,
+      "loss": 0.5106,
+      "step": 1124
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.620541781986747e-06,
+      "loss": 0.4968,
+      "step": 1125
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.619709859178978e-06,
+      "loss": 0.5272,
+      "step": 1126
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.618877061467305e-06,
+      "loss": 0.4883,
+      "step": 1127
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.618043389009443e-06,
+      "loss": 0.4893,
+      "step": 1128
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.617208841963279e-06,
+      "loss": 0.5128,
+      "step": 1129
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.616373420486867e-06,
+      "loss": 0.5,
+      "step": 1130
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.615537124738418e-06,
+      "loss": 0.5321,
+      "step": 1131
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.614699954876317e-06,
+      "loss": 0.5159,
+      "step": 1132
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.613861911059112e-06,
+      "loss": 0.5298,
+      "step": 1133
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.613022993445514e-06,
+      "loss": 0.488,
+      "step": 1134
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.6121832021944e-06,
+      "loss": 0.5319,
+      "step": 1135
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.61134253746482e-06,
+      "loss": 0.5468,
+      "step": 1136
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.610500999415978e-06,
+      "loss": 0.4562,
+      "step": 1137
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.609658588207248e-06,
+      "loss": 0.5169,
+      "step": 1138
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.608815303998176e-06,
+      "loss": 0.5371,
+      "step": 1139
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.60797114694846e-06,
+      "loss": 0.5034,
+      "step": 1140
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.607126117217977e-06,
+      "loss": 0.4987,
+      "step": 1141
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.606280214966759e-06,
+      "loss": 0.4988,
+      "step": 1142
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.605433440355009e-06,
+      "loss": 0.4832,
+      "step": 1143
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.604585793543093e-06,
+      "loss": 0.5243,
+      "step": 1144
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.603737274691543e-06,
+      "loss": 0.4622,
+      "step": 1145
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.602887883961056e-06,
+      "loss": 0.5017,
+      "step": 1146
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.602037621512492e-06,
+      "loss": 0.5069,
+      "step": 1147
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.60118648750688e-06,
+      "loss": 0.4967,
+      "step": 1148
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.60033448210541e-06,
+      "loss": 0.4853,
+      "step": 1149
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.599481605469444e-06,
+      "loss": 0.5233,
+      "step": 1150
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.598627857760496e-06,
+      "loss": 0.4919,
+      "step": 1151
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.59777323914026e-06,
+      "loss": 0.4924,
+      "step": 1152
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.596917749770584e-06,
+      "loss": 0.5148,
+      "step": 1153
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.596061389813486e-06,
+      "loss": 0.5437,
+      "step": 1154
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.59520415943115e-06,
+      "loss": 0.4976,
+      "step": 1155
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.594346058785921e-06,
+      "loss": 0.4901,
+      "step": 1156
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.59348708804031e-06,
+      "loss": 0.5108,
+      "step": 1157
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.592627247356991e-06,
+      "loss": 0.5171,
+      "step": 1158
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.59176653689881e-06,
+      "loss": 0.4943,
+      "step": 1159
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.590904956828769e-06,
+      "loss": 0.489,
+      "step": 1160
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.590042507310038e-06,
+      "loss": 0.4914,
+      "step": 1161
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.589179188505955e-06,
+      "loss": 0.4973,
+      "step": 1162
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.588315000580018e-06,
+      "loss": 0.4916,
+      "step": 1163
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.58744994369589e-06,
+      "loss": 0.5204,
+      "step": 1164
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.586584018017402e-06,
+      "loss": 0.5122,
+      "step": 1165
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.585717223708545e-06,
+      "loss": 0.4958,
+      "step": 1166
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.58484956093348e-06,
+      "loss": 0.498,
+      "step": 1167
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.583981029856529e-06,
+      "loss": 0.4793,
+      "step": 1168
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.583111630642177e-06,
+      "loss": 0.4715,
+      "step": 1169
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.582241363455073e-06,
+      "loss": 0.4886,
+      "step": 1170
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.581370228460038e-06,
+      "loss": 0.4866,
+      "step": 1171
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.580498225822049e-06,
+      "loss": 0.511,
+      "step": 1172
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.579625355706249e-06,
+      "loss": 0.5317,
+      "step": 1173
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.578751618277949e-06,
+      "loss": 0.4588,
+      "step": 1174
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.577877013702619e-06,
+      "loss": 0.4781,
+      "step": 1175
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.5770015421459e-06,
+      "loss": 0.4762,
+      "step": 1176
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.576125203773589e-06,
+      "loss": 0.4796,
+      "step": 1177
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.575247998751655e-06,
+      "loss": 0.5369,
+      "step": 1178
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.574369927246225e-06,
+      "loss": 0.5061,
+      "step": 1179
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.573490989423595e-06,
+      "loss": 0.5076,
+      "step": 1180
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.57261118545022e-06,
+      "loss": 0.5152,
+      "step": 1181
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.571730515492722e-06,
+      "loss": 0.5239,
+      "step": 1182
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.57084897971789e-06,
+      "loss": 0.5164,
+      "step": 1183
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.56996657829267e-06,
+      "loss": 0.4563,
+      "step": 1184
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.569083311384177e-06,
+      "loss": 0.5066,
+      "step": 1185
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.568199179159688e-06,
+      "loss": 0.511,
+      "step": 1186
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.567314181786644e-06,
+      "loss": 0.4948,
+      "step": 1187
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.566428319432652e-06,
+      "loss": 0.5169,
+      "step": 1188
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.56554159226548e-06,
+      "loss": 0.4806,
+      "step": 1189
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.56465400045306e-06,
+      "loss": 0.5103,
+      "step": 1190
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.563765544163492e-06,
+      "loss": 0.534,
+      "step": 1191
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.562876223565031e-06,
+      "loss": 0.4979,
+      "step": 1192
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.561986038826106e-06,
+      "loss": 0.5073,
+      "step": 1193
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.561094990115302e-06,
+      "loss": 0.481,
+      "step": 1194
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.560203077601371e-06,
+      "loss": 0.535,
+      "step": 1195
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.559310301453228e-06,
+      "loss": 0.4692,
+      "step": 1196
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.558416661839951e-06,
+      "loss": 0.4941,
+      "step": 1197
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.557522158930784e-06,
+      "loss": 0.5147,
+      "step": 1198
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.556626792895127e-06,
+      "loss": 0.474,
+      "step": 1199
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.555730563902555e-06,
+      "loss": 0.5089,
+      "step": 1200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.554833472122798e-06,
+      "loss": 0.5232,
+      "step": 1201
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.553935517725749e-06,
+      "loss": 0.4955,
+      "step": 1202
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.553036700881473e-06,
+      "loss": 0.502,
+      "step": 1203
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.552137021760189e-06,
+      "loss": 0.4958,
+      "step": 1204
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.551236480532282e-06,
+      "loss": 0.5153,
+      "step": 1205
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.550335077368303e-06,
+      "loss": 0.5213,
+      "step": 1206
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.549432812438961e-06,
+      "loss": 0.5028,
+      "step": 1207
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.548529685915135e-06,
+      "loss": 0.4799,
+      "step": 1208
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.547625697967862e-06,
+      "loss": 0.4918,
+      "step": 1209
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.546720848768346e-06,
+      "loss": 0.5209,
+      "step": 1210
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.54581513848795e-06,
+      "loss": 0.5071,
+      "step": 1211
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.544908567298201e-06,
+      "loss": 0.4958,
+      "step": 1212
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.544001135370791e-06,
+      "loss": 0.4639,
+      "step": 1213
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.543092842877575e-06,
+      "loss": 0.5067,
+      "step": 1214
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.542183689990567e-06,
+      "loss": 0.4915,
+      "step": 1215
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.541273676881952e-06,
+      "loss": 0.4789,
+      "step": 1216
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.540362803724068e-06,
+      "loss": 0.487,
+      "step": 1217
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.539451070689424e-06,
+      "loss": 0.4892,
+      "step": 1218
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.538538477950686e-06,
+      "loss": 0.4747,
+      "step": 1219
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.537625025680687e-06,
+      "loss": 0.5321,
+      "step": 1220
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.536710714052422e-06,
+      "loss": 0.5278,
+      "step": 1221
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.535795543239045e-06,
+      "loss": 0.5031,
+      "step": 1222
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.53487951341388e-06,
+      "loss": 0.49,
+      "step": 1223
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.533962624750406e-06,
+      "loss": 0.4541,
+      "step": 1224
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.533044877422267e-06,
+      "loss": 0.4585,
+      "step": 1225
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.532126271603275e-06,
+      "loss": 0.522,
+      "step": 1226
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.531206807467395e-06,
+      "loss": 0.5049,
+      "step": 1227
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.530286485188764e-06,
+      "loss": 0.5061,
+      "step": 1228
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.529365304941676e-06,
+      "loss": 0.4902,
+      "step": 1229
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.528443266900586e-06,
+      "loss": 0.5154,
+      "step": 1230
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.52752037124012e-06,
+      "loss": 0.522,
+      "step": 1231
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.526596618135055e-06,
+      "loss": 0.4991,
+      "step": 1232
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.525672007760337e-06,
+      "loss": 0.4821,
+      "step": 1233
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.524746540291076e-06,
+      "loss": 0.4969,
+      "step": 1234
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.523820215902542e-06,
+      "loss": 0.5205,
+      "step": 1235
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.522893034770165e-06,
+      "loss": 0.5031,
+      "step": 1236
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.521964997069538e-06,
+      "loss": 0.4851,
+      "step": 1237
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.521036102976417e-06,
+      "loss": 0.4873,
+      "step": 1238
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.520106352666727e-06,
+      "loss": 0.493,
+      "step": 1239
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.51917574631654e-06,
+      "loss": 0.4823,
+      "step": 1240
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.518244284102108e-06,
+      "loss": 0.5096,
+      "step": 1241
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.517311966199827e-06,
+      "loss": 0.4932,
+      "step": 1242
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.51637879278627e-06,
+      "loss": 0.4899,
+      "step": 1243
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.515444764038165e-06,
+      "loss": 0.4627,
+      "step": 1244
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.514509880132401e-06,
+      "loss": 0.5111,
+      "step": 1245
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.513574141246034e-06,
+      "loss": 0.4717,
+      "step": 1246
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.512637547556275e-06,
+      "loss": 0.4671,
+      "step": 1247
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.511700099240507e-06,
+      "loss": 0.496,
+      "step": 1248
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.510761796476263e-06,
+      "loss": 0.5362,
+      "step": 1249
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.509822639441247e-06,
+      "loss": 0.4927,
+      "step": 1250
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.508882628313316e-06,
+      "loss": 0.4922,
+      "step": 1251
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.507941763270501e-06,
+      "loss": 0.5035,
+      "step": 1252
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.507000044490984e-06,
+      "loss": 0.5028,
+      "step": 1253
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.506057472153112e-06,
+      "loss": 0.5011,
+      "step": 1254
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.505114046435395e-06,
+      "loss": 0.5127,
+      "step": 1255
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.504169767516504e-06,
+      "loss": 0.4774,
+      "step": 1256
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.503224635575273e-06,
+      "loss": 0.479,
+      "step": 1257
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.502278650790692e-06,
+      "loss": 0.5037,
+      "step": 1258
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.501331813341918e-06,
+      "loss": 0.474,
+      "step": 1259
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.50038412340827e-06,
+      "loss": 0.4732,
+      "step": 1260
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.499435581169222e-06,
+      "loss": 0.4879,
+      "step": 1261
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.498486186804418e-06,
+      "loss": 0.5179,
+      "step": 1262
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.497535940493656e-06,
+      "loss": 0.4802,
+      "step": 1263
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.4965848424169e-06,
+      "loss": 0.5395,
+      "step": 1264
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.495632892754275e-06,
+      "loss": 0.5301,
+      "step": 1265
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.494680091686061e-06,
+      "loss": 0.5404,
+      "step": 1266
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.493726439392712e-06,
+      "loss": 0.4749,
+      "step": 1267
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.492771936054828e-06,
+      "loss": 0.5136,
+      "step": 1268
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.491816581853181e-06,
+      "loss": 0.5102,
+      "step": 1269
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.4908603769687e-06,
+      "loss": 0.5307,
+      "step": 1270
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.48990332158248e-06,
+      "loss": 0.4912,
+      "step": 1271
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.488945415875765e-06,
+      "loss": 0.5173,
+      "step": 1272
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.487986660029975e-06,
+      "loss": 0.5184,
+      "step": 1273
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.487027054226681e-06,
+      "loss": 0.504,
+      "step": 1274
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.486066598647618e-06,
+      "loss": 0.5041,
+      "step": 1275
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.485105293474682e-06,
+      "loss": 0.5334,
+      "step": 1276
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.484143138889931e-06,
+      "loss": 0.4866,
+      "step": 1277
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.48318013507558e-06,
+      "loss": 0.491,
+      "step": 1278
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.482216282214011e-06,
+      "loss": 0.48,
+      "step": 1279
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.481251580487762e-06,
+      "loss": 0.5151,
+      "step": 1280
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.480286030079532e-06,
+      "loss": 0.4927,
+      "step": 1281
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.479319631172184e-06,
+      "loss": 0.548,
+      "step": 1282
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.478352383948735e-06,
+      "loss": 0.5228,
+      "step": 1283
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.477384288592371e-06,
+      "loss": 0.5309,
+      "step": 1284
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.476415345286436e-06,
+      "loss": 0.5014,
+      "step": 1285
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.475445554214429e-06,
+      "loss": 0.5211,
+      "step": 1286
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.474474915560016e-06,
+      "loss": 0.5225,
+      "step": 1287
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.473503429507024e-06,
+      "loss": 0.5076,
+      "step": 1288
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.472531096239434e-06,
+      "loss": 0.4998,
+      "step": 1289
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.471557915941393e-06,
+      "loss": 0.4958,
+      "step": 1290
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.470583888797207e-06,
+      "loss": 0.471,
+      "step": 1291
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.469609014991341e-06,
+      "loss": 0.5208,
+      "step": 1292
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.468633294708422e-06,
+      "loss": 0.4993,
+      "step": 1293
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.467656728133239e-06,
+      "loss": 0.4867,
+      "step": 1294
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.466679315450736e-06,
+      "loss": 0.4856,
+      "step": 1295
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.465701056846022e-06,
+      "loss": 0.4955,
+      "step": 1296
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.464721952504363e-06,
+      "loss": 0.5044,
+      "step": 1297
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.46374200261119e-06,
+      "loss": 0.5083,
+      "step": 1298
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.462761207352087e-06,
+      "loss": 0.5111,
+      "step": 1299
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.461779566912805e-06,
+      "loss": 0.5077,
+      "step": 1300
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.46079708147925e-06,
+      "loss": 0.4749,
+      "step": 1301
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.45981375123749e-06,
+      "loss": 0.4989,
+      "step": 1302
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.458829576373753e-06,
+      "loss": 0.5039,
+      "step": 1303
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.45784455707443e-06,
+      "loss": 0.5157,
+      "step": 1304
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.456858693526065e-06,
+      "loss": 0.5234,
+      "step": 1305
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.455871985915367e-06,
+      "loss": 0.5107,
+      "step": 1306
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.454884434429205e-06,
+      "loss": 0.5032,
+      "step": 1307
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.453896039254607e-06,
+      "loss": 0.5286,
+      "step": 1308
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.452906800578755e-06,
+      "loss": 0.4829,
+      "step": 1309
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.451916718589002e-06,
+      "loss": 0.4971,
+      "step": 1310
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.450925793472853e-06,
+      "loss": 0.5361,
+      "step": 1311
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.449934025417974e-06,
+      "loss": 0.4569,
+      "step": 1312
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.448941414612191e-06,
+      "loss": 0.5177,
+      "step": 1313
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.447947961243489e-06,
+      "loss": 0.5232,
+      "step": 1314
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.446953665500013e-06,
+      "loss": 0.4788,
+      "step": 1315
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.445958527570069e-06,
+      "loss": 0.4896,
+      "step": 1316
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.44496254764212e-06,
+      "loss": 0.4922,
+      "step": 1317
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.443965725904792e-06,
+      "loss": 0.5178,
+      "step": 1318
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.442968062546864e-06,
+      "loss": 0.4525,
+      "step": 1319
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.441969557757285e-06,
+      "loss": 0.5226,
+      "step": 1320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.44097021172515e-06,
+      "loss": 0.4682,
+      "step": 1321
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.439970024639723e-06,
+      "loss": 0.5075,
+      "step": 1322
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.438968996690425e-06,
+      "loss": 0.5093,
+      "step": 1323
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.437967128066836e-06,
+      "loss": 0.4893,
+      "step": 1324
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.436964418958692e-06,
+      "loss": 0.5083,
+      "step": 1325
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.435960869555895e-06,
+      "loss": 0.5051,
+      "step": 1326
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.434956480048501e-06,
+      "loss": 0.4693,
+      "step": 1327
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.433951250626725e-06,
+      "loss": 0.4776,
+      "step": 1328
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.432945181480943e-06,
+      "loss": 0.5016,
+      "step": 1329
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.43193827280169e-06,
+      "loss": 0.5021,
+      "step": 1330
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.43093052477966e-06,
+      "loss": 0.4733,
+      "step": 1331
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.429921937605704e-06,
+      "loss": 0.4735,
+      "step": 1332
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.428912511470833e-06,
+      "loss": 0.5372,
+      "step": 1333
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.427902246566218e-06,
+      "loss": 0.4921,
+      "step": 1334
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.426891143083189e-06,
+      "loss": 0.4889,
+      "step": 1335
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.425879201213232e-06,
+      "loss": 0.5025,
+      "step": 1336
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.424866421147996e-06,
+      "loss": 0.5019,
+      "step": 1337
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.423852803079283e-06,
+      "loss": 0.4878,
+      "step": 1338
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.42283834719906e-06,
+      "loss": 0.5191,
+      "step": 1339
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.421823053699449e-06,
+      "loss": 0.5308,
+      "step": 1340
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.420806922772732e-06,
+      "loss": 0.5507,
+      "step": 1341
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.419789954611347e-06,
+      "loss": 0.5287,
+      "step": 1342
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.418772149407895e-06,
+      "loss": 0.4742,
+      "step": 1343
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.41775350735513e-06,
+      "loss": 0.4853,
+      "step": 1344
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.41673402864597e-06,
+      "loss": 0.474,
+      "step": 1345
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.41571371347349e-06,
+      "loss": 0.5053,
+      "step": 1346
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.414692562030919e-06,
+      "loss": 0.5096,
+      "step": 1347
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.41367057451165e-06,
+      "loss": 0.5015,
+      "step": 1348
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.412647751109231e-06,
+      "loss": 0.503,
+      "step": 1349
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.411624092017372e-06,
+      "loss": 0.5237,
+      "step": 1350
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.410599597429937e-06,
+      "loss": 0.5209,
+      "step": 1351
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.409574267540948e-06,
+      "loss": 0.5197,
+      "step": 1352
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.40854810254459e-06,
+      "loss": 0.4986,
+      "step": 1353
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.407521102635203e-06,
+      "loss": 0.4791,
+      "step": 1354
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.406493268007284e-06,
+      "loss": 0.4961,
+      "step": 1355
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.40546459885549e-06,
+      "loss": 0.5149,
+      "step": 1356
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.404435095374636e-06,
+      "loss": 0.4844,
+      "step": 1357
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.403404757759696e-06,
+      "loss": 0.4931,
+      "step": 1358
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.402373586205795e-06,
+      "loss": 0.5225,
+      "step": 1359
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.401341580908227e-06,
+      "loss": 0.5404,
+      "step": 1360
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.400308742062438e-06,
+      "loss": 0.5105,
+      "step": 1361
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.39927506986403e-06,
+      "loss": 0.4858,
+      "step": 1362
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.398240564508766e-06,
+      "loss": 0.53,
+      "step": 1363
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.397205226192565e-06,
+      "loss": 0.5078,
+      "step": 1364
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.396169055111506e-06,
+      "loss": 0.5368,
+      "step": 1365
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.395132051461824e-06,
+      "loss": 0.5219,
+      "step": 1366
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.39409421543991e-06,
+      "loss": 0.5235,
+      "step": 1367
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.393055547242317e-06,
+      "loss": 0.5133,
+      "step": 1368
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.392016047065753e-06,
+      "loss": 0.4744,
+      "step": 1369
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.390975715107083e-06,
+      "loss": 0.4867,
+      "step": 1370
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.38993455156333e-06,
+      "loss": 0.5134,
+      "step": 1371
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.388892556631675e-06,
+      "loss": 0.4798,
+      "step": 1372
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.387849730509458e-06,
+      "loss": 0.5035,
+      "step": 1373
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.386806073394171e-06,
+      "loss": 0.4922,
+      "step": 1374
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.38576158548347e-06,
+      "loss": 0.5045,
+      "step": 1375
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.384716266975165e-06,
+      "loss": 0.5366,
+      "step": 1376
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.383670118067224e-06,
+      "loss": 0.473,
+      "step": 1377
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.382623138957771e-06,
+      "loss": 0.4988,
+      "step": 1378
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.381575329845087e-06,
+      "loss": 0.459,
+      "step": 1379
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.380526690927613e-06,
+      "loss": 0.5133,
+      "step": 1380
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.379477222403946e-06,
+      "loss": 0.5259,
+      "step": 1381
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.378426924472839e-06,
+      "loss": 0.4836,
+      "step": 1382
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.377375797333202e-06,
+      "loss": 0.5157,
+      "step": 1383
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.376323841184103e-06,
+      "loss": 0.486,
+      "step": 1384
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.375271056224769e-06,
+      "loss": 0.5066,
+      "step": 1385
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.374217442654579e-06,
+      "loss": 0.4913,
+      "step": 1386
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.373163000673072e-06,
+      "loss": 0.5136,
+      "step": 1387
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.372107730479944e-06,
+      "loss": 0.4952,
+      "step": 1388
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.371051632275048e-06,
+      "loss": 0.5012,
+      "step": 1389
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.369994706258393e-06,
+      "loss": 0.5226,
+      "step": 1390
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.36893695263014e-06,
+      "loss": 0.5046,
+      "step": 1391
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.367878371590622e-06,
+      "loss": 0.5218,
+      "step": 1392
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.36681896334031e-06,
+      "loss": 0.4892,
+      "step": 1393
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.365758728079844e-06,
+      "loss": 0.5025,
+      "step": 1394
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.364697666010013e-06,
+      "loss": 0.4728,
+      "step": 1395
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.36363577733177e-06,
+      "loss": 0.5159,
+      "step": 1396
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.36257306224622e-06,
+      "loss": 0.4756,
+      "step": 1397
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.361509520954625e-06,
+      "loss": 0.527,
+      "step": 1398
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.360445153658405e-06,
+      "loss": 0.5373,
+      "step": 1399
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.35937996055913e-06,
+      "loss": 0.5354,
+      "step": 1400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.358313941858537e-06,
+      "loss": 0.5164,
+      "step": 1401
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.357247097758515e-06,
+      "loss": 0.5287,
+      "step": 1402
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.356179428461103e-06,
+      "loss": 0.4991,
+      "step": 1403
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.355110934168506e-06,
+      "loss": 0.5296,
+      "step": 1404
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.354041615083077e-06,
+      "loss": 0.5079,
+      "step": 1405
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.352971471407333e-06,
+      "loss": 0.5157,
+      "step": 1406
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.351900503343941e-06,
+      "loss": 0.516,
+      "step": 1407
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.350828711095728e-06,
+      "loss": 0.4996,
+      "step": 1408
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.349756094865672e-06,
+      "loss": 0.5598,
+      "step": 1409
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.348682654856913e-06,
+      "loss": 0.472,
+      "step": 1410
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.347608391272745e-06,
+      "loss": 0.498,
+      "step": 1411
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.346533304316617e-06,
+      "loss": 0.5285,
+      "step": 1412
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.345457394192132e-06,
+      "loss": 0.515,
+      "step": 1413
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.344380661103056e-06,
+      "loss": 0.5196,
+      "step": 1414
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.3433031052533e-06,
+      "loss": 0.4919,
+      "step": 1415
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.342224726846943e-06,
+      "loss": 0.5178,
+      "step": 1416
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.34114552608821e-06,
+      "loss": 0.4903,
+      "step": 1417
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.340065503181487e-06,
+      "loss": 0.5261,
+      "step": 1418
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.338984658331314e-06,
+      "loss": 0.5301,
+      "step": 1419
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.337902991742385e-06,
+      "loss": 0.5102,
+      "step": 1420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.336820503619554e-06,
+      "loss": 0.4877,
+      "step": 1421
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.335737194167826e-06,
+      "loss": 0.4939,
+      "step": 1422
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.334653063592366e-06,
+      "loss": 0.491,
+      "step": 1423
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.33356811209849e-06,
+      "loss": 0.5088,
+      "step": 1424
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.332482339891674e-06,
+      "loss": 0.5166,
+      "step": 1425
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.331395747177542e-06,
+      "loss": 0.5036,
+      "step": 1426
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.330308334161883e-06,
+      "loss": 0.4912,
+      "step": 1427
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.329220101050636e-06,
+      "loss": 0.5181,
+      "step": 1428
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.328131048049895e-06,
+      "loss": 0.4983,
+      "step": 1429
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.327041175365911e-06,
+      "loss": 0.5394,
+      "step": 1430
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.32595048320509e-06,
+      "loss": 0.4872,
+      "step": 1431
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.324858971773991e-06,
+      "loss": 0.4952,
+      "step": 1432
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.323766641279333e-06,
+      "loss": 0.4749,
+      "step": 1433
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.322673491927983e-06,
+      "loss": 0.5093,
+      "step": 1434
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.321579523926971e-06,
+      "loss": 0.4944,
+      "step": 1435
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.320484737483475e-06,
+      "loss": 0.506,
+      "step": 1436
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.319389132804835e-06,
+      "loss": 0.4863,
+      "step": 1437
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.318292710098538e-06,
+      "loss": 0.5148,
+      "step": 1438
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.31719546957223e-06,
+      "loss": 0.4637,
+      "step": 1439
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.316097411433717e-06,
+      "loss": 0.5011,
+      "step": 1440
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.31499853589095e-06,
+      "loss": 0.4876,
+      "step": 1441
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.31389884315204e-06,
+      "loss": 0.4985,
+      "step": 1442
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.312798333425256e-06,
+      "loss": 0.5158,
+      "step": 1443
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.311697006919013e-06,
+      "loss": 0.5149,
+      "step": 1444
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.310594863841891e-06,
+      "loss": 0.4611,
+      "step": 1445
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.309491904402614e-06,
+      "loss": 0.482,
+      "step": 1446
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.308388128810072e-06,
+      "loss": 0.5145,
+      "step": 1447
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.307283537273297e-06,
+      "loss": 0.4787,
+      "step": 1448
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.306178130001486e-06,
+      "loss": 0.497,
+      "step": 1449
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.305071907203987e-06,
+      "loss": 0.4879,
+      "step": 1450
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.303964869090301e-06,
+      "loss": 0.4656,
+      "step": 1451
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.302857015870083e-06,
+      "loss": 0.5128,
+      "step": 1452
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.301748347753147e-06,
+      "loss": 0.489,
+      "step": 1453
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.300638864949455e-06,
+      "loss": 0.4901,
+      "step": 1454
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.29952856766913e-06,
+      "loss": 0.5265,
+      "step": 1455
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.298417456122442e-06,
+      "loss": 0.481,
+      "step": 1456
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.297305530519818e-06,
+      "loss": 0.5169,
+      "step": 1457
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.296192791071846e-06,
+      "loss": 0.5074,
+      "step": 1458
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.295079237989258e-06,
+      "loss": 0.4891,
+      "step": 1459
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.293964871482944e-06,
+      "loss": 0.5274,
+      "step": 1460
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.292849691763952e-06,
+      "loss": 0.4639,
+      "step": 1461
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.291733699043476e-06,
+      "loss": 0.4921,
+      "step": 1462
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.290616893532868e-06,
+      "loss": 0.4692,
+      "step": 1463
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.28949927544364e-06,
+      "loss": 0.514,
+      "step": 1464
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.288380844987445e-06,
+      "loss": 0.4968,
+      "step": 1465
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.287261602376102e-06,
+      "loss": 0.5085,
+      "step": 1466
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.286141547821577e-06,
+      "loss": 0.5311,
+      "step": 1467
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.28502068153599e-06,
+      "loss": 0.4692,
+      "step": 1468
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.28389900373162e-06,
+      "loss": 0.4628,
+      "step": 1469
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.282776514620894e-06,
+      "loss": 0.4677,
+      "step": 1470
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.281653214416395e-06,
+      "loss": 0.4882,
+      "step": 1471
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.280529103330857e-06,
+      "loss": 0.5136,
+      "step": 1472
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.279404181577174e-06,
+      "loss": 0.4966,
+      "step": 1473
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.278278449368387e-06,
+      "loss": 0.4602,
+      "step": 1474
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.277151906917691e-06,
+      "loss": 0.4922,
+      "step": 1475
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.27602455443844e-06,
+      "loss": 0.5106,
+      "step": 1476
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.274896392144134e-06,
+      "loss": 0.4822,
+      "step": 1477
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.273767420248435e-06,
+      "loss": 0.4944,
+      "step": 1478
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.272637638965148e-06,
+      "loss": 0.504,
+      "step": 1479
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.271507048508239e-06,
+      "loss": 0.4977,
+      "step": 1480
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.270375649091825e-06,
+      "loss": 0.51,
+      "step": 1481
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.269243440930177e-06,
+      "loss": 0.4882,
+      "step": 1482
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.268110424237716e-06,
+      "loss": 0.5383,
+      "step": 1483
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.26697659922902e-06,
+      "loss": 0.5345,
+      "step": 1484
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.265841966118818e-06,
+      "loss": 0.5261,
+      "step": 1485
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.264706525121993e-06,
+      "loss": 0.5279,
+      "step": 1486
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.263570276453582e-06,
+      "loss": 0.5218,
+      "step": 1487
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.262433220328768e-06,
+      "loss": 0.5107,
+      "step": 1488
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.2612953569629e-06,
+      "loss": 0.4963,
+      "step": 1489
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.260156686571466e-06,
+      "loss": 0.4876,
+      "step": 1490
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.259017209370116e-06,
+      "loss": 0.5137,
+      "step": 1491
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.257876925574649e-06,
+      "loss": 0.5061,
+      "step": 1492
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.256735835401021e-06,
+      "loss": 0.4645,
+      "step": 1493
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.255593939065331e-06,
+      "loss": 0.506,
+      "step": 1494
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.254451236783844e-06,
+      "loss": 0.5633,
+      "step": 1495
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.253307728772964e-06,
+      "loss": 0.4635,
+      "step": 1496
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.25216341524926e-06,
+      "loss": 0.5111,
+      "step": 1497
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.251018296429443e-06,
+      "loss": 0.5159,
+      "step": 1498
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.249872372530387e-06,
+      "loss": 0.4793,
+      "step": 1499
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.248725643769108e-06,
+      "loss": 0.4935,
+      "step": 1500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.24757811036278e-06,
+      "loss": 0.4805,
+      "step": 1501
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.246429772528732e-06,
+      "loss": 0.5018,
+      "step": 1502
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.245280630484439e-06,
+      "loss": 0.5255,
+      "step": 1503
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.244130684447532e-06,
+      "loss": 0.4977,
+      "step": 1504
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.242979934635792e-06,
+      "loss": 0.5235,
+      "step": 1505
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.241828381267156e-06,
+      "loss": 0.4574,
+      "step": 1506
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.240676024559712e-06,
+      "loss": 0.4518,
+      "step": 1507
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.239522864731696e-06,
+      "loss": 0.505,
+      "step": 1508
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.2383689020015e-06,
+      "loss": 0.5248,
+      "step": 1509
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.23721413658767e-06,
+      "loss": 0.521,
+      "step": 1510
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.236058568708896e-06,
+      "loss": 0.5078,
+      "step": 1511
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.234902198584031e-06,
+      "loss": 0.522,
+      "step": 1512
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.23374502643207e-06,
+      "loss": 0.4914,
+      "step": 1513
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.232587052472166e-06,
+      "loss": 0.467,
+      "step": 1514
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.231428276923622e-06,
+      "loss": 0.495,
+      "step": 1515
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.230268700005893e-06,
+      "loss": 0.4858,
+      "step": 1516
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.229108321938585e-06,
+      "loss": 0.5448,
+      "step": 1517
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.227947142941454e-06,
+      "loss": 0.4823,
+      "step": 1518
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.226785163234413e-06,
+      "loss": 0.5011,
+      "step": 1519
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.225622383037523e-06,
+      "loss": 0.5264,
+      "step": 1520
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.224458802570998e-06,
+      "loss": 0.5074,
+      "step": 1521
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.2232944220552e-06,
+      "loss": 0.5203,
+      "step": 1522
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.222129241710647e-06,
+      "loss": 0.4701,
+      "step": 1523
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.220963261758006e-06,
+      "loss": 0.527,
+      "step": 1524
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.219796482418098e-06,
+      "loss": 0.5117,
+      "step": 1525
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.218628903911892e-06,
+      "loss": 0.5457,
+      "step": 1526
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.21746052646051e-06,
+      "loss": 0.4999,
+      "step": 1527
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.216291350285224e-06,
+      "loss": 0.5099,
+      "step": 1528
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.21512137560746e-06,
+      "loss": 0.5067,
+      "step": 1529
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.213950602648795e-06,
+      "loss": 0.478,
+      "step": 1530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.212779031630953e-06,
+      "loss": 0.4721,
+      "step": 1531
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.211606662775813e-06,
+      "loss": 0.4935,
+      "step": 1532
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.210433496305406e-06,
+      "loss": 0.4756,
+      "step": 1533
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.20925953244191e-06,
+      "loss": 0.5099,
+      "step": 1534
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.208084771407656e-06,
+      "loss": 0.4829,
+      "step": 1535
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.206909213425128e-06,
+      "loss": 0.4859,
+      "step": 1536
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.205732858716957e-06,
+      "loss": 0.4932,
+      "step": 1537
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.20455570750593e-06,
+      "loss": 0.5187,
+      "step": 1538
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.203377760014977e-06,
+      "loss": 0.5117,
+      "step": 1539
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.202199016467188e-06,
+      "loss": 0.5204,
+      "step": 1540
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.201019477085798e-06,
+      "loss": 0.5039,
+      "step": 1541
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.199839142094193e-06,
+      "loss": 0.5127,
+      "step": 1542
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.198658011715913e-06,
+      "loss": 0.4978,
+      "step": 1543
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.197476086174644e-06,
+      "loss": 0.4997,
+      "step": 1544
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.196293365694226e-06,
+      "loss": 0.4802,
+      "step": 1545
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.19510985049865e-06,
+      "loss": 0.5042,
+      "step": 1546
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.193925540812055e-06,
+      "loss": 0.5218,
+      "step": 1547
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.192740436858729e-06,
+      "loss": 0.4848,
+      "step": 1548
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.191554538863116e-06,
+      "loss": 0.4776,
+      "step": 1549
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.190367847049809e-06,
+      "loss": 0.5223,
+      "step": 1550
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.189180361643546e-06,
+      "loss": 0.5007,
+      "step": 1551
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.18799208286922e-06,
+      "loss": 0.4855,
+      "step": 1552
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.186803010951873e-06,
+      "loss": 0.5193,
+      "step": 1553
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.185613146116698e-06,
+      "loss": 0.4692,
+      "step": 1554
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.184422488589038e-06,
+      "loss": 0.443,
+      "step": 1555
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.183231038594384e-06,
+      "loss": 0.4968,
+      "step": 1556
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.182038796358381e-06,
+      "loss": 0.4751,
+      "step": 1557
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.180845762106822e-06,
+      "loss": 0.5048,
+      "step": 1558
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.179651936065646e-06,
+      "loss": 0.4811,
+      "step": 1559
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.17845731846095e-06,
+      "loss": 0.4722,
+      "step": 1560
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.177261909518975e-06,
+      "loss": 0.4921,
+      "step": 1561
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.176065709466114e-06,
+      "loss": 0.4894,
+      "step": 1562
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.174868718528908e-06,
+      "loss": 0.4755,
+      "step": 1563
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.17367093693405e-06,
+      "loss": 0.5286,
+      "step": 1564
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.172472364908384e-06,
+      "loss": 0.4909,
+      "step": 1565
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.171273002678898e-06,
+      "loss": 0.4756,
+      "step": 1566
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.170072850472737e-06,
+      "loss": 0.5229,
+      "step": 1567
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.168871908517189e-06,
+      "loss": 0.4899,
+      "step": 1568
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.167670177039693e-06,
+      "loss": 0.5216,
+      "step": 1569
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.166467656267845e-06,
+      "loss": 0.5073,
+      "step": 1570
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.165264346429379e-06,
+      "loss": 0.5027,
+      "step": 1571
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.164060247752185e-06,
+      "loss": 0.4966,
+      "step": 1572
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.162855360464302e-06,
+      "loss": 0.4655,
+      "step": 1573
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.161649684793918e-06,
+      "loss": 0.5001,
+      "step": 1574
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.160443220969368e-06,
+      "loss": 0.4736,
+      "step": 1575
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.159235969219142e-06,
+      "loss": 0.448,
+      "step": 1576
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.158027929771871e-06,
+      "loss": 0.4776,
+      "step": 1577
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.156819102856343e-06,
+      "loss": 0.4872,
+      "step": 1578
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.155609488701489e-06,
+      "loss": 0.5203,
+      "step": 1579
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.154399087536394e-06,
+      "loss": 0.4931,
+      "step": 1580
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.15318789959029e-06,
+      "loss": 0.5072,
+      "step": 1581
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.151975925092556e-06,
+      "loss": 0.461,
+      "step": 1582
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.150763164272722e-06,
+      "loss": 0.5367,
+      "step": 1583
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.149549617360467e-06,
+      "loss": 0.4938,
+      "step": 1584
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.14833528458562e-06,
+      "loss": 0.5434,
+      "step": 1585
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.147120166178156e-06,
+      "loss": 0.5008,
+      "step": 1586
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.145904262368202e-06,
+      "loss": 0.5027,
+      "step": 1587
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.144687573386028e-06,
+      "loss": 0.5047,
+      "step": 1588
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.143470099462062e-06,
+      "loss": 0.4985,
+      "step": 1589
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.142251840826872e-06,
+      "loss": 0.5122,
+      "step": 1590
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.14103279771118e-06,
+      "loss": 0.5031,
+      "step": 1591
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.139812970345852e-06,
+      "loss": 0.5125,
+      "step": 1592
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.138592358961905e-06,
+      "loss": 0.4786,
+      "step": 1593
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.137370963790508e-06,
+      "loss": 0.4803,
+      "step": 1594
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.13614878506297e-06,
+      "loss": 0.4852,
+      "step": 1595
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.134925823010759e-06,
+      "loss": 0.5109,
+      "step": 1596
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.133702077865481e-06,
+      "loss": 0.5196,
+      "step": 1597
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.132477549858898e-06,
+      "loss": 0.4963,
+      "step": 1598
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.131252239222916e-06,
+      "loss": 0.5095,
+      "step": 1599
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.13002614618959e-06,
+      "loss": 0.4925,
+      "step": 1600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.128799270991126e-06,
+      "loss": 0.4617,
+      "step": 1601
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.127571613859872e-06,
+      "loss": 0.4913,
+      "step": 1602
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.126343175028332e-06,
+      "loss": 0.4759,
+      "step": 1603
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.12511395472915e-06,
+      "loss": 0.534,
+      "step": 1604
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.123883953195125e-06,
+      "loss": 0.4784,
+      "step": 1605
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.122653170659198e-06,
+      "loss": 0.4856,
+      "step": 1606
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.121421607354464e-06,
+      "loss": 0.5103,
+      "step": 1607
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.12018926351416e-06,
+      "loss": 0.4992,
+      "step": 1608
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.118956139371677e-06,
+      "loss": 0.4922,
+      "step": 1609
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.117722235160547e-06,
+      "loss": 0.5207,
+      "step": 1610
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.116487551114453e-06,
+      "loss": 0.5089,
+      "step": 1611
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.115252087467229e-06,
+      "loss": 0.5355,
+      "step": 1612
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.114015844452848e-06,
+      "loss": 0.4634,
+      "step": 1613
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.11277882230544e-06,
+      "loss": 0.5045,
+      "step": 1614
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.111541021259276e-06,
+      "loss": 0.4909,
+      "step": 1615
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.110302441548779e-06,
+      "loss": 0.4663,
+      "step": 1616
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.109063083408518e-06,
+      "loss": 0.4946,
+      "step": 1617
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.107822947073206e-06,
+      "loss": 0.5325,
+      "step": 1618
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.10658203277771e-06,
+      "loss": 0.4837,
+      "step": 1619
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.105340340757037e-06,
+      "loss": 0.4907,
+      "step": 1620
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.104097871246346e-06,
+      "loss": 0.5082,
+      "step": 1621
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.102854624480944e-06,
+      "loss": 0.5006,
+      "step": 1622
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.10161060069628e-06,
+      "loss": 0.4811,
+      "step": 1623
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.100365800127957e-06,
+      "loss": 0.5095,
+      "step": 1624
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.099120223011719e-06,
+      "loss": 0.511,
+      "step": 1625
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.09787386958346e-06,
+      "loss": 0.5394,
+      "step": 1626
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.096626740079223e-06,
+      "loss": 0.5327,
+      "step": 1627
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.095378834735194e-06,
+      "loss": 0.5091,
+      "step": 1628
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.094130153787705e-06,
+      "loss": 0.5087,
+      "step": 1629
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.092880697473244e-06,
+      "loss": 0.5189,
+      "step": 1630
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.091630466028432e-06,
+      "loss": 0.5153,
+      "step": 1631
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.090379459690048e-06,
+      "loss": 0.5009,
+      "step": 1632
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.089127678695015e-06,
+      "loss": 0.482,
+      "step": 1633
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.087875123280399e-06,
+      "loss": 0.4943,
+      "step": 1634
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.086621793683415e-06,
+      "loss": 0.4844,
+      "step": 1635
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.085367690141426e-06,
+      "loss": 0.4907,
+      "step": 1636
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.084112812891944e-06,
+      "loss": 0.5032,
+      "step": 1637
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.082857162172616e-06,
+      "loss": 0.534,
+      "step": 1638
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.08160073822125e-06,
+      "loss": 0.5107,
+      "step": 1639
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.080343541275793e-06,
+      "loss": 0.4766,
+      "step": 1640
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.079085571574336e-06,
+      "loss": 0.5157,
+      "step": 1641
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.077826829355123e-06,
+      "loss": 0.4962,
+      "step": 1642
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.07656731485654e-06,
+      "loss": 0.5078,
+      "step": 1643
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.075307028317118e-06,
+      "loss": 0.5036,
+      "step": 1644
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.074045969975539e-06,
+      "loss": 0.4834,
+      "step": 1645
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.072784140070628e-06,
+      "loss": 0.5438,
+      "step": 1646
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.071521538841356e-06,
+      "loss": 0.4699,
+      "step": 1647
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.070258166526844e-06,
+      "loss": 0.5241,
+      "step": 1648
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.06899402336635e-06,
+      "loss": 0.4893,
+      "step": 1649
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.067729109599288e-06,
+      "loss": 0.5151,
+      "step": 1650
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.066463425465213e-06,
+      "loss": 0.4851,
+      "step": 1651
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.065196971203827e-06,
+      "loss": 0.4836,
+      "step": 1652
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.063929747054976e-06,
+      "loss": 0.5112,
+      "step": 1653
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.062661753258652e-06,
+      "loss": 0.4765,
+      "step": 1654
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.061392990054998e-06,
+      "loss": 0.5212,
+      "step": 1655
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.060123457684298e-06,
+      "loss": 0.5115,
+      "step": 1656
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.05885315638698e-06,
+      "loss": 0.5195,
+      "step": 1657
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.05758208640362e-06,
+      "loss": 0.5213,
+      "step": 1658
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.056310247974944e-06,
+      "loss": 0.5073,
+      "step": 1659
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.055037641341814e-06,
+      "loss": 0.5045,
+      "step": 1660
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.053764266745247e-06,
+      "loss": 0.4895,
+      "step": 1661
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.052490124426397e-06,
+      "loss": 0.517,
+      "step": 1662
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.05121521462657e-06,
+      "loss": 0.4789,
+      "step": 1663
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.049939537587213e-06,
+      "loss": 0.562,
+      "step": 1664
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.048663093549923e-06,
+      "loss": 0.5177,
+      "step": 1665
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.047385882756436e-06,
+      "loss": 0.5168,
+      "step": 1666
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.04610790544864e-06,
+      "loss": 0.4718,
+      "step": 1667
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.044829161868563e-06,
+      "loss": 0.5166,
+      "step": 1668
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.043549652258378e-06,
+      "loss": 0.5035,
+      "step": 1669
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.042269376860408e-06,
+      "loss": 0.5033,
+      "step": 1670
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.040988335917118e-06,
+      "loss": 0.5325,
+      "step": 1671
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.039706529671117e-06,
+      "loss": 0.5165,
+      "step": 1672
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.03842395836516e-06,
+      "loss": 0.4823,
+      "step": 1673
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.037140622242146e-06,
+      "loss": 0.547,
+      "step": 1674
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.03585652154512e-06,
+      "loss": 0.4985,
+      "step": 1675
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.034571656517272e-06,
+      "loss": 0.4775,
+      "step": 1676
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.033286027401938e-06,
+      "loss": 0.4753,
+      "step": 1677
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.031999634442592e-06,
+      "loss": 0.5382,
+      "step": 1678
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.03071247788286e-06,
+      "loss": 0.4951,
+      "step": 1679
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.029424557966514e-06,
+      "loss": 0.4852,
+      "step": 1680
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.028135874937463e-06,
+      "loss": 0.4967,
+      "step": 1681
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.026846429039763e-06,
+      "loss": 0.516,
+      "step": 1682
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.025556220517618e-06,
+      "loss": 0.4923,
+      "step": 1683
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.024265249615373e-06,
+      "loss": 0.5352,
+      "step": 1684
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.02297351657752e-06,
+      "loss": 0.5086,
+      "step": 1685
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.021681021648693e-06,
+      "loss": 0.4979,
+      "step": 1686
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.020387765073672e-06,
+      "loss": 0.4776,
+      "step": 1687
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.019093747097378e-06,
+      "loss": 0.5183,
+      "step": 1688
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.017798967964883e-06,
+      "loss": 0.526,
+      "step": 1689
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.016503427921395e-06,
+      "loss": 0.4755,
+      "step": 1690
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.01520712721227e-06,
+      "loss": 0.4833,
+      "step": 1691
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.013910066083011e-06,
+      "loss": 0.4785,
+      "step": 1692
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.012612244779259e-06,
+      "loss": 0.4693,
+      "step": 1693
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.011313663546802e-06,
+      "loss": 0.5092,
+      "step": 1694
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.010014322631575e-06,
+      "loss": 0.4786,
+      "step": 1695
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.008714222279649e-06,
+      "loss": 0.4975,
+      "step": 1696
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.007413362737247e-06,
+      "loss": 0.5164,
+      "step": 1697
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.006111744250732e-06,
+      "loss": 0.4582,
+      "step": 1698
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.00480936706661e-06,
+      "loss": 0.5023,
+      "step": 1699
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.003506231431533e-06,
+      "loss": 0.5174,
+      "step": 1700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.002202337592293e-06,
+      "loss": 0.4608,
+      "step": 1701
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.00089768579583e-06,
+      "loss": 0.4645,
+      "step": 1702
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.999592276289226e-06,
+      "loss": 0.4994,
+      "step": 1703
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.998286109319707e-06,
+      "loss": 0.5031,
+      "step": 1704
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.996979185134636e-06,
+      "loss": 0.4689,
+      "step": 1705
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.99567150398153e-06,
+      "loss": 0.4717,
+      "step": 1706
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.994363066108044e-06,
+      "loss": 0.5035,
+      "step": 1707
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.993053871761973e-06,
+      "loss": 0.4529,
+      "step": 1708
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.991743921191261e-06,
+      "loss": 0.5067,
+      "step": 1709
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.990433214643995e-06,
+      "loss": 0.4944,
+      "step": 1710
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.989121752368399e-06,
+      "loss": 0.5294,
+      "step": 1711
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.987809534612848e-06,
+      "loss": 0.5041,
+      "step": 1712
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.986496561625855e-06,
+      "loss": 0.5038,
+      "step": 1713
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.985182833656077e-06,
+      "loss": 0.5036,
+      "step": 1714
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.983868350952316e-06,
+      "loss": 0.4623,
+      "step": 1715
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.982553113763512e-06,
+      "loss": 0.4727,
+      "step": 1716
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.981237122338755e-06,
+      "loss": 0.5037,
+      "step": 1717
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.979920376927273e-06,
+      "loss": 0.5044,
+      "step": 1718
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.978602877778437e-06,
+      "loss": 0.4859,
+      "step": 1719
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.977284625141763e-06,
+      "loss": 0.5286,
+      "step": 1720
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.975965619266907e-06,
+      "loss": 0.5091,
+      "step": 1721
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.974645860403667e-06,
+      "loss": 0.5122,
+      "step": 1722
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.97332534880199e-06,
+      "loss": 0.4879,
+      "step": 1723
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.972004084711959e-06,
+      "loss": 0.4987,
+      "step": 1724
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.9706820683838e-06,
+      "loss": 0.5155,
+      "step": 1725
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.969359300067888e-06,
+      "loss": 0.4834,
+      "step": 1726
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.96803578001473e-06,
+      "loss": 0.4851,
+      "step": 1727
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.966711508474983e-06,
+      "loss": 0.545,
+      "step": 1728
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.965386485699445e-06,
+      "loss": 0.5005,
+      "step": 1729
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.964060711939056e-06,
+      "loss": 0.5155,
+      "step": 1730
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.962734187444894e-06,
+      "loss": 0.508,
+      "step": 1731
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.961406912468189e-06,
+      "loss": 0.5151,
+      "step": 1732
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.960078887260301e-06,
+      "loss": 0.4736,
+      "step": 1733
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.958750112072741e-06,
+      "loss": 0.4903,
+      "step": 1734
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.957420587157159e-06,
+      "loss": 0.486,
+      "step": 1735
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.956090312765347e-06,
+      "loss": 0.5385,
+      "step": 1736
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.954759289149239e-06,
+      "loss": 0.534,
+      "step": 1737
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.95342751656091e-06,
+      "loss": 0.4956,
+      "step": 1738
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.95209499525258e-06,
+      "loss": 0.5,
+      "step": 1739
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.950761725476607e-06,
+      "loss": 0.4866,
+      "step": 1740
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.949427707485492e-06,
+      "loss": 0.5164,
+      "step": 1741
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.94809294153188e-06,
+      "loss": 0.5354,
+      "step": 1742
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.946757427868554e-06,
+      "loss": 0.4984,
+      "step": 1743
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.945421166748437e-06,
+      "loss": 0.4946,
+      "step": 1744
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.944084158424605e-06,
+      "loss": 0.5029,
+      "step": 1745
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.94274640315026e-06,
+      "loss": 0.4862,
+      "step": 1746
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.941407901178757e-06,
+      "loss": 0.4835,
+      "step": 1747
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.940068652763588e-06,
+      "loss": 0.493,
+      "step": 1748
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.938728658158384e-06,
+      "loss": 0.5303,
+      "step": 1749
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.937387917616922e-06,
+      "loss": 0.4944,
+      "step": 1750
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.936046431393117e-06,
+      "loss": 0.5131,
+      "step": 1751
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.934704199741028e-06,
+      "loss": 0.5299,
+      "step": 1752
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.933361222914853e-06,
+      "loss": 0.5278,
+      "step": 1753
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.93201750116893e-06,
+      "loss": 0.4859,
+      "step": 1754
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.930673034757741e-06,
+      "loss": 0.5088,
+      "step": 1755
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.929327823935909e-06,
+      "loss": 0.4984,
+      "step": 1756
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.927981868958198e-06,
+      "loss": 0.4925,
+      "step": 1757
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.926635170079506e-06,
+      "loss": 0.4761,
+      "step": 1758
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.925287727554883e-06,
+      "loss": 0.4854,
+      "step": 1759
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.92393954163951e-06,
+      "loss": 0.4778,
+      "step": 1760
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.922590612588718e-06,
+      "loss": 0.5065,
+      "step": 1761
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.921240940657972e-06,
+      "loss": 0.4927,
+      "step": 1762
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.919890526102879e-06,
+      "loss": 0.4441,
+      "step": 1763
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.918539369179186e-06,
+      "loss": 0.4983,
+      "step": 1764
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.917187470142786e-06,
+      "loss": 0.4882,
+      "step": 1765
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.915834829249706e-06,
+      "loss": 0.4953,
+      "step": 1766
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.914481446756114e-06,
+      "loss": 0.4972,
+      "step": 1767
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.913127322918323e-06,
+      "loss": 0.4904,
+      "step": 1768
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.911772457992785e-06,
+      "loss": 0.4667,
+      "step": 1769
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.910416852236088e-06,
+      "loss": 0.4798,
+      "step": 1770
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.909060505904964e-06,
+      "loss": 0.4828,
+      "step": 1771
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.907703419256287e-06,
+      "loss": 0.5276,
+      "step": 1772
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.906345592547068e-06,
+      "loss": 0.5177,
+      "step": 1773
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.904987026034457e-06,
+      "loss": 0.4829,
+      "step": 1774
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.903627719975746e-06,
+      "loss": 0.4994,
+      "step": 1775
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.90226767462837e-06,
+      "loss": 0.5017,
+      "step": 1776
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.900906890249903e-06,
+      "loss": 0.4811,
+      "step": 1777
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.89954536709805e-06,
+      "loss": 0.496,
+      "step": 1778
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.89818310543067e-06,
+      "loss": 0.505,
+      "step": 1779
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.896820105505752e-06,
+      "loss": 0.4579,
+      "step": 1780
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.895456367581428e-06,
+      "loss": 0.4993,
+      "step": 1781
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.894091891915968e-06,
+      "loss": 0.4969,
+      "step": 1782
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.892726678767787e-06,
+      "loss": 0.4902,
+      "step": 1783
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.891360728395432e-06,
+      "loss": 0.514,
+      "step": 1784
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.889994041057598e-06,
+      "loss": 0.4813,
+      "step": 1785
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.888626617013111e-06,
+      "loss": 0.5426,
+      "step": 1786
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.887258456520942e-06,
+      "loss": 0.4949,
+      "step": 1787
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.885889559840201e-06,
+      "loss": 0.4902,
+      "step": 1788
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.884519927230135e-06,
+      "loss": 0.5204,
+      "step": 1789
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.883149558950132e-06,
+      "loss": 0.4722,
+      "step": 1790
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.881778455259721e-06,
+      "loss": 0.5192,
+      "step": 1791
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.88040661641857e-06,
+      "loss": 0.4979,
+      "step": 1792
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.87903404268648e-06,
+      "loss": 0.4866,
+      "step": 1793
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.877660734323398e-06,
+      "loss": 0.499,
+      "step": 1794
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.876286691589408e-06,
+      "loss": 0.4957,
+      "step": 1795
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.874911914744734e-06,
+      "loss": 0.5041,
+      "step": 1796
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.873536404049738e-06,
+      "loss": 0.5047,
+      "step": 1797
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.87216015976492e-06,
+      "loss": 0.4984,
+      "step": 1798
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.870783182150922e-06,
+      "loss": 0.4934,
+      "step": 1799
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.86940547146852e-06,
+      "loss": 0.4861,
+      "step": 1800
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.868027027978634e-06,
+      "loss": 0.5016,
+      "step": 1801
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.86664785194232e-06,
+      "loss": 0.4904,
+      "step": 1802
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.865267943620774e-06,
+      "loss": 0.4936,
+      "step": 1803
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.86388730327533e-06,
+      "loss": 0.4932,
+      "step": 1804
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.862505931167458e-06,
+      "loss": 0.5281,
+      "step": 1805
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.861123827558772e-06,
+      "loss": 0.5197,
+      "step": 1806
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.859740992711022e-06,
+      "loss": 0.5046,
+      "step": 1807
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.858357426886093e-06,
+      "loss": 0.5213,
+      "step": 1808
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.856973130346015e-06,
+      "loss": 0.4588,
+      "step": 1809
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.855588103352952e-06,
+      "loss": 0.499,
+      "step": 1810
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.854202346169207e-06,
+      "loss": 0.4803,
+      "step": 1811
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.852815859057222e-06,
+      "loss": 0.5219,
+      "step": 1812
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.851428642279576e-06,
+      "loss": 0.4963,
+      "step": 1813
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.850040696098989e-06,
+      "loss": 0.4973,
+      "step": 1814
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.848652020778315e-06,
+      "loss": 0.5068,
+      "step": 1815
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.847262616580551e-06,
+      "loss": 0.5155,
+      "step": 1816
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.845872483768827e-06,
+      "loss": 0.4747,
+      "step": 1817
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.844481622606414e-06,
+      "loss": 0.4965,
+      "step": 1818
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.843090033356721e-06,
+      "loss": 0.5044,
+      "step": 1819
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.841697716283292e-06,
+      "loss": 0.499,
+      "step": 1820
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.840304671649814e-06,
+      "loss": 0.4867,
+      "step": 1821
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.838910899720108e-06,
+      "loss": 0.4875,
+      "step": 1822
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.83751640075813e-06,
+      "loss": 0.4965,
+      "step": 1823
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.836121175027984e-06,
+      "loss": 0.511,
+      "step": 1824
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.834725222793899e-06,
+      "loss": 0.5114,
+      "step": 1825
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.83332854432025e-06,
+      "loss": 0.5102,
+      "step": 1826
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.831931139871547e-06,
+      "loss": 0.4854,
+      "step": 1827
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.830533009712437e-06,
+      "loss": 0.4614,
+      "step": 1828
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.829134154107704e-06,
+      "loss": 0.499,
+      "step": 1829
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.827734573322272e-06,
+      "loss": 0.5159,
+      "step": 1830
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.8263342676212e-06,
+      "loss": 0.4898,
+      "step": 1831
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.824933237269684e-06,
+      "loss": 0.4739,
+      "step": 1832
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.82353148253306e-06,
+      "loss": 0.5066,
+      "step": 1833
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.822129003676798e-06,
+      "loss": 0.5079,
+      "step": 1834
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.820725800966508e-06,
+      "loss": 0.4902,
+      "step": 1835
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.819321874667932e-06,
+      "loss": 0.515,
+      "step": 1836
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.817917225046956e-06,
+      "loss": 0.4814,
+      "step": 1837
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.8165118523696e-06,
+      "loss": 0.5103,
+      "step": 1838
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.81510575690202e-06,
+      "loss": 0.5055,
+      "step": 1839
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.813698938910507e-06,
+      "loss": 0.4969,
+      "step": 1840
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.812291398661494e-06,
+      "loss": 0.4925,
+      "step": 1841
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.810883136421547e-06,
+      "loss": 0.4884,
+      "step": 1842
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.809474152457369e-06,
+      "loss": 0.528,
+      "step": 1843
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.808064447035803e-06,
+      "loss": 0.513,
+      "step": 1844
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.806654020423822e-06,
+      "loss": 0.4787,
+      "step": 1845
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.805242872888545e-06,
+      "loss": 0.4702,
+      "step": 1846
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.803831004697216e-06,
+      "loss": 0.5181,
+      "step": 1847
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.802418416117227e-06,
+      "loss": 0.4838,
+      "step": 1848
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.801005107416098e-06,
+      "loss": 0.4704,
+      "step": 1849
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.799591078861491e-06,
+      "loss": 0.4815,
+      "step": 1850
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.798176330721201e-06,
+      "loss": 0.4961,
+      "step": 1851
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.79676086326316e-06,
+      "loss": 0.5088,
+      "step": 1852
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.795344676755436e-06,
+      "loss": 0.4741,
+      "step": 1853
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.79392777146623e-06,
+      "loss": 0.4754,
+      "step": 1854
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.79251014766389e-06,
+      "loss": 0.4844,
+      "step": 1855
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.791091805616888e-06,
+      "loss": 0.5122,
+      "step": 1856
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.789672745593838e-06,
+      "loss": 0.4995,
+      "step": 1857
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.788252967863487e-06,
+      "loss": 0.496,
+      "step": 1858
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.786832472694723e-06,
+      "loss": 0.5045,
+      "step": 1859
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.785411260356565e-06,
+      "loss": 0.5069,
+      "step": 1860
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.783989331118167e-06,
+      "loss": 0.5241,
+      "step": 1861
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.782566685248824e-06,
+      "loss": 0.5283,
+      "step": 1862
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.781143323017962e-06,
+      "loss": 0.4668,
+      "step": 1863
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.779719244695145e-06,
+      "loss": 0.511,
+      "step": 1864
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.778294450550076e-06,
+      "loss": 0.4982,
+      "step": 1865
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.776868940852583e-06,
+      "loss": 0.5075,
+      "step": 1866
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.77544271587264e-06,
+      "loss": 0.5093,
+      "step": 1867
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.774015775880354e-06,
+      "loss": 0.47,
+      "step": 1868
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.772588121145962e-06,
+      "loss": 0.5008,
+      "step": 1869
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.771159751939844e-06,
+      "loss": 0.4989,
+      "step": 1870
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.769730668532508e-06,
+      "loss": 0.5273,
+      "step": 1871
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.768300871194607e-06,
+      "loss": 0.4798,
+      "step": 1872
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.766870360196917e-06,
+      "loss": 0.5058,
+      "step": 1873
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.765439135810358e-06,
+      "loss": 0.4884,
+      "step": 1874
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.764007198305983e-06,
+      "loss": 0.4753,
+      "step": 1875
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.762574547954979e-06,
+      "loss": 0.501,
+      "step": 1876
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.76114118502867e-06,
+      "loss": 0.5073,
+      "step": 1877
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.75970710979851e-06,
+      "loss": 0.5096,
+      "step": 1878
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.758272322536095e-06,
+      "loss": 0.4837,
+      "step": 1879
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.756836823513149e-06,
+      "loss": 0.4884,
+      "step": 1880
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.755400613001534e-06,
+      "loss": 0.5146,
+      "step": 1881
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.753963691273253e-06,
+      "loss": 0.4802,
+      "step": 1882
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.75252605860043e-06,
+      "loss": 0.4658,
+      "step": 1883
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.751087715255333e-06,
+      "loss": 0.503,
+      "step": 1884
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.749648661510366e-06,
+      "loss": 0.483,
+      "step": 1885
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.748208897638062e-06,
+      "loss": 0.502,
+      "step": 1886
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.746768423911089e-06,
+      "loss": 0.4938,
+      "step": 1887
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.745327240602254e-06,
+      "loss": 0.4769,
+      "step": 1888
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.743885347984494e-06,
+      "loss": 0.5177,
+      "step": 1889
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.742442746330882e-06,
+      "loss": 0.4703,
+      "step": 1890
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.740999435914626e-06,
+      "loss": 0.4864,
+      "step": 1891
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.739555417009067e-06,
+      "loss": 0.4702,
+      "step": 1892
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.73811068988768e-06,
+      "loss": 0.5062,
+      "step": 1893
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.736665254824075e-06,
+      "loss": 0.5148,
+      "step": 1894
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.735219112091996e-06,
+      "loss": 0.5013,
+      "step": 1895
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.733772261965321e-06,
+      "loss": 0.4886,
+      "step": 1896
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.73232470471806e-06,
+      "loss": 0.4982,
+      "step": 1897
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.730876440624361e-06,
+      "loss": 0.4967,
+      "step": 1898
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.729427469958503e-06,
+      "loss": 0.4785,
+      "step": 1899
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.727977792994899e-06,
+      "loss": 0.471,
+      "step": 1900
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.726527410008096e-06,
+      "loss": 0.4919,
+      "step": 1901
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.725076321272775e-06,
+      "loss": 0.5073,
+      "step": 1902
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.72362452706375e-06,
+      "loss": 0.4938,
+      "step": 1903
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.72217202765597e-06,
+      "loss": 0.5084,
+      "step": 1904
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.720718823324517e-06,
+      "loss": 0.4916,
+      "step": 1905
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.719264914344604e-06,
+      "loss": 0.5023,
+      "step": 1906
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.717810300991581e-06,
+      "loss": 0.5193,
+      "step": 1907
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.716354983540929e-06,
+      "loss": 0.4888,
+      "step": 1908
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.714898962268266e-06,
+      "loss": 0.5091,
+      "step": 1909
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.713442237449338e-06,
+      "loss": 0.4635,
+      "step": 1910
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.711984809360027e-06,
+      "loss": 0.494,
+      "step": 1911
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.710526678276349e-06,
+      "loss": 0.5205,
+      "step": 1912
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.709067844474454e-06,
+      "loss": 0.526,
+      "step": 1913
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.707608308230622e-06,
+      "loss": 0.4954,
+      "step": 1914
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.706148069821264e-06,
+      "loss": 0.486,
+      "step": 1915
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.704687129522931e-06,
+      "loss": 0.4929,
+      "step": 1916
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.7032254876123e-06,
+      "loss": 0.4713,
+      "step": 1917
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.70176314436619e-06,
+      "loss": 0.478,
+      "step": 1918
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.700300100061542e-06,
+      "loss": 0.48,
+      "step": 1919
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.698836354975436e-06,
+      "loss": 0.493,
+      "step": 1920
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.697371909385082e-06,
+      "loss": 0.4837,
+      "step": 1921
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.695906763567826e-06,
+      "loss": 0.5049,
+      "step": 1922
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.694440917801144e-06,
+      "loss": 0.4899,
+      "step": 1923
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.692974372362646e-06,
+      "loss": 0.4787,
+      "step": 1924
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.691507127530072e-06,
+      "loss": 0.4798,
+      "step": 1925
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.690039183581298e-06,
+      "loss": 0.4781,
+      "step": 1926
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.68857054079433e-06,
+      "loss": 0.5054,
+      "step": 1927
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.687101199447307e-06,
+      "loss": 0.5413,
+      "step": 1928
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.6856311598185e-06,
+      "loss": 0.5119,
+      "step": 1929
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.684160422186311e-06,
+      "loss": 0.4923,
+      "step": 1930
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.68268898682928e-06,
+      "loss": 0.5081,
+      "step": 1931
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.681216854026071e-06,
+      "loss": 0.4751,
+      "step": 1932
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.679744024055485e-06,
+      "loss": 0.5107,
+      "step": 1933
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.678270497196455e-06,
+      "loss": 0.4937,
+      "step": 1934
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.676796273728043e-06,
+      "loss": 0.4948,
+      "step": 1935
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.675321353929446e-06,
+      "loss": 0.4929,
+      "step": 1936
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.673845738079992e-06,
+      "loss": 0.4801,
+      "step": 1937
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.67236942645914e-06,
+      "loss": 0.5169,
+      "step": 1938
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.670892419346481e-06,
+      "loss": 0.4785,
+      "step": 1939
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.669414717021739e-06,
+      "loss": 0.4859,
+      "step": 1940
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.667936319764767e-06,
+      "loss": 0.4883,
+      "step": 1941
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.666457227855554e-06,
+      "loss": 0.4822,
+      "step": 1942
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.664977441574216e-06,
+      "loss": 0.5097,
+      "step": 1943
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.663496961201003e-06,
+      "loss": 0.4588,
+      "step": 1944
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.662015787016294e-06,
+      "loss": 0.502,
+      "step": 1945
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.660533919300604e-06,
+      "loss": 0.4865,
+      "step": 1946
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.659051358334575e-06,
+      "loss": 0.4869,
+      "step": 1947
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.657568104398983e-06,
+      "loss": 0.5124,
+      "step": 1948
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.656084157774734e-06,
+      "loss": 0.4892,
+      "step": 1949
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.654599518742863e-06,
+      "loss": 0.5197,
+      "step": 1950
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.653114187584542e-06,
+      "loss": 0.4765,
+      "step": 1951
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.651628164581067e-06,
+      "loss": 0.497,
+      "step": 1952
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.650141450013869e-06,
+      "loss": 0.4943,
+      "step": 1953
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.648654044164513e-06,
+      "loss": 0.4861,
+      "step": 1954
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.647165947314686e-06,
+      "loss": 0.5076,
+      "step": 1955
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.645677159746216e-06,
+      "loss": 0.4832,
+      "step": 1956
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.644187681741056e-06,
+      "loss": 0.4983,
+      "step": 1957
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.642697513581288e-06,
+      "loss": 0.4658,
+      "step": 1958
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.641206655549133e-06,
+      "loss": 0.4975,
+      "step": 1959
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.639715107926931e-06,
+      "loss": 0.4946,
+      "step": 1960
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.638222870997161e-06,
+      "loss": 0.4902,
+      "step": 1961
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.636729945042434e-06,
+      "loss": 0.4947,
+      "step": 1962
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.635236330345482e-06,
+      "loss": 0.5133,
+      "step": 1963
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.633742027189178e-06,
+      "loss": 0.5043,
+      "step": 1964
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.632247035856518e-06,
+      "loss": 0.4985,
+      "step": 1965
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.630751356630634e-06,
+      "loss": 0.4979,
+      "step": 1966
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.62925498979478e-06,
+      "loss": 0.5054,
+      "step": 1967
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.627757935632348e-06,
+      "loss": 0.4682,
+      "step": 1968
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.626260194426862e-06,
+      "loss": 0.4888,
+      "step": 1969
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.624761766461965e-06,
+      "loss": 0.5011,
+      "step": 1970
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.62326265202144e-06,
+      "loss": 0.4946,
+      "step": 1971
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.621762851389196e-06,
+      "loss": 0.4957,
+      "step": 1972
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.620262364849275e-06,
+      "loss": 0.4818,
+      "step": 1973
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.618761192685843e-06,
+      "loss": 0.5183,
+      "step": 1974
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.617259335183203e-06,
+      "loss": 0.5154,
+      "step": 1975
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.615756792625783e-06,
+      "loss": 0.5184,
+      "step": 1976
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.61425356529814e-06,
+      "loss": 0.5058,
+      "step": 1977
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.612749653484969e-06,
+      "loss": 0.469,
+      "step": 1978
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.61124505747108e-06,
+      "loss": 0.4884,
+      "step": 1979
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.609739777541427e-06,
+      "loss": 0.489,
+      "step": 1980
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.608233813981087e-06,
+      "loss": 0.5402,
+      "step": 1981
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.606727167075262e-06,
+      "loss": 0.511,
+      "step": 1982
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.605219837109296e-06,
+      "loss": 0.4727,
+      "step": 1983
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.603711824368648e-06,
+      "loss": 0.4629,
+      "step": 1984
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.602203129138914e-06,
+      "loss": 0.5005,
+      "step": 1985
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.600693751705824e-06,
+      "loss": 0.5093,
+      "step": 1986
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.599183692355226e-06,
+      "loss": 0.4981,
+      "step": 1987
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.597672951373104e-06,
+      "loss": 0.4608,
+      "step": 1988
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.59616152904557e-06,
+      "loss": 0.4644,
+      "step": 1989
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.594649425658865e-06,
+      "loss": 0.4983,
+      "step": 1990
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.593136641499358e-06,
+      "loss": 0.4725,
+      "step": 1991
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.591623176853548e-06,
+      "loss": 0.4979,
+      "step": 1992
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.590109032008063e-06,
+      "loss": 0.4939,
+      "step": 1993
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.588594207249658e-06,
+      "loss": 0.4805,
+      "step": 1994
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.58707870286522e-06,
+      "loss": 0.5176,
+      "step": 1995
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.585562519141764e-06,
+      "loss": 0.4799,
+      "step": 1996
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.584045656366426e-06,
+      "loss": 0.4645,
+      "step": 1997
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.582528114826484e-06,
+      "loss": 0.4702,
+      "step": 1998
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.581009894809337e-06,
+      "loss": 0.4861,
+      "step": 1999
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.579490996602509e-06,
+      "loss": 0.467,
+      "step": 2000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.577971420493661e-06,
+      "loss": 0.5035,
+      "step": 2001
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.576451166770575e-06,
+      "loss": 0.4847,
+      "step": 2002
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.574930235721163e-06,
+      "loss": 0.4668,
+      "step": 2003
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.573408627633472e-06,
+      "loss": 0.4629,
+      "step": 2004
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.571886342795667e-06,
+      "loss": 0.5059,
+      "step": 2005
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.570363381496047e-06,
+      "loss": 0.5141,
+      "step": 2006
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.568839744023041e-06,
+      "loss": 0.4563,
+      "step": 2007
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.567315430665199e-06,
+      "loss": 0.4839,
+      "step": 2008
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.565790441711207e-06,
+      "loss": 0.5049,
+      "step": 2009
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.564264777449872e-06,
+      "loss": 0.5152,
+      "step": 2010
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.562738438170131e-06,
+      "loss": 0.4777,
+      "step": 2011
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.561211424161053e-06,
+      "loss": 0.5018,
+      "step": 2012
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.559683735711832e-06,
+      "loss": 0.4932,
+      "step": 2013
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.558155373111787e-06,
+      "loss": 0.4889,
+      "step": 2014
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.556626336650366e-06,
+      "loss": 0.5004,
+      "step": 2015
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.55509662661715e-06,
+      "loss": 0.5037,
+      "step": 2016
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.55356624330184e-06,
+      "loss": 0.489,
+      "step": 2017
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.552035186994268e-06,
+      "loss": 0.4923,
+      "step": 2018
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.550503457984391e-06,
+      "loss": 0.4668,
+      "step": 2019
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.5489710565623e-06,
+      "loss": 0.4732,
+      "step": 2020
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.547437983018207e-06,
+      "loss": 0.5022,
+      "step": 2021
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.545904237642456e-06,
+      "loss": 0.4723,
+      "step": 2022
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.544369820725508e-06,
+      "loss": 0.4839,
+      "step": 2023
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.542834732557964e-06,
+      "loss": 0.5159,
+      "step": 2024
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.541298973430548e-06,
+      "loss": 0.5301,
+      "step": 2025
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.539762543634104e-06,
+      "loss": 0.4897,
+      "step": 2026
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.538225443459616e-06,
+      "loss": 0.5045,
+      "step": 2027
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.536687673198182e-06,
+      "loss": 0.5026,
+      "step": 2028
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.535149233141036e-06,
+      "loss": 0.4641,
+      "step": 2029
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.533610123579532e-06,
+      "loss": 0.4886,
+      "step": 2030
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.532070344805158e-06,
+      "loss": 0.4955,
+      "step": 2031
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.530529897109525e-06,
+      "loss": 0.4878,
+      "step": 2032
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.528988780784369e-06,
+      "loss": 0.5018,
+      "step": 2033
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.527446996121554e-06,
+      "loss": 0.4974,
+      "step": 2034
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.525904543413073e-06,
+      "loss": 0.4868,
+      "step": 2035
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.524361422951041e-06,
+      "loss": 0.5053,
+      "step": 2036
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.522817635027704e-06,
+      "loss": 0.4865,
+      "step": 2037
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.52127317993543e-06,
+      "loss": 0.456,
+      "step": 2038
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.519728057966719e-06,
+      "loss": 0.4942,
+      "step": 2039
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.51818226941419e-06,
+      "loss": 0.4841,
+      "step": 2040
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.516635814570596e-06,
+      "loss": 0.4924,
+      "step": 2041
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.515088693728811e-06,
+      "loss": 0.5137,
+      "step": 2042
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.513540907181836e-06,
+      "loss": 0.5086,
+      "step": 2043
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.5119924552228e-06,
+      "loss": 0.5011,
+      "step": 2044
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.510443338144953e-06,
+      "loss": 0.5321,
+      "step": 2045
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.508893556241678e-06,
+      "loss": 0.5158,
+      "step": 2046
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.507343109806482e-06,
+      "loss": 0.4995,
+      "step": 2047
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.50579199913299e-06,
+      "loss": 0.4516,
+      "step": 2048
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.504240224514965e-06,
+      "loss": 0.5207,
+      "step": 2049
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.502687786246289e-06,
+      "loss": 0.4837,
+      "step": 2050
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.501134684620966e-06,
+      "loss": 0.4914,
+      "step": 2051
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.499580919933137e-06,
+      "loss": 0.4877,
+      "step": 2052
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.498026492477058e-06,
+      "loss": 0.5009,
+      "step": 2053
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.496471402547116e-06,
+      "loss": 0.4487,
+      "step": 2054
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.494915650437818e-06,
+      "loss": 0.5015,
+      "step": 2055
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.493359236443803e-06,
+      "loss": 0.5009,
+      "step": 2056
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.491802160859835e-06,
+      "loss": 0.5225,
+      "step": 2057
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.490244423980796e-06,
+      "loss": 0.4834,
+      "step": 2058
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.4886860261017e-06,
+      "loss": 0.4675,
+      "step": 2059
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.487126967517685e-06,
+      "loss": 0.4754,
+      "step": 2060
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.485567248524012e-06,
+      "loss": 0.5355,
+      "step": 2061
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.484006869416071e-06,
+      "loss": 0.4913,
+      "step": 2062
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.48244583048937e-06,
+      "loss": 0.4924,
+      "step": 2063
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.480884132039551e-06,
+      "loss": 0.5085,
+      "step": 2064
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.479321774362373e-06,
+      "loss": 0.5002,
+      "step": 2065
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.477758757753727e-06,
+      "loss": 0.4849,
+      "step": 2066
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.47619508250962e-06,
+      "loss": 0.4901,
+      "step": 2067
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.474630748926192e-06,
+      "loss": 0.4969,
+      "step": 2068
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.473065757299703e-06,
+      "loss": 0.4978,
+      "step": 2069
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.47150010792654e-06,
+      "loss": 0.5074,
+      "step": 2070
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.469933801103211e-06,
+      "loss": 0.4688,
+      "step": 2071
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.468366837126355e-06,
+      "loss": 0.459,
+      "step": 2072
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.466799216292732e-06,
+      "loss": 0.4976,
+      "step": 2073
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.46523093889922e-06,
+      "loss": 0.4944,
+      "step": 2074
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.463662005242833e-06,
+      "loss": 0.4676,
+      "step": 2075
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.462092415620701e-06,
+      "loss": 0.4719,
+      "step": 2076
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.46052217033008e-06,
+      "loss": 0.4993,
+      "step": 2077
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.458951269668352e-06,
+      "loss": 0.4838,
+      "step": 2078
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.457379713933025e-06,
+      "loss": 0.4497,
+      "step": 2079
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.455807503421722e-06,
+      "loss": 0.501,
+      "step": 2080
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.454234638432201e-06,
+      "loss": 0.4982,
+      "step": 2081
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.452661119262338e-06,
+      "loss": 0.482,
+      "step": 2082
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.451086946210134e-06,
+      "loss": 0.4917,
+      "step": 2083
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.449512119573712e-06,
+      "loss": 0.4958,
+      "step": 2084
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.447936639651322e-06,
+      "loss": 0.5198,
+      "step": 2085
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.446360506741338e-06,
+      "loss": 0.5091,
+      "step": 2086
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.444783721142253e-06,
+      "loss": 0.4921,
+      "step": 2087
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.443206283152688e-06,
+      "loss": 0.4719,
+      "step": 2088
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.441628193071384e-06,
+      "loss": 0.4869,
+      "step": 2089
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.440049451197212e-06,
+      "loss": 0.4673,
+      "step": 2090
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.43847005782916e-06,
+      "loss": 0.4917,
+      "step": 2091
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.436890013266342e-06,
+      "loss": 0.4824,
+      "step": 2092
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.435309317807992e-06,
+      "loss": 0.4719,
+      "step": 2093
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.43372797175347e-06,
+      "loss": 0.4808,
+      "step": 2094
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.432145975402266e-06,
+      "loss": 0.4778,
+      "step": 2095
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.43056332905398e-06,
+      "loss": 0.481,
+      "step": 2096
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.42898003300834e-06,
+      "loss": 0.5107,
+      "step": 2097
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.427396087565208e-06,
+      "loss": 0.5169,
+      "step": 2098
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.42581149302455e-06,
+      "loss": 0.498,
+      "step": 2099
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.424226249686467e-06,
+      "loss": 0.5103,
+      "step": 2100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.422640357851182e-06,
+      "loss": 0.4782,
+      "step": 2101
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.42105381781904e-06,
+      "loss": 0.525,
+      "step": 2102
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.419466629890505e-06,
+      "loss": 0.4944,
+      "step": 2103
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.417878794366168e-06,
+      "loss": 0.4979,
+      "step": 2104
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.416290311546741e-06,
+      "loss": 0.4746,
+      "step": 2105
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.41470118173306e-06,
+      "loss": 0.5148,
+      "step": 2106
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.413111405226082e-06,
+      "loss": 0.4859,
+      "step": 2107
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.411520982326886e-06,
+      "loss": 0.4877,
+      "step": 2108
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.409929913336675e-06,
+      "loss": 0.507,
+      "step": 2109
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.408338198556772e-06,
+      "loss": 0.4705,
+      "step": 2110
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.406745838288628e-06,
+      "loss": 0.4859,
+      "step": 2111
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.405152832833808e-06,
+      "loss": 0.4966,
+      "step": 2112
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.403559182494004e-06,
+      "loss": 0.4893,
+      "step": 2113
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.401964887571033e-06,
+      "loss": 0.5173,
+      "step": 2114
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.400369948366826e-06,
+      "loss": 0.5105,
+      "step": 2115
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.398774365183443e-06,
+      "loss": 0.5189,
+      "step": 2116
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.397178138323063e-06,
+      "loss": 0.4584,
+      "step": 2117
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.395581268087989e-06,
+      "loss": 0.4601,
+      "step": 2118
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.393983754780644e-06,
+      "loss": 0.5169,
+      "step": 2119
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.39238559870357e-06,
+      "loss": 0.5127,
+      "step": 2120
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.390786800159436e-06,
+      "loss": 0.4814,
+      "step": 2121
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.389187359451032e-06,
+      "loss": 0.4907,
+      "step": 2122
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.387587276881266e-06,
+      "loss": 0.4907,
+      "step": 2123
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.38598655275317e-06,
+      "loss": 0.472,
+      "step": 2124
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.384385187369898e-06,
+      "loss": 0.488,
+      "step": 2125
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.382783181034724e-06,
+      "loss": 0.4649,
+      "step": 2126
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.381180534051045e-06,
+      "loss": 0.5218,
+      "step": 2127
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.379577246722377e-06,
+      "loss": 0.4794,
+      "step": 2128
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.377973319352358e-06,
+      "loss": 0.4857,
+      "step": 2129
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.37636875224475e-06,
+      "loss": 0.5018,
+      "step": 2130
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.374763545703433e-06,
+      "loss": 0.4987,
+      "step": 2131
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.373157700032411e-06,
+      "loss": 0.4927,
+      "step": 2132
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.371551215535803e-06,
+      "loss": 0.4833,
+      "step": 2133
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.369944092517856e-06,
+      "loss": 0.5036,
+      "step": 2134
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.368336331282935e-06,
+      "loss": 0.5008,
+      "step": 2135
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.366727932135523e-06,
+      "loss": 0.4717,
+      "step": 2136
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.36511889538023e-06,
+      "loss": 0.5014,
+      "step": 2137
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.363509221321786e-06,
+      "loss": 0.5042,
+      "step": 2138
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.361898910265032e-06,
+      "loss": 0.5041,
+      "step": 2139
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.36028796251494e-06,
+      "loss": 0.4813,
+      "step": 2140
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.358676378376605e-06,
+      "loss": 0.5371,
+      "step": 2141
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.357064158155227e-06,
+      "loss": 0.5122,
+      "step": 2142
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.355451302156143e-06,
+      "loss": 0.5346,
+      "step": 2143
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.353837810684803e-06,
+      "loss": 0.5074,
+      "step": 2144
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.352223684046775e-06,
+      "loss": 0.4826,
+      "step": 2145
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.350608922547753e-06,
+      "loss": 0.4956,
+      "step": 2146
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.348993526493549e-06,
+      "loss": 0.4848,
+      "step": 2147
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.347377496190094e-06,
+      "loss": 0.483,
+      "step": 2148
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.34576083194344e-06,
+      "loss": 0.511,
+      "step": 2149
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.344143534059758e-06,
+      "loss": 0.5144,
+      "step": 2150
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.342525602845343e-06,
+      "loss": 0.5002,
+      "step": 2151
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.340907038606602e-06,
+      "loss": 0.5181,
+      "step": 2152
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.339287841650073e-06,
+      "loss": 0.4552,
+      "step": 2153
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.337668012282404e-06,
+      "loss": 0.4842,
+      "step": 2154
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.336047550810365e-06,
+      "loss": 0.4898,
+      "step": 2155
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.334426457540852e-06,
+      "loss": 0.5213,
+      "step": 2156
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.332804732780871e-06,
+      "loss": 0.4859,
+      "step": 2157
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.331182376837555e-06,
+      "loss": 0.4962,
+      "step": 2158
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.329559390018152e-06,
+      "loss": 0.4995,
+      "step": 2159
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.327935772630033e-06,
+      "loss": 0.4752,
+      "step": 2160
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.326311524980687e-06,
+      "loss": 0.5117,
+      "step": 2161
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.324686647377722e-06,
+      "loss": 0.4752,
+      "step": 2162
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.323061140128865e-06,
+      "loss": 0.4836,
+      "step": 2163
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.321435003541965e-06,
+      "loss": 0.482,
+      "step": 2164
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.319808237924982e-06,
+      "loss": 0.5075,
+      "step": 2165
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.318180843586007e-06,
+      "loss": 0.4928,
+      "step": 2166
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.316552820833242e-06,
+      "loss": 0.4788,
+      "step": 2167
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.314924169975009e-06,
+      "loss": 0.47,
+      "step": 2168
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.313294891319752e-06,
+      "loss": 0.508,
+      "step": 2169
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.311664985176034e-06,
+      "loss": 0.4892,
+      "step": 2170
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.310034451852529e-06,
+      "loss": 0.5304,
+      "step": 2171
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.30840329165804e-06,
+      "loss": 0.4894,
+      "step": 2172
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.306771504901482e-06,
+      "loss": 0.5108,
+      "step": 2173
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.305139091891893e-06,
+      "loss": 0.4848,
+      "step": 2174
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.303506052938427e-06,
+      "loss": 0.4873,
+      "step": 2175
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.301872388350355e-06,
+      "loss": 0.4909,
+      "step": 2176
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.300238098437071e-06,
+      "loss": 0.4885,
+      "step": 2177
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.298603183508084e-06,
+      "loss": 0.5109,
+      "step": 2178
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.296967643873023e-06,
+      "loss": 0.5219,
+      "step": 2179
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.295331479841637e-06,
+      "loss": 0.4812,
+      "step": 2180
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.293694691723786e-06,
+      "loss": 0.4827,
+      "step": 2181
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.292057279829457e-06,
+      "loss": 0.4763,
+      "step": 2182
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.29041924446875e-06,
+      "loss": 0.5264,
+      "step": 2183
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.288780585951885e-06,
+      "loss": 0.4849,
+      "step": 2184
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.287141304589197e-06,
+      "loss": 0.4968,
+      "step": 2185
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.285501400691145e-06,
+      "loss": 0.5068,
+      "step": 2186
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.283860874568299e-06,
+      "loss": 0.5037,
+      "step": 2187
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.282219726531354e-06,
+      "loss": 0.5019,
+      "step": 2188
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.280577956891113e-06,
+      "loss": 0.4988,
+      "step": 2189
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.278935565958508e-06,
+      "loss": 0.5008,
+      "step": 2190
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.27729255404458e-06,
+      "loss": 0.4859,
+      "step": 2191
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.275648921460493e-06,
+      "loss": 0.474,
+      "step": 2192
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.274004668517526e-06,
+      "loss": 0.5065,
+      "step": 2193
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.272359795527074e-06,
+      "loss": 0.4816,
+      "step": 2194
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.270714302800655e-06,
+      "loss": 0.52,
+      "step": 2195
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.269068190649899e-06,
+      "loss": 0.4902,
+      "step": 2196
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.267421459386551e-06,
+      "loss": 0.5129,
+      "step": 2197
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.265774109322484e-06,
+      "loss": 0.5186,
+      "step": 2198
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.264126140769675e-06,
+      "loss": 0.4992,
+      "step": 2199
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.262477554040232e-06,
+      "loss": 0.4752,
+      "step": 2200
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.260828349446363e-06,
+      "loss": 0.499,
+      "step": 2201
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.259178527300412e-06,
+      "loss": 0.485,
+      "step": 2202
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.257528087914825e-06,
+      "loss": 0.4867,
+      "step": 2203
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.255877031602173e-06,
+      "loss": 0.5179,
+      "step": 2204
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.25422535867514e-06,
+      "loss": 0.4998,
+      "step": 2205
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.25257306944653e-06,
+      "loss": 0.4583,
+      "step": 2206
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.250920164229259e-06,
+      "loss": 0.5036,
+      "step": 2207
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.249266643336366e-06,
+      "loss": 0.4799,
+      "step": 2208
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.247612507081e-06,
+      "loss": 0.4957,
+      "step": 2209
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.24595775577643e-06,
+      "loss": 0.5127,
+      "step": 2210
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.244302389736043e-06,
+      "loss": 0.5001,
+      "step": 2211
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.24264640927334e-06,
+      "loss": 0.5018,
+      "step": 2212
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.240989814701938e-06,
+      "loss": 0.4797,
+      "step": 2213
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.23933260633557e-06,
+      "loss": 0.518,
+      "step": 2214
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.23767478448809e-06,
+      "loss": 0.4552,
+      "step": 2215
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.236016349473464e-06,
+      "loss": 0.5088,
+      "step": 2216
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.23435730160577e-06,
+      "loss": 0.4799,
+      "step": 2217
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.232697641199211e-06,
+      "loss": 0.4867,
+      "step": 2218
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.231037368568104e-06,
+      "loss": 0.5083,
+      "step": 2219
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.229376484026875e-06,
+      "loss": 0.4813,
+      "step": 2220
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.227714987890074e-06,
+      "loss": 0.5021,
+      "step": 2221
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.22605288047236e-06,
+      "loss": 0.519,
+      "step": 2222
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.224390162088516e-06,
+      "loss": 0.4941,
+      "step": 2223
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.222726833053432e-06,
+      "loss": 0.4429,
+      "step": 2224
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.221062893682117e-06,
+      "loss": 0.4852,
+      "step": 2225
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.219398344289702e-06,
+      "loss": 0.4931,
+      "step": 2226
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.21773318519142e-06,
+      "loss": 0.4696,
+      "step": 2227
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.216067416702636e-06,
+      "loss": 0.5268,
+      "step": 2228
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.214401039138812e-06,
+      "loss": 0.5218,
+      "step": 2229
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.212734052815543e-06,
+      "loss": 0.5138,
+      "step": 2230
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.211066458048528e-06,
+      "loss": 0.5256,
+      "step": 2231
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.209398255153584e-06,
+      "loss": 0.5179,
+      "step": 2232
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.207729444446645e-06,
+      "loss": 0.5125,
+      "step": 2233
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.206060026243755e-06,
+      "loss": 0.4716,
+      "step": 2234
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.204390000861082e-06,
+      "loss": 0.4981,
+      "step": 2235
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.202719368614902e-06,
+      "loss": 0.4861,
+      "step": 2236
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.201048129821605e-06,
+      "loss": 0.5056,
+      "step": 2237
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.199376284797704e-06,
+      "loss": 0.5126,
+      "step": 2238
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.197703833859817e-06,
+      "loss": 0.4556,
+      "step": 2239
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.196030777324681e-06,
+      "loss": 0.4969,
+      "step": 2240
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.194357115509153e-06,
+      "loss": 0.4912,
+      "step": 2241
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.192682848730194e-06,
+      "loss": 0.4831,
+      "step": 2242
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.191007977304887e-06,
+      "loss": 0.5016,
+      "step": 2243
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.189332501550428e-06,
+      "loss": 0.4787,
+      "step": 2244
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.187656421784126e-06,
+      "loss": 0.4967,
+      "step": 2245
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.185979738323406e-06,
+      "loss": 0.5061,
+      "step": 2246
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.184302451485808e-06,
+      "loss": 0.5072,
+      "step": 2247
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.182624561588983e-06,
+      "loss": 0.4999,
+      "step": 2248
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.180946068950698e-06,
+      "loss": 0.5248,
+      "step": 2249
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.179266973888837e-06,
+      "loss": 0.4728,
+      "step": 2250
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.177587276721392e-06,
+      "loss": 0.5002,
+      "step": 2251
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.175906977766476e-06,
+      "loss": 0.5066,
+      "step": 2252
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.17422607734231e-06,
+      "loss": 0.4825,
+      "step": 2253
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.172544575767235e-06,
+      "loss": 0.5005,
+      "step": 2254
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.170862473359698e-06,
+      "loss": 0.4717,
+      "step": 2255
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.169179770438264e-06,
+      "loss": 0.5238,
+      "step": 2256
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.167496467321614e-06,
+      "loss": 0.5262,
+      "step": 2257
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.16581256432854e-06,
+      "loss": 0.4626,
+      "step": 2258
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.164128061777947e-06,
+      "loss": 0.4967,
+      "step": 2259
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.162442959988856e-06,
+      "loss": 0.5058,
+      "step": 2260
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.160757259280401e-06,
+      "loss": 0.5045,
+      "step": 2261
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.159070959971824e-06,
+      "loss": 0.4852,
+      "step": 2262
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.15738406238249e-06,
+      "loss": 0.4863,
+      "step": 2263
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.155696566831868e-06,
+      "loss": 0.4934,
+      "step": 2264
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.154008473639547e-06,
+      "loss": 0.4642,
+      "step": 2265
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.152319783125225e-06,
+      "loss": 0.4938,
+      "step": 2266
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.150630495608717e-06,
+      "loss": 0.4935,
+      "step": 2267
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.148940611409947e-06,
+      "loss": 0.5085,
+      "step": 2268
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.147250130848954e-06,
+      "loss": 0.4833,
+      "step": 2269
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.145559054245889e-06,
+      "loss": 0.4884,
+      "step": 2270
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.143867381921018e-06,
+      "loss": 0.5039,
+      "step": 2271
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.142175114194717e-06,
+      "loss": 0.5012,
+      "step": 2272
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.140482251387478e-06,
+      "loss": 0.5097,
+      "step": 2273
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.138788793819903e-06,
+      "loss": 0.5112,
+      "step": 2274
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.137094741812708e-06,
+      "loss": 0.5009,
+      "step": 2275
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.135400095686719e-06,
+      "loss": 0.481,
+      "step": 2276
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.133704855762878e-06,
+      "loss": 0.4856,
+      "step": 2277
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.13200902236224e-06,
+      "loss": 0.4884,
+      "step": 2278
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.130312595805967e-06,
+      "loss": 0.4873,
+      "step": 2279
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.12861557641534e-06,
+      "loss": 0.5266,
+      "step": 2280
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.126917964511746e-06,
+      "loss": 0.504,
+      "step": 2281
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.125219760416688e-06,
+      "loss": 0.4987,
+      "step": 2282
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.123520964451782e-06,
+      "loss": 0.5039,
+      "step": 2283
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.121821576938755e-06,
+      "loss": 0.4857,
+      "step": 2284
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.120121598199444e-06,
+      "loss": 0.476,
+      "step": 2285
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.118421028555799e-06,
+      "loss": 0.5169,
+      "step": 2286
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.116719868329885e-06,
+      "loss": 0.4822,
+      "step": 2287
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.115018117843874e-06,
+      "loss": 0.5113,
+      "step": 2288
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.11331577742005e-06,
+      "loss": 0.5055,
+      "step": 2289
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.111612847380817e-06,
+      "loss": 0.5197,
+      "step": 2290
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.109909328048679e-06,
+      "loss": 0.5123,
+      "step": 2291
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.108205219746259e-06,
+      "loss": 0.4868,
+      "step": 2292
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.10650052279629e-06,
+      "loss": 0.4985,
+      "step": 2293
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.104795237521617e-06,
+      "loss": 0.5273,
+      "step": 2294
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.103089364245191e-06,
+      "loss": 0.4665,
+      "step": 2295
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.101382903290085e-06,
+      "loss": 0.4591,
+      "step": 2296
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.099675854979475e-06,
+      "loss": 0.5326,
+      "step": 2297
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.097968219636646e-06,
+      "loss": 0.489,
+      "step": 2298
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.096259997585004e-06,
+      "loss": 0.5103,
+      "step": 2299
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.094551189148059e-06,
+      "loss": 0.468,
+      "step": 2300
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.092841794649435e-06,
+      "loss": 0.5126,
+      "step": 2301
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.09113181441286e-06,
+      "loss": 0.4946,
+      "step": 2302
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.089421248762187e-06,
+      "loss": 0.5208,
+      "step": 2303
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.087710098021368e-06,
+      "loss": 0.4994,
+      "step": 2304
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.085998362514467e-06,
+      "loss": 0.49,
+      "step": 2305
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.084286042565664e-06,
+      "loss": 0.5046,
+      "step": 2306
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.082573138499246e-06,
+      "loss": 0.5343,
+      "step": 2307
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.080859650639609e-06,
+      "loss": 0.5089,
+      "step": 2308
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.079145579311266e-06,
+      "loss": 0.4995,
+      "step": 2309
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.077430924838834e-06,
+      "loss": 0.5243,
+      "step": 2310
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.075715687547044e-06,
+      "loss": 0.471,
+      "step": 2311
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.073999867760736e-06,
+      "loss": 0.4839,
+      "step": 2312
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.072283465804859e-06,
+      "loss": 0.5243,
+      "step": 2313
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.070566482004475e-06,
+      "loss": 0.5178,
+      "step": 2314
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.068848916684757e-06,
+      "loss": 0.4866,
+      "step": 2315
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.067130770170984e-06,
+      "loss": 0.496,
+      "step": 2316
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.065412042788545e-06,
+      "loss": 0.5029,
+      "step": 2317
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.063692734862948e-06,
+      "loss": 0.4915,
+      "step": 2318
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.061972846719796e-06,
+      "loss": 0.4931,
+      "step": 2319
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.060252378684817e-06,
+      "loss": 0.5209,
+      "step": 2320
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.058531331083838e-06,
+      "loss": 0.4907,
+      "step": 2321
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.056809704242801e-06,
+      "loss": 0.518,
+      "step": 2322
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.055087498487756e-06,
+      "loss": 0.4777,
+      "step": 2323
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.053364714144862e-06,
+      "loss": 0.499,
+      "step": 2324
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.05164135154039e-06,
+      "loss": 0.4645,
+      "step": 2325
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.04991741100072e-06,
+      "loss": 0.4829,
+      "step": 2326
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.048192892852337e-06,
+      "loss": 0.4851,
+      "step": 2327
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.046467797421843e-06,
+      "loss": 0.5106,
+      "step": 2328
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.044742125035942e-06,
+      "loss": 0.4639,
+      "step": 2329
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.043015876021454e-06,
+      "loss": 0.5059,
+      "step": 2330
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.041289050705302e-06,
+      "loss": 0.4821,
+      "step": 2331
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.03956164941452e-06,
+      "loss": 0.4737,
+      "step": 2332
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.037833672476253e-06,
+      "loss": 0.4704,
+      "step": 2333
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.036105120217756e-06,
+      "loss": 0.4886,
+      "step": 2334
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.03437599296639e-06,
+      "loss": 0.4908,
+      "step": 2335
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.032646291049624e-06,
+      "loss": 0.5158,
+      "step": 2336
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.030916014795038e-06,
+      "loss": 0.5019,
+      "step": 2337
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.029185164530322e-06,
+      "loss": 0.4814,
+      "step": 2338
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.027453740583275e-06,
+      "loss": 0.4914,
+      "step": 2339
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.025721743281798e-06,
+      "loss": 0.4929,
+      "step": 2340
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.023989172953907e-06,
+      "loss": 0.5069,
+      "step": 2341
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.022256029927726e-06,
+      "loss": 0.4852,
+      "step": 2342
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.020522314531484e-06,
+      "loss": 0.4819,
+      "step": 2343
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.018788027093525e-06,
+      "loss": 0.4828,
+      "step": 2344
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.017053167942294e-06,
+      "loss": 0.4956,
+      "step": 2345
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.01531773740635e-06,
+      "loss": 0.4842,
+      "step": 2346
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.01358173581435e-06,
+      "loss": 0.4698,
+      "step": 2347
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.011845163495079e-06,
+      "loss": 0.4864,
+      "step": 2348
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.010108020777406e-06,
+      "loss": 0.5051,
+      "step": 2349
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.008370307990327e-06,
+      "loss": 0.4981,
+      "step": 2350
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.006632025462937e-06,
+      "loss": 0.4913,
+      "step": 2351
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.004893173524438e-06,
+      "loss": 0.4542,
+      "step": 2352
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.003153752504146e-06,
+      "loss": 0.4826,
+      "step": 2353
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.001413762731483e-06,
+      "loss": 0.4775,
+      "step": 2354
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.999673204535972e-06,
+      "loss": 0.4806,
+      "step": 2355
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.99793207824725e-06,
+      "loss": 0.5266,
+      "step": 2356
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.996190384195062e-06,
+      "loss": 0.502,
+      "step": 2357
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.994448122709259e-06,
+      "loss": 0.4951,
+      "step": 2358
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.992705294119796e-06,
+      "loss": 0.4798,
+      "step": 2359
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.990961898756743e-06,
+      "loss": 0.504,
+      "step": 2360
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.989217936950271e-06,
+      "loss": 0.501,
+      "step": 2361
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.987473409030659e-06,
+      "loss": 0.5101,
+      "step": 2362
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.985728315328296e-06,
+      "loss": 0.5109,
+      "step": 2363
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.983982656173676e-06,
+      "loss": 0.4744,
+      "step": 2364
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.982236431897402e-06,
+      "loss": 0.534,
+      "step": 2365
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.98048964283018e-06,
+      "loss": 0.4763,
+      "step": 2366
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.978742289302828e-06,
+      "loss": 0.4601,
+      "step": 2367
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.976994371646268e-06,
+      "loss": 0.4806,
+      "step": 2368
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.97524589019153e-06,
+      "loss": 0.4917,
+      "step": 2369
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.973496845269748e-06,
+      "loss": 0.4698,
+      "step": 2370
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.971747237212169e-06,
+      "loss": 0.4908,
+      "step": 2371
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.969997066350139e-06,
+      "loss": 0.4764,
+      "step": 2372
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.968246333015115e-06,
+      "loss": 0.4688,
+      "step": 2373
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.966495037538663e-06,
+      "loss": 0.4993,
+      "step": 2374
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.964743180252446e-06,
+      "loss": 0.5056,
+      "step": 2375
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.962990761488243e-06,
+      "loss": 0.5218,
+      "step": 2376
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.961237781577935e-06,
+      "loss": 0.4968,
+      "step": 2377
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.959484240853512e-06,
+      "loss": 0.5153,
+      "step": 2378
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.957730139647068e-06,
+      "loss": 0.4997,
+      "step": 2379
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.9559754782908e-06,
+      "loss": 0.4939,
+      "step": 2380
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.95422025711702e-06,
+      "loss": 0.4691,
+      "step": 2381
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.952464476458137e-06,
+      "loss": 0.4939,
+      "step": 2382
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.95070813664667e-06,
+      "loss": 0.4858,
+      "step": 2383
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.948951238015244e-06,
+      "loss": 0.4709,
+      "step": 2384
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.94719378089659e-06,
+      "loss": 0.4986,
+      "step": 2385
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.945435765623543e-06,
+      "loss": 0.4794,
+      "step": 2386
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.943677192529044e-06,
+      "loss": 0.495,
+      "step": 2387
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.941918061946144e-06,
+      "loss": 0.4954,
+      "step": 2388
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.940158374207991e-06,
+      "loss": 0.4673,
+      "step": 2389
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.938398129647847e-06,
+      "loss": 0.4705,
+      "step": 2390
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.936637328599075e-06,
+      "loss": 0.4965,
+      "step": 2391
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.934875971395147e-06,
+      "loss": 0.524,
+      "step": 2392
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.93311405836963e-06,
+      "loss": 0.4878,
+      "step": 2393
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.931351589856212e-06,
+      "loss": 0.509,
+      "step": 2394
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.929588566188675e-06,
+      "loss": 0.4791,
+      "step": 2395
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.92782498770091e-06,
+      "loss": 0.4917,
+      "step": 2396
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.926060854726911e-06,
+      "loss": 0.5021,
+      "step": 2397
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.92429616760078e-06,
+      "loss": 0.4775,
+      "step": 2398
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.92253092665672e-06,
+      "loss": 0.4901,
+      "step": 2399
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.920765132229044e-06,
+      "loss": 0.4874,
+      "step": 2400
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.918998784652165e-06,
+      "loss": 0.5074,
+      "step": 2401
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.917231884260605e-06,
+      "loss": 0.5032,
+      "step": 2402
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.915464431388986e-06,
+      "loss": 0.4779,
+      "step": 2403
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.91369642637204e-06,
+      "loss": 0.509,
+      "step": 2404
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.911927869544598e-06,
+      "loss": 0.4697,
+      "step": 2405
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.910158761241597e-06,
+      "loss": 0.4983,
+      "step": 2406
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.908389101798083e-06,
+      "loss": 0.4664,
+      "step": 2407
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.906618891549203e-06,
+      "loss": 0.4841,
+      "step": 2408
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.904848130830207e-06,
+      "loss": 0.4775,
+      "step": 2409
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.90307681997645e-06,
+      "loss": 0.4991,
+      "step": 2410
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.901304959323393e-06,
+      "loss": 0.5236,
+      "step": 2411
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.899532549206599e-06,
+      "loss": 0.4783,
+      "step": 2412
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.897759589961735e-06,
+      "loss": 0.4783,
+      "step": 2413
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.895986081924576e-06,
+      "loss": 0.5108,
+      "step": 2414
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.894212025430996e-06,
+      "loss": 0.4938,
+      "step": 2415
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.892437420816973e-06,
+      "loss": 0.4737,
+      "step": 2416
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.890662268418595e-06,
+      "loss": 0.4729,
+      "step": 2417
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.888886568572045e-06,
+      "loss": 0.4919,
+      "step": 2418
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.887110321613616e-06,
+      "loss": 0.4569,
+      "step": 2419
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.885333527879702e-06,
+      "loss": 0.4771,
+      "step": 2420
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.883556187706802e-06,
+      "loss": 0.5046,
+      "step": 2421
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.881778301431517e-06,
+      "loss": 0.4872,
+      "step": 2422
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.879999869390553e-06,
+      "loss": 0.4922,
+      "step": 2423
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.878220891920716e-06,
+      "loss": 0.48,
+      "step": 2424
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.876441369358918e-06,
+      "loss": 0.4792,
+      "step": 2425
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.874661302042179e-06,
+      "loss": 0.4991,
+      "step": 2426
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.87288069030761e-06,
+      "loss": 0.5377,
+      "step": 2427
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.871099534492437e-06,
+      "loss": 0.4888,
+      "step": 2428
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.869317834933982e-06,
+      "loss": 0.4548,
+      "step": 2429
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.867535591969675e-06,
+      "loss": 0.5004,
+      "step": 2430
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.865752805937042e-06,
+      "loss": 0.5137,
+      "step": 2431
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.863969477173717e-06,
+      "loss": 0.4959,
+      "step": 2432
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.862185606017438e-06,
+      "loss": 0.5158,
+      "step": 2433
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.860401192806044e-06,
+      "loss": 0.4957,
+      "step": 2434
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.858616237877472e-06,
+      "loss": 0.4723,
+      "step": 2435
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.85683074156977e-06,
+      "loss": 0.4472,
+      "step": 2436
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.855044704221082e-06,
+      "loss": 0.5023,
+      "step": 2437
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.853258126169655e-06,
+      "loss": 0.4649,
+      "step": 2438
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.851471007753846e-06,
+      "loss": 0.5043,
+      "step": 2439
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.849683349312105e-06,
+      "loss": 0.4945,
+      "step": 2440
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.847895151182985e-06,
+      "loss": 0.5002,
+      "step": 2441
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.84610641370515e-06,
+      "loss": 0.4701,
+      "step": 2442
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.844317137217354e-06,
+      "loss": 0.4946,
+      "step": 2443
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.842527322058463e-06,
+      "loss": 0.5095,
+      "step": 2444
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.84073696856744e-06,
+      "loss": 0.5034,
+      "step": 2445
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.838946077083349e-06,
+      "loss": 0.4793,
+      "step": 2446
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.837154647945364e-06,
+      "loss": 0.4676,
+      "step": 2447
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.835362681492749e-06,
+      "loss": 0.5164,
+      "step": 2448
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.833570178064878e-06,
+      "loss": 0.5048,
+      "step": 2449
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.831777138001224e-06,
+      "loss": 0.503,
+      "step": 2450
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.829983561641364e-06,
+      "loss": 0.5012,
+      "step": 2451
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.82818944932497e-06,
+      "loss": 0.5087,
+      "step": 2452
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.826394801391824e-06,
+      "loss": 0.5064,
+      "step": 2453
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.824599618181804e-06,
+      "loss": 0.5161,
+      "step": 2454
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.822803900034893e-06,
+      "loss": 0.4718,
+      "step": 2455
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.821007647291168e-06,
+      "loss": 0.4799,
+      "step": 2456
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.819210860290817e-06,
+      "loss": 0.506,
+      "step": 2457
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.817413539374121e-06,
+      "loss": 0.4879,
+      "step": 2458
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.81561568488147e-06,
+      "loss": 0.4988,
+      "step": 2459
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.813817297153347e-06,
+      "loss": 0.5012,
+      "step": 2460
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.812018376530341e-06,
+      "loss": 0.476,
+      "step": 2461
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.810218923353143e-06,
+      "loss": 0.5007,
+      "step": 2462
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.808418937962538e-06,
+      "loss": 0.5014,
+      "step": 2463
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.806618420699418e-06,
+      "loss": 0.5214,
+      "step": 2464
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.804817371904778e-06,
+      "loss": 0.486,
+      "step": 2465
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.803015791919704e-06,
+      "loss": 0.4974,
+      "step": 2466
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.801213681085391e-06,
+      "loss": 0.4774,
+      "step": 2467
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.799411039743132e-06,
+      "loss": 0.5032,
+      "step": 2468
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.79760786823432e-06,
+      "loss": 0.4766,
+      "step": 2469
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.795804166900448e-06,
+      "loss": 0.479,
+      "step": 2470
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.793999936083108e-06,
+      "loss": 0.5023,
+      "step": 2471
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.792195176123999e-06,
+      "loss": 0.492,
+      "step": 2472
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.790389887364913e-06,
+      "loss": 0.5003,
+      "step": 2473
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.788584070147745e-06,
+      "loss": 0.4889,
+      "step": 2474
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.786777724814489e-06,
+      "loss": 0.4931,
+      "step": 2475
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.784970851707241e-06,
+      "loss": 0.4899,
+      "step": 2476
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.783163451168195e-06,
+      "loss": 0.4852,
+      "step": 2477
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.781355523539647e-06,
+      "loss": 0.496,
+      "step": 2478
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.779547069163988e-06,
+      "loss": 0.4788,
+      "step": 2479
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.777738088383718e-06,
+      "loss": 0.4774,
+      "step": 2480
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.775928581541423e-06,
+      "loss": 0.5068,
+      "step": 2481
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.774118548979803e-06,
+      "loss": 0.5317,
+      "step": 2482
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.77230799104165e-06,
+      "loss": 0.5004,
+      "step": 2483
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.770496908069854e-06,
+      "loss": 0.4954,
+      "step": 2484
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.768685300407409e-06,
+      "loss": 0.5091,
+      "step": 2485
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.766873168397403e-06,
+      "loss": 0.488,
+      "step": 2486
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.765060512383034e-06,
+      "loss": 0.4843,
+      "step": 2487
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.763247332707585e-06,
+      "loss": 0.4974,
+      "step": 2488
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.761433629714448e-06,
+      "loss": 0.4662,
+      "step": 2489
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.759619403747112e-06,
+      "loss": 0.5285,
+      "step": 2490
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.757804655149161e-06,
+      "loss": 0.4932,
+      "step": 2491
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.755989384264284e-06,
+      "loss": 0.4809,
+      "step": 2492
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.754173591436267e-06,
+      "loss": 0.5207,
+      "step": 2493
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.75235727700899e-06,
+      "loss": 0.488,
+      "step": 2494
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.75054044132644e-06,
+      "loss": 0.5072,
+      "step": 2495
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.748723084732694e-06,
+      "loss": 0.4713,
+      "step": 2496
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.746905207571936e-06,
+      "loss": 0.5122,
+      "step": 2497
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.745086810188445e-06,
+      "loss": 0.4669,
+      "step": 2498
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.743267892926596e-06,
+      "loss": 0.4768,
+      "step": 2499
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.741448456130864e-06,
+      "loss": 0.5019,
+      "step": 2500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.739628500145826e-06,
+      "loss": 0.497,
+      "step": 2501
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.737808025316154e-06,
+      "loss": 0.497,
+      "step": 2502
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.735987031986613e-06,
+      "loss": 0.5052,
+      "step": 2503
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.734165520502082e-06,
+      "loss": 0.4939,
+      "step": 2504
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.73234349120752e-06,
+      "loss": 0.4917,
+      "step": 2505
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.730520944447994e-06,
+      "loss": 0.4758,
+      "step": 2506
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.72869788056867e-06,
+      "loss": 0.4853,
+      "step": 2507
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.726874299914805e-06,
+      "loss": 0.4678,
+      "step": 2508
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.72505020283176e-06,
+      "loss": 0.4728,
+      "step": 2509
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.72322558966499e-06,
+      "loss": 0.4761,
+      "step": 2510
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.721400460760052e-06,
+      "loss": 0.4568,
+      "step": 2511
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.719574816462597e-06,
+      "loss": 0.479,
+      "step": 2512
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.717748657118374e-06,
+      "loss": 0.5086,
+      "step": 2513
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.715921983073231e-06,
+      "loss": 0.5089,
+      "step": 2514
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.714094794673112e-06,
+      "loss": 0.5122,
+      "step": 2515
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.71226709226406e-06,
+      "loss": 0.5041,
+      "step": 2516
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.710438876192215e-06,
+      "loss": 0.4945,
+      "step": 2517
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.708610146803812e-06,
+      "loss": 0.5252,
+      "step": 2518
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.706780904445185e-06,
+      "loss": 0.4786,
+      "step": 2519
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.704951149462768e-06,
+      "loss": 0.4779,
+      "step": 2520
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.703120882203086e-06,
+      "loss": 0.4922,
+      "step": 2521
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.701290103012765e-06,
+      "loss": 0.537,
+      "step": 2522
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.699458812238528e-06,
+      "loss": 0.4873,
+      "step": 2523
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.697627010227194e-06,
+      "loss": 0.4955,
+      "step": 2524
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.695794697325678e-06,
+      "loss": 0.477,
+      "step": 2525
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.693961873880993e-06,
+      "loss": 0.4657,
+      "step": 2526
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.692128540240248e-06,
+      "loss": 0.501,
+      "step": 2527
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.690294696750651e-06,
+      "loss": 0.4687,
+      "step": 2528
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.688460343759502e-06,
+      "loss": 0.5239,
+      "step": 2529
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.686625481614201e-06,
+      "loss": 0.4889,
+      "step": 2530
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.684790110662243e-06,
+      "loss": 0.5037,
+      "step": 2531
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.682954231251221e-06,
+      "loss": 0.5099,
+      "step": 2532
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.681117843728822e-06,
+      "loss": 0.4652,
+      "step": 2533
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.679280948442832e-06,
+      "loss": 0.4894,
+      "step": 2534
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.677443545741127e-06,
+      "loss": 0.4911,
+      "step": 2535
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.675605635971688e-06,
+      "loss": 0.4735,
+      "step": 2536
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.673767219482588e-06,
+      "loss": 0.4505,
+      "step": 2537
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.671928296621992e-06,
+      "loss": 0.4647,
+      "step": 2538
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.670088867738166e-06,
+      "loss": 0.5169,
+      "step": 2539
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.668248933179471e-06,
+      "loss": 0.4902,
+      "step": 2540
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.666408493294364e-06,
+      "loss": 0.4917,
+      "step": 2541
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.664567548431393e-06,
+      "loss": 0.4871,
+      "step": 2542
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.662726098939211e-06,
+      "loss": 0.4774,
+      "step": 2543
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.660884145166556e-06,
+      "loss": 0.4763,
+      "step": 2544
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.659041687462268e-06,
+      "loss": 0.4817,
+      "step": 2545
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.657198726175279e-06,
+      "loss": 0.4436,
+      "step": 2546
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.655355261654625e-06,
+      "loss": 0.4917,
+      "step": 2547
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.653511294249423e-06,
+      "loss": 0.4938,
+      "step": 2548
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.651666824308895e-06,
+      "loss": 0.4923,
+      "step": 2549
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.64982185218236e-06,
+      "loss": 0.5017,
+      "step": 2550
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.647976378219223e-06,
+      "loss": 0.5111,
+      "step": 2551
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.646130402768991e-06,
+      "loss": 0.5189,
+      "step": 2552
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.644283926181265e-06,
+      "loss": 0.4814,
+      "step": 2553
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.64243694880574e-06,
+      "loss": 0.5259,
+      "step": 2554
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.640589470992204e-06,
+      "loss": 0.4661,
+      "step": 2555
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.638741493090543e-06,
+      "loss": 0.4893,
+      "step": 2556
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.636893015450735e-06,
+      "loss": 0.4919,
+      "step": 2557
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.635044038422857e-06,
+      "loss": 0.4753,
+      "step": 2558
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.633194562357076e-06,
+      "loss": 0.4848,
+      "step": 2559
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.631344587603653e-06,
+      "loss": 0.4823,
+      "step": 2560
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.629494114512948e-06,
+      "loss": 0.4959,
+      "step": 2561
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.627643143435412e-06,
+      "loss": 0.4526,
+      "step": 2562
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.6257916747215924e-06,
+      "loss": 0.4573,
+      "step": 2563
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.623939708722129e-06,
+      "loss": 0.4873,
+      "step": 2564
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.622087245787755e-06,
+      "loss": 0.5207,
+      "step": 2565
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.620234286269301e-06,
+      "loss": 0.4539,
+      "step": 2566
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.618380830517689e-06,
+      "loss": 0.5031,
+      "step": 2567
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.616526878883937e-06,
+      "loss": 0.4265,
+      "step": 2568
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.614672431719152e-06,
+      "loss": 0.4644,
+      "step": 2569
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.612817489374544e-06,
+      "loss": 0.4727,
+      "step": 2570
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.610962052201407e-06,
+      "loss": 0.453,
+      "step": 2571
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.609106120551134e-06,
+      "loss": 0.471,
+      "step": 2572
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.607249694775212e-06,
+      "loss": 0.461,
+      "step": 2573
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.60539277522522e-06,
+      "loss": 0.5018,
+      "step": 2574
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.603535362252829e-06,
+      "loss": 0.4931,
+      "step": 2575
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.6016774562098075e-06,
+      "loss": 0.4835,
+      "step": 2576
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.599819057448013e-06,
+      "loss": 0.5095,
+      "step": 2577
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.597960166319399e-06,
+      "loss": 0.5092,
+      "step": 2578
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.596100783176015e-06,
+      "loss": 0.509,
+      "step": 2579
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.594240908369994e-06,
+      "loss": 0.4765,
+      "step": 2580
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.592380542253572e-06,
+      "loss": 0.4715,
+      "step": 2581
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.590519685179078e-06,
+      "loss": 0.4772,
+      "step": 2582
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.588658337498924e-06,
+      "loss": 0.495,
+      "step": 2583
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.586796499565625e-06,
+      "loss": 0.4987,
+      "step": 2584
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5849341717317835e-06,
+      "loss": 0.4652,
+      "step": 2585
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5830713543501e-06,
+      "loss": 0.4804,
+      "step": 2586
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5812080477733585e-06,
+      "loss": 0.5087,
+      "step": 2587
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.579344252354446e-06,
+      "loss": 0.4732,
+      "step": 2588
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.577479968446337e-06,
+      "loss": 0.4628,
+      "step": 2589
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.575615196402097e-06,
+      "loss": 0.4993,
+      "step": 2590
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.573749936574888e-06,
+      "loss": 0.5048,
+      "step": 2591
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.57188418931796e-06,
+      "loss": 0.4651,
+      "step": 2592
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.570017954984659e-06,
+      "loss": 0.5156,
+      "step": 2593
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.568151233928422e-06,
+      "loss": 0.5021,
+      "step": 2594
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5662840265027794e-06,
+      "loss": 0.469,
+      "step": 2595
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.564416333061349e-06,
+      "loss": 0.5127,
+      "step": 2596
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.562548153957845e-06,
+      "loss": 0.4597,
+      "step": 2597
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5606794895460744e-06,
+      "loss": 0.4939,
+      "step": 2598
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.558810340179932e-06,
+      "loss": 0.5069,
+      "step": 2599
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.55694070621341e-06,
+      "loss": 0.5011,
+      "step": 2600
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.555070588000585e-06,
+      "loss": 0.4746,
+      "step": 2601
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.553199985895632e-06,
+      "loss": 0.489,
+      "step": 2602
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.551328900252814e-06,
+      "loss": 0.496,
+      "step": 2603
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.549457331426486e-06,
+      "loss": 0.4621,
+      "step": 2604
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.547585279771098e-06,
+      "loss": 0.5236,
+      "step": 2605
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.545712745641187e-06,
+      "loss": 0.5005,
+      "step": 2606
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.543839729391383e-06,
+      "loss": 0.4994,
+      "step": 2607
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.541966231376403e-06,
+      "loss": 0.4864,
+      "step": 2608
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5400922519510655e-06,
+      "loss": 0.4673,
+      "step": 2609
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5382177914702735e-06,
+      "loss": 0.4866,
+      "step": 2610
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.536342850289018e-06,
+      "loss": 0.5065,
+      "step": 2611
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5344674287623886e-06,
+      "loss": 0.4775,
+      "step": 2612
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.532591527245561e-06,
+      "loss": 0.4747,
+      "step": 2613
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.530715146093799e-06,
+      "loss": 0.4905,
+      "step": 2614
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5288382856624675e-06,
+      "loss": 0.5099,
+      "step": 2615
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5269609463070135e-06,
+      "loss": 0.4949,
+      "step": 2616
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.525083128382974e-06,
+      "loss": 0.4908,
+      "step": 2617
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.523204832245983e-06,
+      "loss": 0.4807,
+      "step": 2618
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.52132605825176e-06,
+      "loss": 0.4765,
+      "step": 2619
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.519446806756117e-06,
+      "loss": 0.5246,
+      "step": 2620
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.517567078114957e-06,
+      "loss": 0.5095,
+      "step": 2621
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.515686872684271e-06,
+      "loss": 0.4884,
+      "step": 2622
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.513806190820143e-06,
+      "loss": 0.5037,
+      "step": 2623
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.511925032878746e-06,
+      "loss": 0.4769,
+      "step": 2624
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.510043399216342e-06,
+      "loss": 0.4866,
+      "step": 2625
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.508161290189287e-06,
+      "loss": 0.4996,
+      "step": 2626
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5062787061540195e-06,
+      "loss": 0.4734,
+      "step": 2627
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.504395647467077e-06,
+      "loss": 0.4878,
+      "step": 2628
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.502512114485081e-06,
+      "loss": 0.4969,
+      "step": 2629
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.500628107564743e-06,
+      "loss": 0.5043,
+      "step": 2630
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.49874362706287e-06,
+      "loss": 0.4761,
+      "step": 2631
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.496858673336349e-06,
+      "loss": 0.5028,
+      "step": 2632
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.494973246742168e-06,
+      "loss": 0.4783,
+      "step": 2633
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.4930873476373925e-06,
+      "loss": 0.5162,
+      "step": 2634
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.491200976379186e-06,
+      "loss": 0.5091,
+      "step": 2635
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.489314133324801e-06,
+      "loss": 0.4927,
+      "step": 2636
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.487426818831576e-06,
+      "loss": 0.4974,
+      "step": 2637
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.4855390332569375e-06,
+      "loss": 0.4968,
+      "step": 2638
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.483650776958409e-06,
+      "loss": 0.4727,
+      "step": 2639
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.481762050293593e-06,
+      "loss": 0.4778,
+      "step": 2640
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.479872853620189e-06,
+      "loss": 0.5042,
+      "step": 2641
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.477983187295982e-06,
+      "loss": 0.4743,
+      "step": 2642
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.476093051678846e-06,
+      "loss": 0.5101,
+      "step": 2643
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4742024471267445e-06,
+      "loss": 0.4687,
+      "step": 2644
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.47231137399773e-06,
+      "loss": 0.4488,
+      "step": 2645
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.470419832649944e-06,
+      "loss": 0.4709,
+      "step": 2646
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4685278234416146e-06,
+      "loss": 0.512,
+      "step": 2647
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4666353467310605e-06,
+      "loss": 0.4765,
+      "step": 2648
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.46474240287669e-06,
+      "loss": 0.4812,
+      "step": 2649
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.462848992236996e-06,
+      "loss": 0.5011,
+      "step": 2650
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.460955115170564e-06,
+      "loss": 0.4651,
+      "step": 2651
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.459060772036066e-06,
+      "loss": 0.5075,
+      "step": 2652
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.45716596319226e-06,
+      "loss": 0.4955,
+      "step": 2653
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.455270688997996e-06,
+      "loss": 0.502,
+      "step": 2654
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4533749498122124e-06,
+      "loss": 0.4834,
+      "step": 2655
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.45147874599393e-06,
+      "loss": 0.5186,
+      "step": 2656
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4495820779022645e-06,
+      "loss": 0.493,
+      "step": 2657
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.447684945896414e-06,
+      "loss": 0.4641,
+      "step": 2658
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.445787350335669e-06,
+      "loss": 0.4614,
+      "step": 2659
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.443889291579405e-06,
+      "loss": 0.5186,
+      "step": 2660
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.441990769987086e-06,
+      "loss": 0.4785,
+      "step": 2661
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4400917859182645e-06,
+      "loss": 0.4691,
+      "step": 2662
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.438192339732576e-06,
+      "loss": 0.4727,
+      "step": 2663
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.436292431789752e-06,
+      "loss": 0.479,
+      "step": 2664
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.434392062449604e-06,
+      "loss": 0.4651,
+      "step": 2665
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4324912320720325e-06,
+      "loss": 0.4831,
+      "step": 2666
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.43058994101703e-06,
+      "loss": 0.5149,
+      "step": 2667
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.428688189644668e-06,
+      "loss": 0.4841,
+      "step": 2668
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.426785978315113e-06,
+      "loss": 0.4895,
+      "step": 2669
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.424883307388615e-06,
+      "loss": 0.5055,
+      "step": 2670
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.422980177225512e-06,
+      "loss": 0.5041,
+      "step": 2671
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.421076588186225e-06,
+      "loss": 0.5195,
+      "step": 2672
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.419172540631267e-06,
+      "loss": 0.5152,
+      "step": 2673
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.417268034921238e-06,
+      "loss": 0.502,
+      "step": 2674
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.415363071416821e-06,
+      "loss": 0.4881,
+      "step": 2675
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.413457650478787e-06,
+      "loss": 0.4988,
+      "step": 2676
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.411551772467998e-06,
+      "loss": 0.488,
+      "step": 2677
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.409645437745393e-06,
+      "loss": 0.4822,
+      "step": 2678
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.407738646672007e-06,
+      "loss": 0.5142,
+      "step": 2679
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.405831399608958e-06,
+      "loss": 0.4942,
+      "step": 2680
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.403923696917448e-06,
+      "loss": 0.4852,
+      "step": 2681
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.402015538958769e-06,
+      "loss": 0.4733,
+      "step": 2682
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.400106926094297e-06,
+      "loss": 0.481,
+      "step": 2683
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.398197858685494e-06,
+      "loss": 0.4983,
+      "step": 2684
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.39628833709391e-06,
+      "loss": 0.4979,
+      "step": 2685
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3943783616811784e-06,
+      "loss": 0.5144,
+      "step": 2686
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.392467932809021e-06,
+      "loss": 0.5058,
+      "step": 2687
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.390557050839244e-06,
+      "loss": 0.4611,
+      "step": 2688
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.388645716133741e-06,
+      "loss": 0.4884,
+      "step": 2689
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3867339290544884e-06,
+      "loss": 0.5136,
+      "step": 2690
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.384821689963552e-06,
+      "loss": 0.4905,
+      "step": 2691
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.382908999223078e-06,
+      "loss": 0.4986,
+      "step": 2692
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.380995857195305e-06,
+      "loss": 0.4772,
+      "step": 2693
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3790822642425515e-06,
+      "loss": 0.4858,
+      "step": 2694
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.377168220727224e-06,
+      "loss": 0.4895,
+      "step": 2695
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.375253727011814e-06,
+      "loss": 0.489,
+      "step": 2696
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.373338783458899e-06,
+      "loss": 0.4899,
+      "step": 2697
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.371423390431139e-06,
+      "loss": 0.5233,
+      "step": 2698
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.369507548291277e-06,
+      "loss": 0.4999,
+      "step": 2699
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.367591257402151e-06,
+      "loss": 0.487,
+      "step": 2700
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.365674518126678e-06,
+      "loss": 0.4803,
+      "step": 2701
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.363757330827854e-06,
+      "loss": 0.4425,
+      "step": 2702
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.361839695868771e-06,
+      "loss": 0.4877,
+      "step": 2703
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.359921613612597e-06,
+      "loss": 0.4954,
+      "step": 2704
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.358003084422589e-06,
+      "loss": 0.5015,
+      "step": 2705
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.356084108662089e-06,
+      "loss": 0.4662,
+      "step": 2706
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.354164686694523e-06,
+      "loss": 0.5061,
+      "step": 2707
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.352244818883396e-06,
+      "loss": 0.5022,
+      "step": 2708
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.350324505592308e-06,
+      "loss": 0.4847,
+      "step": 2709
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3484037471849335e-06,
+      "loss": 0.4942,
+      "step": 2710
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3464825440250375e-06,
+      "loss": 0.503,
+      "step": 2711
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.344560896476465e-06,
+      "loss": 0.4646,
+      "step": 2712
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.34263880490315e-06,
+      "loss": 0.5337,
+      "step": 2713
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.340716269669107e-06,
+      "loss": 0.4884,
+      "step": 2714
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.338793291138434e-06,
+      "loss": 0.5001,
+      "step": 2715
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.336869869675317e-06,
+      "loss": 0.502,
+      "step": 2716
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.33494600564402e-06,
+      "loss": 0.4763,
+      "step": 2717
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.333021699408897e-06,
+      "loss": 0.5287,
+      "step": 2718
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.331096951334381e-06,
+      "loss": 0.4834,
+      "step": 2719
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.32917176178499e-06,
+      "loss": 0.4887,
+      "step": 2720
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.327246131125328e-06,
+      "loss": 0.4696,
+      "step": 2721
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.3253200597200805e-06,
+      "loss": 0.5018,
+      "step": 2722
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.323393547934014e-06,
+      "loss": 0.5052,
+      "step": 2723
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.321466596131985e-06,
+      "loss": 0.4971,
+      "step": 2724
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.319539204678925e-06,
+      "loss": 0.5297,
+      "step": 2725
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.317611373939856e-06,
+      "loss": 0.496,
+      "step": 2726
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.315683104279881e-06,
+      "loss": 0.4884,
+      "step": 2727
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.313754396064181e-06,
+      "loss": 0.4995,
+      "step": 2728
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.311825249658029e-06,
+      "loss": 0.501,
+      "step": 2729
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.309895665426775e-06,
+      "loss": 0.4746,
+      "step": 2730
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.307965643735852e-06,
+      "loss": 0.4719,
+      "step": 2731
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.306035184950779e-06,
+      "loss": 0.4648,
+      "step": 2732
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.304104289437157e-06,
+      "loss": 0.5107,
+      "step": 2733
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.302172957560666e-06,
+      "loss": 0.4756,
+      "step": 2734
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.300241189687072e-06,
+      "loss": 0.4934,
+      "step": 2735
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.298308986182224e-06,
+      "loss": 0.5016,
+      "step": 2736
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.296376347412053e-06,
+      "loss": 0.508,
+      "step": 2737
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.294443273742569e-06,
+      "loss": 0.4861,
+      "step": 2738
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.292509765539869e-06,
+      "loss": 0.488,
+      "step": 2739
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.290575823170132e-06,
+      "loss": 0.4951,
+      "step": 2740
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.288641446999614e-06,
+      "loss": 0.5212,
+      "step": 2741
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.28670663739466e-06,
+      "loss": 0.4838,
+      "step": 2742
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.284771394721694e-06,
+      "loss": 0.5004,
+      "step": 2743
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.282835719347219e-06,
+      "loss": 0.5182,
+      "step": 2744
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.280899611637827e-06,
+      "loss": 0.4897,
+      "step": 2745
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.278963071960185e-06,
+      "loss": 0.4649,
+      "step": 2746
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.277026100681046e-06,
+      "loss": 0.4961,
+      "step": 2747
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.275088698167242e-06,
+      "loss": 0.483,
+      "step": 2748
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.27315086478569e-06,
+      "loss": 0.4962,
+      "step": 2749
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.271212600903384e-06,
+      "loss": 0.4999,
+      "step": 2750
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.269273906887405e-06,
+      "loss": 0.5188,
+      "step": 2751
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.267334783104911e-06,
+      "loss": 0.4852,
+      "step": 2752
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.265395229923143e-06,
+      "loss": 0.5111,
+      "step": 2753
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.263455247709423e-06,
+      "loss": 0.4536,
+      "step": 2754
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.261514836831157e-06,
+      "loss": 0.5172,
+      "step": 2755
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2595739976558275e-06,
+      "loss": 0.5147,
+      "step": 2756
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2576327305510006e-06,
+      "loss": 0.4717,
+      "step": 2757
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.255691035884323e-06,
+      "loss": 0.4817,
+      "step": 2758
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.253748914023524e-06,
+      "loss": 0.4837,
+      "step": 2759
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.251806365336411e-06,
+      "loss": 0.4876,
+      "step": 2760
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.249863390190876e-06,
+      "loss": 0.4903,
+      "step": 2761
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.247919988954885e-06,
+      "loss": 0.4803,
+      "step": 2762
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.245976161996495e-06,
+      "loss": 0.5131,
+      "step": 2763
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.244031909683832e-06,
+      "loss": 0.4805,
+      "step": 2764
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.242087232385111e-06,
+      "loss": 0.4782,
+      "step": 2765
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.240142130468626e-06,
+      "loss": 0.472,
+      "step": 2766
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.238196604302748e-06,
+      "loss": 0.4651,
+      "step": 2767
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.23625065425593e-06,
+      "loss": 0.4973,
+      "step": 2768
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.234304280696709e-06,
+      "loss": 0.4852,
+      "step": 2769
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.232357483993696e-06,
+      "loss": 0.503,
+      "step": 2770
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.230410264515586e-06,
+      "loss": 0.5052,
+      "step": 2771
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.228462622631155e-06,
+      "loss": 0.4858,
+      "step": 2772
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.226514558709254e-06,
+      "loss": 0.4677,
+      "step": 2773
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.224566073118821e-06,
+      "loss": 0.5251,
+      "step": 2774
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2226171662288665e-06,
+      "loss": 0.5114,
+      "step": 2775
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.220667838408487e-06,
+      "loss": 0.4883,
+      "step": 2776
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.218718090026853e-06,
+      "loss": 0.4812,
+      "step": 2777
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.21676792145322e-06,
+      "loss": 0.4832,
+      "step": 2778
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.214817333056921e-06,
+      "loss": 0.4959,
+      "step": 2779
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.212866325207366e-06,
+      "loss": 0.5027,
+      "step": 2780
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2109148982740485e-06,
+      "loss": 0.4999,
+      "step": 2781
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.208963052626539e-06,
+      "loss": 0.4756,
+      "step": 2782
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.207010788634489e-06,
+      "loss": 0.5106,
+      "step": 2783
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.205058106667626e-06,
+      "loss": 0.4892,
+      "step": 2784
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.203105007095761e-06,
+      "loss": 0.4813,
+      "step": 2785
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2011514902887796e-06,
+      "loss": 0.4796,
+      "step": 2786
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.199197556616648e-06,
+      "loss": 0.5329,
+      "step": 2787
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.197243206449417e-06,
+      "loss": 0.511,
+      "step": 2788
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.195288440157207e-06,
+      "loss": 0.5312,
+      "step": 2789
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.19333325811022e-06,
+      "loss": 0.4874,
+      "step": 2790
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.191377660678743e-06,
+      "loss": 0.4701,
+      "step": 2791
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.189421648233134e-06,
+      "loss": 0.4751,
+      "step": 2792
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.187465221143832e-06,
+      "loss": 0.4991,
+      "step": 2793
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.185508379781358e-06,
+      "loss": 0.5108,
+      "step": 2794
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.183551124516305e-06,
+      "loss": 0.4957,
+      "step": 2795
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.181593455719349e-06,
+      "loss": 0.476,
+      "step": 2796
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.179635373761245e-06,
+      "loss": 0.4974,
+      "step": 2797
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.177676879012821e-06,
+      "loss": 0.4891,
+      "step": 2798
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.17571797184499e-06,
+      "loss": 0.4915,
+      "step": 2799
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.173758652628738e-06,
+      "loss": 0.4636,
+      "step": 2800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.171798921735132e-06,
+      "loss": 0.4843,
+      "step": 2801
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.169838779535313e-06,
+      "loss": 0.4407,
+      "step": 2802
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.167878226400504e-06,
+      "loss": 0.499,
+      "step": 2803
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.165917262702006e-06,
+      "loss": 0.4867,
+      "step": 2804
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.163955888811194e-06,
+      "loss": 0.4976,
+      "step": 2805
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.161994105099523e-06,
+      "loss": 0.5079,
+      "step": 2806
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.160031911938527e-06,
+      "loss": 0.4808,
+      "step": 2807
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.158069309699816e-06,
+      "loss": 0.504,
+      "step": 2808
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1561062987550746e-06,
+      "loss": 0.5192,
+      "step": 2809
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1541428794760715e-06,
+      "loss": 0.5025,
+      "step": 2810
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1521790522346465e-06,
+      "loss": 0.4826,
+      "step": 2811
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.15021481740272e-06,
+      "loss": 0.529,
+      "step": 2812
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1482501753522895e-06,
+      "loss": 0.5351,
+      "step": 2813
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.146285126455425e-06,
+      "loss": 0.4597,
+      "step": 2814
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.144319671084283e-06,
+      "loss": 0.4901,
+      "step": 2815
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.142353809611087e-06,
+      "loss": 0.4917,
+      "step": 2816
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.140387542408142e-06,
+      "loss": 0.4978,
+      "step": 2817
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.138420869847833e-06,
+      "loss": 0.5073,
+      "step": 2818
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1364537923026155e-06,
+      "loss": 0.4724,
+      "step": 2819
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1344863101450245e-06,
+      "loss": 0.4824,
+      "step": 2820
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.132518423747673e-06,
+      "loss": 0.472,
+      "step": 2821
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.130550133483249e-06,
+      "loss": 0.4876,
+      "step": 2822
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1285814397245155e-06,
+      "loss": 0.4933,
+      "step": 2823
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.126612342844315e-06,
+      "loss": 0.4896,
+      "step": 2824
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.124642843215566e-06,
+      "loss": 0.4918,
+      "step": 2825
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.12267294121126e-06,
+      "loss": 0.4902,
+      "step": 2826
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.120702637204468e-06,
+      "loss": 0.4578,
+      "step": 2827
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.118731931568339e-06,
+      "loss": 0.4919,
+      "step": 2828
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.116760824676088e-06,
+      "loss": 0.4843,
+      "step": 2829
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.114789316901021e-06,
+      "loss": 0.4839,
+      "step": 2830
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.112817408616508e-06,
+      "loss": 0.4739,
+      "step": 2831
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.110845100195997e-06,
+      "loss": 0.5041,
+      "step": 2832
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1088723920130185e-06,
+      "loss": 0.4812,
+      "step": 2833
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.106899284441171e-06,
+      "loss": 0.4825,
+      "step": 2834
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.104925777854131e-06,
+      "loss": 0.5132,
+      "step": 2835
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1029518726256544e-06,
+      "loss": 0.4836,
+      "step": 2836
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.100977569129566e-06,
+      "loss": 0.4917,
+      "step": 2837
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.09900286773977e-06,
+      "loss": 0.5015,
+      "step": 2838
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.097027768830247e-06,
+      "loss": 0.4835,
+      "step": 2839
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.095052272775049e-06,
+      "loss": 0.4792,
+      "step": 2840
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.093076379948306e-06,
+      "loss": 0.4865,
+      "step": 2841
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.091100090724223e-06,
+      "loss": 0.5083,
+      "step": 2842
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.089123405477081e-06,
+      "loss": 0.497,
+      "step": 2843
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.087146324581231e-06,
+      "loss": 0.4931,
+      "step": 2844
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.085168848411105e-06,
+      "loss": 0.5127,
+      "step": 2845
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.083190977341208e-06,
+      "loss": 0.4947,
+      "step": 2846
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.081212711746117e-06,
+      "loss": 0.4922,
+      "step": 2847
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.079234052000487e-06,
+      "loss": 0.5046,
+      "step": 2848
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.077254998479048e-06,
+      "loss": 0.467,
+      "step": 2849
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.0752755515566e-06,
+      "loss": 0.5009,
+      "step": 2850
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.073295711608025e-06,
+      "loss": 0.5038,
+      "step": 2851
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.071315479008269e-06,
+      "loss": 0.5108,
+      "step": 2852
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.069334854132364e-06,
+      "loss": 0.4807,
+      "step": 2853
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.067353837355409e-06,
+      "loss": 0.4822,
+      "step": 2854
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.065372429052576e-06,
+      "loss": 0.4915,
+      "step": 2855
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.063390629599117e-06,
+      "loss": 0.5117,
+      "step": 2856
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.061408439370356e-06,
+      "loss": 0.4985,
+      "step": 2857
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.059425858741688e-06,
+      "loss": 0.4951,
+      "step": 2858
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.057442888088586e-06,
+      "loss": 0.4861,
+      "step": 2859
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.055459527786594e-06,
+      "loss": 0.4969,
+      "step": 2860
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.053475778211328e-06,
+      "loss": 0.4873,
+      "step": 2861
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.051491639738485e-06,
+      "loss": 0.4831,
+      "step": 2862
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.04950711274383e-06,
+      "loss": 0.4568,
+      "step": 2863
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.047522197603202e-06,
+      "loss": 0.5056,
+      "step": 2864
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.045536894692512e-06,
+      "loss": 0.5284,
+      "step": 2865
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.043551204387751e-06,
+      "loss": 0.4928,
+      "step": 2866
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.041565127064976e-06,
+      "loss": 0.4967,
+      "step": 2867
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.03957866310032e-06,
+      "loss": 0.489,
+      "step": 2868
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.037591812869992e-06,
+      "loss": 0.5158,
+      "step": 2869
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.035604576750271e-06,
+      "loss": 0.493,
+      "step": 2870
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.0336169551175085e-06,
+      "loss": 0.5008,
+      "step": 2871
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.031628948348131e-06,
+      "loss": 0.5144,
+      "step": 2872
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.029640556818638e-06,
+      "loss": 0.4986,
+      "step": 2873
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.027651780905598e-06,
+      "loss": 0.4659,
+      "step": 2874
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.025662620985661e-06,
+      "loss": 0.4897,
+      "step": 2875
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.023673077435539e-06,
+      "loss": 0.4995,
+      "step": 2876
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.021683150632024e-06,
+      "loss": 0.52,
+      "step": 2877
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.019692840951977e-06,
+      "loss": 0.4792,
+      "step": 2878
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.017702148772335e-06,
+      "loss": 0.4884,
+      "step": 2879
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.015711074470104e-06,
+      "loss": 0.5156,
+      "step": 2880
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.013719618422363e-06,
+      "loss": 0.4977,
+      "step": 2881
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.0117277810062655e-06,
+      "loss": 0.4602,
+      "step": 2882
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.009735562599035e-06,
+      "loss": 0.4689,
+      "step": 2883
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.007742963577967e-06,
+      "loss": 0.4908,
+      "step": 2884
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.005749984320432e-06,
+      "loss": 0.461,
+      "step": 2885
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.003756625203868e-06,
+      "loss": 0.4829,
+      "step": 2886
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.001762886605789e-06,
+      "loss": 0.5019,
+      "step": 2887
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9997687689037804e-06,
+      "loss": 0.5075,
+      "step": 2888
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.997774272475495e-06,
+      "loss": 0.4887,
+      "step": 2889
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.995779397698663e-06,
+      "loss": 0.4838,
+      "step": 2890
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.993784144951082e-06,
+      "loss": 0.4623,
+      "step": 2891
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9917885146106245e-06,
+      "loss": 0.502,
+      "step": 2892
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.989792507055233e-06,
+      "loss": 0.471,
+      "step": 2893
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9877961226629205e-06,
+      "loss": 0.4544,
+      "step": 2894
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.985799361811773e-06,
+      "loss": 0.4584,
+      "step": 2895
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.983802224879949e-06,
+      "loss": 0.4942,
+      "step": 2896
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9818047122456725e-06,
+      "loss": 0.5004,
+      "step": 2897
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.979806824287243e-06,
+      "loss": 0.4984,
+      "step": 2898
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.977808561383035e-06,
+      "loss": 0.4789,
+      "step": 2899
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.975809923911484e-06,
+      "loss": 0.5126,
+      "step": 2900
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.973810912251105e-06,
+      "loss": 0.4683,
+      "step": 2901
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.971811526780482e-06,
+      "loss": 0.4821,
+      "step": 2902
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.969811767878266e-06,
+      "loss": 0.5013,
+      "step": 2903
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.967811635923183e-06,
+      "loss": 0.4783,
+      "step": 2904
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.965811131294027e-06,
+      "loss": 0.4885,
+      "step": 2905
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.963810254369665e-06,
+      "loss": 0.4922,
+      "step": 2906
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.961809005529031e-06,
+      "loss": 0.4876,
+      "step": 2907
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9598073851511325e-06,
+      "loss": 0.4928,
+      "step": 2908
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.957805393615049e-06,
+      "loss": 0.4932,
+      "step": 2909
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.955803031299925e-06,
+      "loss": 0.4792,
+      "step": 2910
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.953800298584977e-06,
+      "loss": 0.4782,
+      "step": 2911
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.951797195849497e-06,
+      "loss": 0.4389,
+      "step": 2912
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.949793723472838e-06,
+      "loss": 0.4621,
+      "step": 2913
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.947789881834428e-06,
+      "loss": 0.4898,
+      "step": 2914
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.94578567131377e-06,
+      "loss": 0.4842,
+      "step": 2915
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.943781092290425e-06,
+      "loss": 0.4762,
+      "step": 2916
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.941776145144032e-06,
+      "loss": 0.5151,
+      "step": 2917
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.939770830254301e-06,
+      "loss": 0.478,
+      "step": 2918
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9377651480010055e-06,
+      "loss": 0.4937,
+      "step": 2919
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.935759098763993e-06,
+      "loss": 0.476,
+      "step": 2920
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.933752682923179e-06,
+      "loss": 0.497,
+      "step": 2921
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9317459008585485e-06,
+      "loss": 0.4926,
+      "step": 2922
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.929738752950155e-06,
+      "loss": 0.4793,
+      "step": 2923
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.927731239578125e-06,
+      "loss": 0.5098,
+      "step": 2924
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9257233611226495e-06,
+      "loss": 0.4777,
+      "step": 2925
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9237151179639895e-06,
+      "loss": 0.4729,
+      "step": 2926
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.921706510482479e-06,
+      "loss": 0.4882,
+      "step": 2927
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.919697539058516e-06,
+      "loss": 0.4744,
+      "step": 2928
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.917688204072571e-06,
+      "loss": 0.4767,
+      "step": 2929
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.915678505905181e-06,
+      "loss": 0.4373,
+      "step": 2930
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.913668444936954e-06,
+      "loss": 0.4779,
+      "step": 2931
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.911658021548564e-06,
+      "loss": 0.4602,
+      "step": 2932
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.909647236120754e-06,
+      "loss": 0.4776,
+      "step": 2933
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.907636089034342e-06,
+      "loss": 0.4958,
+      "step": 2934
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.905624580670202e-06,
+      "loss": 0.5092,
+      "step": 2935
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.903612711409289e-06,
+      "loss": 0.5016,
+      "step": 2936
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.901600481632619e-06,
+      "loss": 0.4854,
+      "step": 2937
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.8995878917212776e-06,
+      "loss": 0.4512,
+      "step": 2938
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.89757494205642e-06,
+      "loss": 0.4819,
+      "step": 2939
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.895561633019269e-06,
+      "loss": 0.494,
+      "step": 2940
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.893547964991114e-06,
+      "loss": 0.4554,
+      "step": 2941
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.891533938353316e-06,
+      "loss": 0.5121,
+      "step": 2942
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.889519553487298e-06,
+      "loss": 0.5154,
+      "step": 2943
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8875048107745566e-06,
+      "loss": 0.4896,
+      "step": 2944
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.885489710596655e-06,
+      "loss": 0.4961,
+      "step": 2945
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.883474253335221e-06,
+      "loss": 0.4905,
+      "step": 2946
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.881458439371951e-06,
+      "loss": 0.4692,
+      "step": 2947
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.879442269088613e-06,
+      "loss": 0.4837,
+      "step": 2948
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.877425742867037e-06,
+      "loss": 0.4637,
+      "step": 2949
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.875408861089125e-06,
+      "loss": 0.4838,
+      "step": 2950
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8733916241368435e-06,
+      "loss": 0.5119,
+      "step": 2951
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.871374032392226e-06,
+      "loss": 0.4566,
+      "step": 2952
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.869356086237374e-06,
+      "loss": 0.5044,
+      "step": 2953
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.867337786054459e-06,
+      "loss": 0.5089,
+      "step": 2954
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.865319132225714e-06,
+      "loss": 0.4864,
+      "step": 2955
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.863300125133442e-06,
+      "loss": 0.5041,
+      "step": 2956
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.861280765160016e-06,
+      "loss": 0.5185,
+      "step": 2957
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.859261052687869e-06,
+      "loss": 0.4628,
+      "step": 2958
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.857240988099503e-06,
+      "loss": 0.4816,
+      "step": 2959
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.855220571777493e-06,
+      "loss": 0.4939,
+      "step": 2960
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8531998041044735e-06,
+      "loss": 0.4713,
+      "step": 2961
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.851178685463144e-06,
+      "loss": 0.4896,
+      "step": 2962
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.849157216236279e-06,
+      "loss": 0.4922,
+      "step": 2963
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.847135396806714e-06,
+      "loss": 0.4665,
+      "step": 2964
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.845113227557348e-06,
+      "loss": 0.501,
+      "step": 2965
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.843090708871154e-06,
+      "loss": 0.5065,
+      "step": 2966
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.841067841131165e-06,
+      "loss": 0.4598,
+      "step": 2967
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.839044624720479e-06,
+      "loss": 0.516,
+      "step": 2968
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8370210600222665e-06,
+      "loss": 0.5007,
+      "step": 2969
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.83499714741976e-06,
+      "loss": 0.479,
+      "step": 2970
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.832972887296255e-06,
+      "loss": 0.4969,
+      "step": 2971
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.830948280035119e-06,
+      "loss": 0.4771,
+      "step": 2972
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.828923326019784e-06,
+      "loss": 0.468,
+      "step": 2973
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.826898025633741e-06,
+      "loss": 0.4729,
+      "step": 2974
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.824872379260556e-06,
+      "loss": 0.4666,
+      "step": 2975
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.822846387283856e-06,
+      "loss": 0.5228,
+      "step": 2976
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.820820050087329e-06,
+      "loss": 0.4788,
+      "step": 2977
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.818793368054737e-06,
+      "loss": 0.446,
+      "step": 2978
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8167663415699035e-06,
+      "loss": 0.502,
+      "step": 2979
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.814738971016715e-06,
+      "loss": 0.496,
+      "step": 2980
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.812711256779126e-06,
+      "loss": 0.4958,
+      "step": 2981
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.810683199241154e-06,
+      "loss": 0.5046,
+      "step": 2982
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.808654798786884e-06,
+      "loss": 0.4862,
+      "step": 2983
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.806626055800465e-06,
+      "loss": 0.4758,
+      "step": 2984
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.80459697066611e-06,
+      "loss": 0.4739,
+      "step": 2985
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8025675437680975e-06,
+      "loss": 0.5164,
+      "step": 2986
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.80053777549077e-06,
+      "loss": 0.498,
+      "step": 2987
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.798507666218536e-06,
+      "loss": 0.5124,
+      "step": 2988
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.796477216335868e-06,
+      "loss": 0.4874,
+      "step": 2989
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.7944464262273015e-06,
+      "loss": 0.5182,
+      "step": 2990
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.792415296277439e-06,
+      "loss": 0.452,
+      "step": 2991
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.790383826870945e-06,
+      "loss": 0.499,
+      "step": 2992
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.788352018392552e-06,
+      "loss": 0.4905,
+      "step": 2993
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.786319871227052e-06,
+      "loss": 0.4916,
+      "step": 2994
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.784287385759301e-06,
+      "loss": 0.4872,
+      "step": 2995
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.782254562374226e-06,
+      "loss": 0.474,
+      "step": 2996
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.780221401456812e-06,
+      "loss": 0.5107,
+      "step": 2997
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.778187903392106e-06,
+      "loss": 0.507,
+      "step": 2998
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.776154068565226e-06,
+      "loss": 0.4846,
+      "step": 2999
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.774119897361349e-06,
+      "loss": 0.466,
+      "step": 3000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.772085390165712e-06,
+      "loss": 0.5184,
+      "step": 3001
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.770050547363629e-06,
+      "loss": 0.4716,
+      "step": 3002
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.768015369340462e-06,
+      "loss": 0.4943,
+      "step": 3003
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.765979856481644e-06,
+      "loss": 0.4772,
+      "step": 3004
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.763944009172674e-06,
+      "loss": 0.4942,
+      "step": 3005
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.761907827799108e-06,
+      "loss": 0.4667,
+      "step": 3006
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.7598713127465685e-06,
+      "loss": 0.5037,
+      "step": 3007
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.757834464400742e-06,
+      "loss": 0.4964,
+      "step": 3008
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.755797283147378e-06,
+      "loss": 0.512,
+      "step": 3009
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.753759769372286e-06,
+      "loss": 0.486,
+      "step": 3010
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.751721923461341e-06,
+      "loss": 0.4623,
+      "step": 3011
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.7496837458004814e-06,
+      "loss": 0.5156,
+      "step": 3012
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.747645236775706e-06,
+      "loss": 0.4835,
+      "step": 3013
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.74560639677308e-06,
+      "loss": 0.4897,
+      "step": 3014
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.743567226178729e-06,
+      "loss": 0.4679,
+      "step": 3015
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.74152772537884e-06,
+      "loss": 0.5021,
+      "step": 3016
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.739487894759664e-06,
+      "loss": 0.4151,
+      "step": 3017
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.737447734707516e-06,
+      "loss": 0.5246,
+      "step": 3018
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.735407245608769e-06,
+      "loss": 0.4773,
+      "step": 3019
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.7333664278498655e-06,
+      "loss": 0.4835,
+      "step": 3020
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.7313252818173e-06,
+      "loss": 0.5012,
+      "step": 3021
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.729283807897639e-06,
+      "loss": 0.4597,
+      "step": 3022
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.727242006477509e-06,
+      "loss": 0.4707,
+      "step": 3023
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.725199877943591e-06,
+      "loss": 0.5002,
+      "step": 3024
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.723157422682637e-06,
+      "loss": 0.4992,
+      "step": 3025
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.721114641081459e-06,
+      "loss": 0.493,
+      "step": 3026
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.719071533526924e-06,
+      "loss": 0.5114,
+      "step": 3027
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.71702810040597e-06,
+      "loss": 0.4794,
+      "step": 3028
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.714984342105594e-06,
+      "loss": 0.4676,
+      "step": 3029
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.712940259012849e-06,
+      "loss": 0.5108,
+      "step": 3030
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.710895851514856e-06,
+      "loss": 0.5052,
+      "step": 3031
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.708851119998796e-06,
+      "loss": 0.5079,
+      "step": 3032
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.706806064851909e-06,
+      "loss": 0.5197,
+      "step": 3033
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.704760686461498e-06,
+      "loss": 0.4881,
+      "step": 3034
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.702714985214927e-06,
+      "loss": 0.4974,
+      "step": 3035
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.700668961499624e-06,
+      "loss": 0.5164,
+      "step": 3036
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.69862261570307e-06,
+      "loss": 0.4573,
+      "step": 3037
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.696575948212815e-06,
+      "loss": 0.511,
+      "step": 3038
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.694528959416471e-06,
+      "loss": 0.4562,
+      "step": 3039
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6924816497016996e-06,
+      "loss": 0.4578,
+      "step": 3040
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.690434019456236e-06,
+      "loss": 0.5037,
+      "step": 3041
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.688386069067869e-06,
+      "loss": 0.4899,
+      "step": 3042
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.686337798924451e-06,
+      "loss": 0.5142,
+      "step": 3043
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.68428920941389e-06,
+      "loss": 0.4931,
+      "step": 3044
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.682240300924163e-06,
+      "loss": 0.506,
+      "step": 3045
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.680191073843299e-06,
+      "loss": 0.5025,
+      "step": 3046
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.678141528559393e-06,
+      "loss": 0.4859,
+      "step": 3047
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.676091665460599e-06,
+      "loss": 0.4565,
+      "step": 3048
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.674041484935128e-06,
+      "loss": 0.48,
+      "step": 3049
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.671990987371253e-06,
+      "loss": 0.4838,
+      "step": 3050
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6699401731573095e-06,
+      "loss": 0.4795,
+      "step": 3051
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.667889042681691e-06,
+      "loss": 0.4858,
+      "step": 3052
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.66583759633285e-06,
+      "loss": 0.507,
+      "step": 3053
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.663785834499301e-06,
+      "loss": 0.4822,
+      "step": 3054
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.661733757569616e-06,
+      "loss": 0.4867,
+      "step": 3055
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.659681365932427e-06,
+      "loss": 0.4971,
+      "step": 3056
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6576286599764286e-06,
+      "loss": 0.4908,
+      "step": 3057
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.655575640090371e-06,
+      "loss": 0.4825,
+      "step": 3058
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6535223066630635e-06,
+      "loss": 0.5121,
+      "step": 3059
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.65146866008338e-06,
+      "loss": 0.4834,
+      "step": 3060
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.649414700740249e-06,
+      "loss": 0.4612,
+      "step": 3061
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.647360429022659e-06,
+      "loss": 0.506,
+      "step": 3062
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.64530584531966e-06,
+      "loss": 0.5147,
+      "step": 3063
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.643250950020358e-06,
+      "loss": 0.4941,
+      "step": 3064
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.64119574351392e-06,
+      "loss": 0.5053,
+      "step": 3065
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.639140226189571e-06,
+      "loss": 0.5172,
+      "step": 3066
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.637084398436597e-06,
+      "loss": 0.5051,
+      "step": 3067
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.635028260644338e-06,
+      "loss": 0.4999,
+      "step": 3068
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.632971813202197e-06,
+      "loss": 0.4775,
+      "step": 3069
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.630915056499636e-06,
+      "loss": 0.4793,
+      "step": 3070
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.628857990926172e-06,
+      "loss": 0.4878,
+      "step": 3071
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.626800616871383e-06,
+      "loss": 0.4639,
+      "step": 3072
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.624742934724904e-06,
+      "loss": 0.4873,
+      "step": 3073
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.62268494487643e-06,
+      "loss": 0.4469,
+      "step": 3074
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6206266477157144e-06,
+      "loss": 0.4631,
+      "step": 3075
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6185680436325674e-06,
+      "loss": 0.4776,
+      "step": 3076
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.616509133016855e-06,
+      "loss": 0.4865,
+      "step": 3077
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.614449916258509e-06,
+      "loss": 0.4769,
+      "step": 3078
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.61239039374751e-06,
+      "loss": 0.4792,
+      "step": 3079
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.610330565873903e-06,
+      "loss": 0.4587,
+      "step": 3080
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.608270433027789e-06,
+      "loss": 0.4723,
+      "step": 3081
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.606209995599324e-06,
+      "loss": 0.5062,
+      "step": 3082
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.604149253978724e-06,
+      "loss": 0.4894,
+      "step": 3083
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.602088208556268e-06,
+      "loss": 0.4844,
+      "step": 3084
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.600026859722281e-06,
+      "loss": 0.4636,
+      "step": 3085
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.5979652078671515e-06,
+      "loss": 0.5245,
+      "step": 3086
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.5959032533813305e-06,
+      "loss": 0.4347,
+      "step": 3087
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.593840996655318e-06,
+      "loss": 0.47,
+      "step": 3088
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.591778438079674e-06,
+      "loss": 0.4829,
+      "step": 3089
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.589715578045018e-06,
+      "loss": 0.4741,
+      "step": 3090
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.587652416942025e-06,
+      "loss": 0.5028,
+      "step": 3091
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.585588955161424e-06,
+      "loss": 0.4743,
+      "step": 3092
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.583525193094007e-06,
+      "loss": 0.4925,
+      "step": 3093
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.581461131130619e-06,
+      "loss": 0.4982,
+      "step": 3094
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.579396769662161e-06,
+      "loss": 0.4917,
+      "step": 3095
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.577332109079593e-06,
+      "loss": 0.4679,
+      "step": 3096
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.575267149773932e-06,
+      "loss": 0.4952,
+      "step": 3097
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5732018921362475e-06,
+      "loss": 0.513,
+      "step": 3098
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5711363365576706e-06,
+      "loss": 0.4955,
+      "step": 3099
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.569070483429387e-06,
+      "loss": 0.4864,
+      "step": 3100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.567004333142636e-06,
+      "loss": 0.4744,
+      "step": 3101
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.564937886088718e-06,
+      "loss": 0.4755,
+      "step": 3102
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5628711426589866e-06,
+      "loss": 0.4868,
+      "step": 3103
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.56080410324485e-06,
+      "loss": 0.4929,
+      "step": 3104
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5587367682377765e-06,
+      "loss": 0.4669,
+      "step": 3105
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5566691380292895e-06,
+      "loss": 0.4418,
+      "step": 3106
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.554601213010965e-06,
+      "loss": 0.5054,
+      "step": 3107
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5525329935744375e-06,
+      "loss": 0.4618,
+      "step": 3108
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.550464480111397e-06,
+      "loss": 0.4805,
+      "step": 3109
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.54839567301359e-06,
+      "loss": 0.4711,
+      "step": 3110
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.546326572672817e-06,
+      "loss": 0.471,
+      "step": 3111
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5442571794809326e-06,
+      "loss": 0.4626,
+      "step": 3112
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.542187493829851e-06,
+      "loss": 0.4919,
+      "step": 3113
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.540117516111541e-06,
+      "loss": 0.4944,
+      "step": 3114
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.538047246718023e-06,
+      "loss": 0.4624,
+      "step": 3115
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.535976686041374e-06,
+      "loss": 0.4976,
+      "step": 3116
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.533905834473732e-06,
+      "loss": 0.4909,
+      "step": 3117
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.53183469240728e-06,
+      "loss": 0.501,
+      "step": 3118
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.529763260234265e-06,
+      "loss": 0.4893,
+      "step": 3119
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.527691538346983e-06,
+      "loss": 0.5017,
+      "step": 3120
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.525619527137789e-06,
+      "loss": 0.4784,
+      "step": 3121
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5235472269990875e-06,
+      "loss": 0.4824,
+      "step": 3122
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.521474638323346e-06,
+      "loss": 0.5263,
+      "step": 3123
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.519401761503079e-06,
+      "loss": 0.47,
+      "step": 3124
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.517328596930856e-06,
+      "loss": 0.4793,
+      "step": 3125
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.515255144999309e-06,
+      "loss": 0.4795,
+      "step": 3126
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.513181406101116e-06,
+      "loss": 0.5189,
+      "step": 3127
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5111073806290095e-06,
+      "loss": 0.4716,
+      "step": 3128
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.509033068975782e-06,
+      "loss": 0.4812,
+      "step": 3129
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.506958471534277e-06,
+      "loss": 0.4895,
+      "step": 3130
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5048835886973905e-06,
+      "loss": 0.5027,
+      "step": 3131
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.502808420858075e-06,
+      "loss": 0.4657,
+      "step": 3132
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.500732968409339e-06,
+      "loss": 0.5217,
+      "step": 3133
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.4986572317442376e-06,
+      "loss": 0.4911,
+      "step": 3134
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.4965812112558864e-06,
+      "loss": 0.5014,
+      "step": 3135
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.494504907337453e-06,
+      "loss": 0.5061,
+      "step": 3136
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.4924283203821585e-06,
+      "loss": 0.4989,
+      "step": 3137
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.490351450783275e-06,
+      "loss": 0.4771,
+      "step": 3138
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.488274298934134e-06,
+      "loss": 0.4887,
+      "step": 3139
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.486196865228113e-06,
+      "loss": 0.4663,
+      "step": 3140
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.48411915005865e-06,
+      "loss": 0.5136,
+      "step": 3141
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.482041153819233e-06,
+      "loss": 0.444,
+      "step": 3142
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.4799628769034015e-06,
+      "loss": 0.4843,
+      "step": 3143
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.4778843197047516e-06,
+      "loss": 0.4724,
+      "step": 3144
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.475805482616931e-06,
+      "loss": 0.5154,
+      "step": 3145
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.473726366033638e-06,
+      "loss": 0.4815,
+      "step": 3146
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.471646970348629e-06,
+      "loss": 0.4993,
+      "step": 3147
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.46956729595571e-06,
+      "loss": 0.4881,
+      "step": 3148
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.467487343248739e-06,
+      "loss": 0.4647,
+      "step": 3149
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.4654071126216285e-06,
+      "loss": 0.4725,
+      "step": 3150
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.463326604468343e-06,
+      "loss": 0.4856,
+      "step": 3151
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.4612458191828985e-06,
+      "loss": 0.4903,
+      "step": 3152
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.4591647571593674e-06,
+      "loss": 0.4583,
+      "step": 3153
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.457083418791869e-06,
+      "loss": 0.5329,
+      "step": 3154
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.455001804474578e-06,
+      "loss": 0.4514,
+      "step": 3155
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.452919914601724e-06,
+      "loss": 0.4963,
+      "step": 3156
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.450837749567581e-06,
+      "loss": 0.4863,
+      "step": 3157
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.448755309766483e-06,
+      "loss": 0.4913,
+      "step": 3158
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.446672595592813e-06,
+      "loss": 0.4922,
+      "step": 3159
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.444589607441006e-06,
+      "loss": 0.4972,
+      "step": 3160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.442506345705546e-06,
+      "loss": 0.508,
+      "step": 3161
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.440422810780974e-06,
+      "loss": 0.5163,
+      "step": 3162
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.438339003061881e-06,
+      "loss": 0.5245,
+      "step": 3163
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.436254922942906e-06,
+      "loss": 0.4961,
+      "step": 3164
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.434170570818745e-06,
+      "loss": 0.4891,
+      "step": 3165
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.432085947084143e-06,
+      "loss": 0.4411,
+      "step": 3166
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.430001052133896e-06,
+      "loss": 0.5122,
+      "step": 3167
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.427915886362852e-06,
+      "loss": 0.4856,
+      "step": 3168
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.4258304501659105e-06,
+      "loss": 0.5211,
+      "step": 3169
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.423744743938021e-06,
+      "loss": 0.4815,
+      "step": 3170
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.421658768074185e-06,
+      "loss": 0.4746,
+      "step": 3171
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.4195725229694576e-06,
+      "loss": 0.5001,
+      "step": 3172
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.417486009018938e-06,
+      "loss": 0.4894,
+      "step": 3173
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.415399226617784e-06,
+      "loss": 0.5012,
+      "step": 3174
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.413312176161202e-06,
+      "loss": 0.4837,
+      "step": 3175
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.4112248580444445e-06,
+      "loss": 0.4737,
+      "step": 3176
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.40913727266282e-06,
+      "loss": 0.4974,
+      "step": 3177
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.407049420411688e-06,
+      "loss": 0.4926,
+      "step": 3178
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.404961301686452e-06,
+      "loss": 0.4635,
+      "step": 3179
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.402872916882573e-06,
+      "loss": 0.4994,
+      "step": 3180
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.400784266395562e-06,
+      "loss": 0.5061,
+      "step": 3181
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.398695350620974e-06,
+      "loss": 0.5024,
+      "step": 3182
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.39660616995442e-06,
+      "loss": 0.4897,
+      "step": 3183
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.394516724791562e-06,
+      "loss": 0.5042,
+      "step": 3184
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.392427015528107e-06,
+      "loss": 0.4933,
+      "step": 3185
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.390337042559814e-06,
+      "loss": 0.4757,
+      "step": 3186
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.388246806282495e-06,
+      "loss": 0.4391,
+      "step": 3187
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.386156307092007e-06,
+      "loss": 0.4707,
+      "step": 3188
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.38406554538426e-06,
+      "loss": 0.4826,
+      "step": 3189
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.381974521555214e-06,
+      "loss": 0.4518,
+      "step": 3190
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.379883236000879e-06,
+      "loss": 0.4701,
+      "step": 3191
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.377791689117309e-06,
+      "loss": 0.5167,
+      "step": 3192
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.375699881300614e-06,
+      "loss": 0.4791,
+      "step": 3193
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.373607812946952e-06,
+      "loss": 0.5212,
+      "step": 3194
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.371515484452527e-06,
+      "loss": 0.4962,
+      "step": 3195
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3694228962135976e-06,
+      "loss": 0.4651,
+      "step": 3196
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.367330048626467e-06,
+      "loss": 0.4691,
+      "step": 3197
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.365236942087489e-06,
+      "loss": 0.486,
+      "step": 3198
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.363143576993067e-06,
+      "loss": 0.4872,
+      "step": 3199
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.361049953739653e-06,
+      "loss": 0.4962,
+      "step": 3200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3589560727237485e-06,
+      "loss": 0.4915,
+      "step": 3201
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.356861934341903e-06,
+      "loss": 0.4665,
+      "step": 3202
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.354767538990715e-06,
+      "loss": 0.4966,
+      "step": 3203
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.352672887066831e-06,
+      "loss": 0.4773,
+      "step": 3204
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.350577978966947e-06,
+      "loss": 0.4864,
+      "step": 3205
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.348482815087806e-06,
+      "loss": 0.4978,
+      "step": 3206
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.346387395826204e-06,
+      "loss": 0.4761,
+      "step": 3207
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.344291721578979e-06,
+      "loss": 0.4847,
+      "step": 3208
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.342195792743022e-06,
+      "loss": 0.4795,
+      "step": 3209
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.340099609715269e-06,
+      "loss": 0.4667,
+      "step": 3210
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3380031728927086e-06,
+      "loss": 0.4631,
+      "step": 3211
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.335906482672369e-06,
+      "loss": 0.4637,
+      "step": 3212
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.333809539451339e-06,
+      "loss": 0.4633,
+      "step": 3213
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.331712343626744e-06,
+      "loss": 0.5021,
+      "step": 3214
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.329614895595761e-06,
+      "loss": 0.497,
+      "step": 3215
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.327517195755617e-06,
+      "loss": 0.4704,
+      "step": 3216
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.325419244503585e-06,
+      "loss": 0.4756,
+      "step": 3217
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.323321042236986e-06,
+      "loss": 0.4966,
+      "step": 3218
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.321222589353185e-06,
+      "loss": 0.4591,
+      "step": 3219
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.319123886249602e-06,
+      "loss": 0.4758,
+      "step": 3220
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.317024933323697e-06,
+      "loss": 0.4856,
+      "step": 3221
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.314925730972979e-06,
+      "loss": 0.4819,
+      "step": 3222
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.31282627959501e-06,
+      "loss": 0.4931,
+      "step": 3223
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.310726579587393e-06,
+      "loss": 0.498,
+      "step": 3224
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.308626631347776e-06,
+      "loss": 0.4875,
+      "step": 3225
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.306526435273862e-06,
+      "loss": 0.481,
+      "step": 3226
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.304425991763396e-06,
+      "loss": 0.4983,
+      "step": 3227
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.30232530121417e-06,
+      "loss": 0.5001,
+      "step": 3228
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.300224364024021e-06,
+      "loss": 0.4537,
+      "step": 3229
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.298123180590841e-06,
+      "loss": 0.4738,
+      "step": 3230
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.296021751312556e-06,
+      "loss": 0.5225,
+      "step": 3231
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.29392007658715e-06,
+      "loss": 0.4931,
+      "step": 3232
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.291818156812647e-06,
+      "loss": 0.4687,
+      "step": 3233
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.2897159923871174e-06,
+      "loss": 0.497,
+      "step": 3234
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.287613583708682e-06,
+      "loss": 0.4689,
+      "step": 3235
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.285510931175508e-06,
+      "loss": 0.5051,
+      "step": 3236
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.2834080351858e-06,
+      "loss": 0.4917,
+      "step": 3237
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.281304896137821e-06,
+      "loss": 0.4609,
+      "step": 3238
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.279201514429869e-06,
+      "loss": 0.4716,
+      "step": 3239
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2770978904602965e-06,
+      "loss": 0.468,
+      "step": 3240
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.274994024627498e-06,
+      "loss": 0.4828,
+      "step": 3241
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.272889917329912e-06,
+      "loss": 0.4739,
+      "step": 3242
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.270785568966027e-06,
+      "loss": 0.4728,
+      "step": 3243
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.268680979934376e-06,
+      "loss": 0.4693,
+      "step": 3244
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.266576150633534e-06,
+      "loss": 0.4495,
+      "step": 3245
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.264471081462125e-06,
+      "loss": 0.5069,
+      "step": 3246
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2623657728188194e-06,
+      "loss": 0.4823,
+      "step": 3247
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.260260225102328e-06,
+      "loss": 0.4727,
+      "step": 3248
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2581544387114126e-06,
+      "loss": 0.4617,
+      "step": 3249
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.256048414044877e-06,
+      "loss": 0.4889,
+      "step": 3250
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.253942151501571e-06,
+      "loss": 0.5017,
+      "step": 3251
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.251835651480387e-06,
+      "loss": 0.5023,
+      "step": 3252
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.249728914380267e-06,
+      "loss": 0.4857,
+      "step": 3253
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.247621940600196e-06,
+      "loss": 0.4953,
+      "step": 3254
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2455147305392e-06,
+      "loss": 0.4693,
+      "step": 3255
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.243407284596357e-06,
+      "loss": 0.478,
+      "step": 3256
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.241299603170784e-06,
+      "loss": 0.4837,
+      "step": 3257
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.239191686661642e-06,
+      "loss": 0.4569,
+      "step": 3258
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2370835354681425e-06,
+      "loss": 0.4839,
+      "step": 3259
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.234975149989537e-06,
+      "loss": 0.4678,
+      "step": 3260
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.232866530625119e-06,
+      "loss": 0.4586,
+      "step": 3261
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.230757677774235e-06,
+      "loss": 0.4604,
+      "step": 3262
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.228648591836266e-06,
+      "loss": 0.4903,
+      "step": 3263
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.226539273210642e-06,
+      "loss": 0.484,
+      "step": 3264
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.224429722296836e-06,
+      "loss": 0.4673,
+      "step": 3265
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2223199394943675e-06,
+      "loss": 0.4628,
+      "step": 3266
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.220209925202796e-06,
+      "loss": 0.5215,
+      "step": 3267
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.218099679821726e-06,
+      "loss": 0.492,
+      "step": 3268
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.215989203750809e-06,
+      "loss": 0.48,
+      "step": 3269
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.213878497389734e-06,
+      "loss": 0.5097,
+      "step": 3270
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.211767561138241e-06,
+      "loss": 0.4702,
+      "step": 3271
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2096563953961085e-06,
+      "loss": 0.4817,
+      "step": 3272
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2075450005631575e-06,
+      "loss": 0.4706,
+      "step": 3273
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.205433377039256e-06,
+      "loss": 0.4943,
+      "step": 3274
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.203321525224316e-06,
+      "loss": 0.4834,
+      "step": 3275
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.201209445518288e-06,
+      "loss": 0.4825,
+      "step": 3276
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.199097138321169e-06,
+      "loss": 0.4449,
+      "step": 3277
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.196984604032998e-06,
+      "loss": 0.4692,
+      "step": 3278
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.194871843053857e-06,
+      "loss": 0.4895,
+      "step": 3279
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.192758855783874e-06,
+      "loss": 0.496,
+      "step": 3280
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1906456426232145e-06,
+      "loss": 0.4549,
+      "step": 3281
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.188532203972088e-06,
+      "loss": 0.4846,
+      "step": 3282
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1864185402307535e-06,
+      "loss": 0.5138,
+      "step": 3283
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.184304651799502e-06,
+      "loss": 0.4803,
+      "step": 3284
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.182190539078675e-06,
+      "loss": 0.5088,
+      "step": 3285
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.180076202468654e-06,
+      "loss": 0.4749,
+      "step": 3286
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1779616423698615e-06,
+      "loss": 0.4667,
+      "step": 3287
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1758468591827635e-06,
+      "loss": 0.4897,
+      "step": 3288
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.17373185330787e-06,
+      "loss": 0.5068,
+      "step": 3289
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.171616625145731e-06,
+      "loss": 0.4982,
+      "step": 3290
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.169501175096936e-06,
+      "loss": 0.4535,
+      "step": 3291
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1673855035621225e-06,
+      "loss": 0.4966,
+      "step": 3292
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.165269610941967e-06,
+      "loss": 0.459,
+      "step": 3293
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.163153497637186e-06,
+      "loss": 0.4739,
+      "step": 3294
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.161037164048542e-06,
+      "loss": 0.4694,
+      "step": 3295
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.158920610576836e-06,
+      "loss": 0.471,
+      "step": 3296
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.156803837622911e-06,
+      "loss": 0.4976,
+      "step": 3297
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.154686845587653e-06,
+      "loss": 0.5117,
+      "step": 3298
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.152569634871987e-06,
+      "loss": 0.5202,
+      "step": 3299
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.150452205876885e-06,
+      "loss": 0.502,
+      "step": 3300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.148334559003351e-06,
+      "loss": 0.4751,
+      "step": 3301
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1462166946524385e-06,
+      "loss": 0.4874,
+      "step": 3302
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.144098613225239e-06,
+      "loss": 0.4825,
+      "step": 3303
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1419803151228855e-06,
+      "loss": 0.4789,
+      "step": 3304
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.139861800746553e-06,
+      "loss": 0.4541,
+      "step": 3305
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.137743070497454e-06,
+      "loss": 0.5027,
+      "step": 3306
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.135624124776845e-06,
+      "loss": 0.5026,
+      "step": 3307
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.133504963986025e-06,
+      "loss": 0.4923,
+      "step": 3308
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.131385588526329e-06,
+      "loss": 0.5194,
+      "step": 3309
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1292659987991345e-06,
+      "loss": 0.49,
+      "step": 3310
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.127146195205862e-06,
+      "loss": 0.4868,
+      "step": 3311
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1250261781479696e-06,
+      "loss": 0.5161,
+      "step": 3312
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.122905948026956e-06,
+      "loss": 0.4926,
+      "step": 3313
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.120785505244363e-06,
+      "loss": 0.4895,
+      "step": 3314
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.118664850201769e-06,
+      "loss": 0.4958,
+      "step": 3315
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.116543983300796e-06,
+      "loss": 0.4792,
+      "step": 3316
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.114422904943103e-06,
+      "loss": 0.4881,
+      "step": 3317
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.11230161553039e-06,
+      "loss": 0.494,
+      "step": 3318
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.110180115464399e-06,
+      "loss": 0.4823,
+      "step": 3319
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.108058405146908e-06,
+      "loss": 0.5304,
+      "step": 3320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.105936484979742e-06,
+      "loss": 0.4828,
+      "step": 3321
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.1038143553647565e-06,
+      "loss": 0.492,
+      "step": 3322
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.1016920167038516e-06,
+      "loss": 0.5097,
+      "step": 3323
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.099569469398968e-06,
+      "loss": 0.4531,
+      "step": 3324
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0974467138520844e-06,
+      "loss": 0.4712,
+      "step": 3325
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.095323750465217e-06,
+      "loss": 0.4729,
+      "step": 3326
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0932005796404256e-06,
+      "loss": 0.4639,
+      "step": 3327
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0910772017798046e-06,
+      "loss": 0.4789,
+      "step": 3328
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.088953617285493e-06,
+      "loss": 0.4723,
+      "step": 3329
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.086829826559663e-06,
+      "loss": 0.4646,
+      "step": 3330
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.084705830004529e-06,
+      "loss": 0.4988,
+      "step": 3331
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.082581628022347e-06,
+      "loss": 0.4921,
+      "step": 3332
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.080457221015407e-06,
+      "loss": 0.4585,
+      "step": 3333
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.078332609386039e-06,
+      "loss": 0.4858,
+      "step": 3334
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.076207793536615e-06,
+      "loss": 0.4499,
+      "step": 3335
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.07408277386954e-06,
+      "loss": 0.4697,
+      "step": 3336
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0719575507872655e-06,
+      "loss": 0.4487,
+      "step": 3337
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.069832124692274e-06,
+      "loss": 0.5265,
+      "step": 3338
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.067706495987089e-06,
+      "loss": 0.48,
+      "step": 3339
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.065580665074273e-06,
+      "loss": 0.4778,
+      "step": 3340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.063454632356428e-06,
+      "loss": 0.4993,
+      "step": 3341
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0613283982361915e-06,
+      "loss": 0.469,
+      "step": 3342
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.059201963116242e-06,
+      "loss": 0.5223,
+      "step": 3343
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.057075327399292e-06,
+      "loss": 0.5052,
+      "step": 3344
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.054948491488098e-06,
+      "loss": 0.4832,
+      "step": 3345
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.052821455785445e-06,
+      "loss": 0.5057,
+      "step": 3346
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.050694220694167e-06,
+      "loss": 0.494,
+      "step": 3347
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.048566786617131e-06,
+      "loss": 0.4933,
+      "step": 3348
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.046439153957236e-06,
+      "loss": 0.4901,
+      "step": 3349
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.044311323117429e-06,
+      "loss": 0.4996,
+      "step": 3350
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.042183294500686e-06,
+      "loss": 0.4749,
+      "step": 3351
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.040055068510026e-06,
+      "loss": 0.4531,
+      "step": 3352
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.037926645548501e-06,
+      "loss": 0.4611,
+      "step": 3353
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.035798026019204e-06,
+      "loss": 0.5043,
+      "step": 3354
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0336692103252635e-06,
+      "loss": 0.4871,
+      "step": 3355
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0315401988698445e-06,
+      "loss": 0.4829,
+      "step": 3356
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.029410992056151e-06,
+      "loss": 0.4836,
+      "step": 3357
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.027281590287421e-06,
+      "loss": 0.4635,
+      "step": 3358
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.025151993966935e-06,
+      "loss": 0.4821,
+      "step": 3359
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.023022203498004e-06,
+      "loss": 0.4682,
+      "step": 3360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.020892219283977e-06,
+      "loss": 0.446,
+      "step": 3361
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.018762041728243e-06,
+      "loss": 0.5007,
+      "step": 3362
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.016631671234228e-06,
+      "loss": 0.491,
+      "step": 3363
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0145011082053885e-06,
+      "loss": 0.4984,
+      "step": 3364
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.012370353045223e-06,
+      "loss": 0.4661,
+      "step": 3365
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.010239406157263e-06,
+      "loss": 0.4711,
+      "step": 3366
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.008108267945081e-06,
+      "loss": 0.5132,
+      "step": 3367
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0059769388122815e-06,
+      "loss": 0.4694,
+      "step": 3368
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.003845419162504e-06,
+      "loss": 0.5017,
+      "step": 3369
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.001713709399428e-06,
+      "loss": 0.508,
+      "step": 3370
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.99958180992677e-06,
+      "loss": 0.5013,
+      "step": 3371
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.997449721148276e-06,
+      "loss": 0.4873,
+      "step": 3372
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.995317443467733e-06,
+      "loss": 0.4859,
+      "step": 3373
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.993184977288964e-06,
+      "loss": 0.4656,
+      "step": 3374
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9910523230158235e-06,
+      "loss": 0.5167,
+      "step": 3375
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.988919481052205e-06,
+      "loss": 0.4846,
+      "step": 3376
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.98678645180204e-06,
+      "loss": 0.4682,
+      "step": 3377
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.98465323566929e-06,
+      "loss": 0.4895,
+      "step": 3378
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.982519833057952e-06,
+      "loss": 0.4808,
+      "step": 3379
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.980386244372063e-06,
+      "loss": 0.5202,
+      "step": 3380
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.978252470015694e-06,
+      "loss": 0.465,
+      "step": 3381
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.976118510392945e-06,
+      "loss": 0.4975,
+      "step": 3382
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.973984365907962e-06,
+      "loss": 0.4859,
+      "step": 3383
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.971850036964917e-06,
+      "loss": 0.5071,
+      "step": 3384
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.969715523968018e-06,
+      "loss": 0.473,
+      "step": 3385
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.967580827321515e-06,
+      "loss": 0.498,
+      "step": 3386
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.965445947429684e-06,
+      "loss": 0.472,
+      "step": 3387
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.963310884696839e-06,
+      "loss": 0.4633,
+      "step": 3388
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.961175639527331e-06,
+      "loss": 0.4976,
+      "step": 3389
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.959040212325542e-06,
+      "loss": 0.4856,
+      "step": 3390
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.95690460349589e-06,
+      "loss": 0.4985,
+      "step": 3391
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.954768813442828e-06,
+      "loss": 0.4831,
+      "step": 3392
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9526328425708425e-06,
+      "loss": 0.4867,
+      "step": 3393
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.950496691284454e-06,
+      "loss": 0.4582,
+      "step": 3394
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.948360359988218e-06,
+      "loss": 0.458,
+      "step": 3395
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.946223849086724e-06,
+      "loss": 0.4735,
+      "step": 3396
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.944087158984594e-06,
+      "loss": 0.506,
+      "step": 3397
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.941950290086485e-06,
+      "loss": 0.4959,
+      "step": 3398
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.939813242797091e-06,
+      "loss": 0.4983,
+      "step": 3399
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.937676017521133e-06,
+      "loss": 0.4666,
+      "step": 3400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.935538614663372e-06,
+      "loss": 0.5114,
+      "step": 3401
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.933401034628598e-06,
+      "loss": 0.5156,
+      "step": 3402
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.93126327782164e-06,
+      "loss": 0.4897,
+      "step": 3403
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.929125344647354e-06,
+      "loss": 0.4988,
+      "step": 3404
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.926987235510633e-06,
+      "loss": 0.4892,
+      "step": 3405
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.924848950816403e-06,
+      "loss": 0.4848,
+      "step": 3406
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.922710490969625e-06,
+      "loss": 0.4735,
+      "step": 3407
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.920571856375288e-06,
+      "loss": 0.4684,
+      "step": 3408
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.91843304743842e-06,
+      "loss": 0.4755,
+      "step": 3409
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.91629406456408e-06,
+      "loss": 0.4584,
+      "step": 3410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.914154908157355e-06,
+      "loss": 0.4675,
+      "step": 3411
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.912015578623375e-06,
+      "loss": 0.5059,
+      "step": 3412
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.909876076367294e-06,
+      "loss": 0.5019,
+      "step": 3413
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9077364017943005e-06,
+      "loss": 0.4787,
+      "step": 3414
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.905596555309619e-06,
+      "loss": 0.4946,
+      "step": 3415
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9034565373185035e-06,
+      "loss": 0.4949,
+      "step": 3416
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.901316348226243e-06,
+      "loss": 0.467,
+      "step": 3417
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8991759884381514e-06,
+      "loss": 0.4748,
+      "step": 3418
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.897035458359589e-06,
+      "loss": 0.5293,
+      "step": 3419
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.894894758395936e-06,
+      "loss": 0.502,
+      "step": 3420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.892753888952608e-06,
+      "loss": 0.4367,
+      "step": 3421
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.890612850435055e-06,
+      "loss": 0.4996,
+      "step": 3422
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8884716432487585e-06,
+      "loss": 0.4301,
+      "step": 3423
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8863302677992294e-06,
+      "loss": 0.4799,
+      "step": 3424
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.884188724492012e-06,
+      "loss": 0.4715,
+      "step": 3425
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.882047013732687e-06,
+      "loss": 0.4716,
+      "step": 3426
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8799051359268555e-06,
+      "loss": 0.4478,
+      "step": 3427
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.877763091480161e-06,
+      "loss": 0.5173,
+      "step": 3428
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.875620880798275e-06,
+      "loss": 0.4913,
+      "step": 3429
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.873478504286899e-06,
+      "loss": 0.4815,
+      "step": 3430
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.871335962351767e-06,
+      "loss": 0.5022,
+      "step": 3431
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.869193255398645e-06,
+      "loss": 0.4753,
+      "step": 3432
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.867050383833328e-06,
+      "loss": 0.4807,
+      "step": 3433
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.864907348061646e-06,
+      "loss": 0.4959,
+      "step": 3434
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.862764148489458e-06,
+      "loss": 0.4963,
+      "step": 3435
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.860620785522651e-06,
+      "loss": 0.5082,
+      "step": 3436
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.858477259567149e-06,
+      "loss": 0.4773,
+      "step": 3437
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8563335710289025e-06,
+      "loss": 0.4906,
+      "step": 3438
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.854189720313894e-06,
+      "loss": 0.4927,
+      "step": 3439
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8520457078281356e-06,
+      "loss": 0.5161,
+      "step": 3440
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.849901533977674e-06,
+      "loss": 0.4364,
+      "step": 3441
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.847757199168581e-06,
+      "loss": 0.4759,
+      "step": 3442
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8456127038069645e-06,
+      "loss": 0.5042,
+      "step": 3443
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8434680482989545e-06,
+      "loss": 0.4789,
+      "step": 3444
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.841323233050722e-06,
+      "loss": 0.5075,
+      "step": 3445
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8391782584684605e-06,
+      "loss": 0.471,
+      "step": 3446
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.837033124958397e-06,
+      "loss": 0.4594,
+      "step": 3447
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.834887832926788e-06,
+      "loss": 0.4531,
+      "step": 3448
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8327423827799165e-06,
+      "loss": 0.4738,
+      "step": 3449
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.830596774924101e-06,
+      "loss": 0.4484,
+      "step": 3450
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.82845100976569e-06,
+      "loss": 0.4782,
+      "step": 3451
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.826305087711055e-06,
+      "loss": 0.4813,
+      "step": 3452
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.824159009166604e-06,
+      "loss": 0.4894,
+      "step": 3453
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.822012774538771e-06,
+      "loss": 0.4967,
+      "step": 3454
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.819866384234021e-06,
+      "loss": 0.4759,
+      "step": 3455
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.81771983865885e-06,
+      "loss": 0.4779,
+      "step": 3456
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.815573138219779e-06,
+      "loss": 0.4717,
+      "step": 3457
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8134262833233614e-06,
+      "loss": 0.4785,
+      "step": 3458
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.811279274376181e-06,
+      "loss": 0.512,
+      "step": 3459
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.809132111784847e-06,
+      "loss": 0.46,
+      "step": 3460
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.806984795956001e-06,
+      "loss": 0.4601,
+      "step": 3461
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8048373272963144e-06,
+      "loss": 0.4998,
+      "step": 3462
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.802689706212481e-06,
+      "loss": 0.5102,
+      "step": 3463
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8005419331112325e-06,
+      "loss": 0.5195,
+      "step": 3464
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.798394008399323e-06,
+      "loss": 0.5079,
+      "step": 3465
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.796245932483537e-06,
+      "loss": 0.4763,
+      "step": 3466
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.79409770577069e-06,
+      "loss": 0.4684,
+      "step": 3467
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.791949328667621e-06,
+      "loss": 0.5075,
+      "step": 3468
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.789800801581202e-06,
+      "loss": 0.4742,
+      "step": 3469
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.787652124918331e-06,
+      "loss": 0.4801,
+      "step": 3470
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.785503299085938e-06,
+      "loss": 0.5036,
+      "step": 3471
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.783354324490976e-06,
+      "loss": 0.4682,
+      "step": 3472
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.781205201540428e-06,
+      "loss": 0.4696,
+      "step": 3473
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.779055930641308e-06,
+      "loss": 0.5183,
+      "step": 3474
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.776906512200654e-06,
+      "loss": 0.5162,
+      "step": 3475
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.774756946625533e-06,
+      "loss": 0.4764,
+      "step": 3476
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.772607234323043e-06,
+      "loss": 0.4838,
+      "step": 3477
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.770457375700307e-06,
+      "loss": 0.468,
+      "step": 3478
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.768307371164472e-06,
+      "loss": 0.4889,
+      "step": 3479
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.766157221122724e-06,
+      "loss": 0.4508,
+      "step": 3480
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.764006925982263e-06,
+      "loss": 0.4947,
+      "step": 3481
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.761856486150324e-06,
+      "loss": 0.498,
+      "step": 3482
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.75970590203417e-06,
+      "loss": 0.4469,
+      "step": 3483
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.757555174041089e-06,
+      "loss": 0.497,
+      "step": 3484
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7554043025783945e-06,
+      "loss": 0.4916,
+      "step": 3485
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.753253288053431e-06,
+      "loss": 0.4668,
+      "step": 3486
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.75110213087357e-06,
+      "loss": 0.4656,
+      "step": 3487
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.748950831446205e-06,
+      "loss": 0.5056,
+      "step": 3488
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.746799390178763e-06,
+      "loss": 0.4873,
+      "step": 3489
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.744647807478693e-06,
+      "loss": 0.481,
+      "step": 3490
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.742496083753474e-06,
+      "loss": 0.4657,
+      "step": 3491
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.740344219410607e-06,
+      "loss": 0.4498,
+      "step": 3492
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.738192214857629e-06,
+      "loss": 0.4626,
+      "step": 3493
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.73604007050209e-06,
+      "loss": 0.4928,
+      "step": 3494
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.733887786751581e-06,
+      "loss": 0.4784,
+      "step": 3495
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.731735364013706e-06,
+      "loss": 0.4951,
+      "step": 3496
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.729582802696106e-06,
+      "loss": 0.4986,
+      "step": 3497
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.727430103206443e-06,
+      "loss": 0.4944,
+      "step": 3498
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.725277265952405e-06,
+      "loss": 0.4663,
+      "step": 3499
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.723124291341707e-06,
+      "loss": 0.496,
+      "step": 3500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.720971179782092e-06,
+      "loss": 0.4791,
+      "step": 3501
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.718817931681326e-06,
+      "loss": 0.4742,
+      "step": 3502
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.716664547447201e-06,
+      "loss": 0.4819,
+      "step": 3503
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.714511027487538e-06,
+      "loss": 0.5132,
+      "step": 3504
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.71235737221018e-06,
+      "loss": 0.4747,
+      "step": 3505
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7102035820229966e-06,
+      "loss": 0.4821,
+      "step": 3506
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.708049657333885e-06,
+      "loss": 0.4892,
+      "step": 3507
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.705895598550765e-06,
+      "loss": 0.51,
+      "step": 3508
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.703741406081582e-06,
+      "loss": 0.4727,
+      "step": 3509
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.701587080334311e-06,
+      "loss": 0.479,
+      "step": 3510
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.699432621716947e-06,
+      "loss": 0.4869,
+      "step": 3511
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.697278030637512e-06,
+      "loss": 0.5123,
+      "step": 3512
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.695123307504055e-06,
+      "loss": 0.4843,
+      "step": 3513
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.692968452724647e-06,
+      "loss": 0.4977,
+      "step": 3514
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.690813466707384e-06,
+      "loss": 0.5083,
+      "step": 3515
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.68865834986039e-06,
+      "loss": 0.5061,
+      "step": 3516
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.686503102591813e-06,
+      "loss": 0.4737,
+      "step": 3517
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.68434772530982e-06,
+      "loss": 0.4609,
+      "step": 3518
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.682192218422613e-06,
+      "loss": 0.4866,
+      "step": 3519
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.68003658233841e-06,
+      "loss": 0.508,
+      "step": 3520
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.677880817465454e-06,
+      "loss": 0.5229,
+      "step": 3521
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.675724924212018e-06,
+      "loss": 0.5033,
+      "step": 3522
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.673568902986395e-06,
+      "loss": 0.4828,
+      "step": 3523
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.671412754196902e-06,
+      "loss": 0.5015,
+      "step": 3524
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.669256478251881e-06,
+      "loss": 0.5105,
+      "step": 3525
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6671000755597015e-06,
+      "loss": 0.4679,
+      "step": 3526
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.664943546528752e-06,
+      "loss": 0.4656,
+      "step": 3527
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.662786891567445e-06,
+      "loss": 0.4845,
+      "step": 3528
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.660630111084222e-06,
+      "loss": 0.4258,
+      "step": 3529
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.658473205487542e-06,
+      "loss": 0.4853,
+      "step": 3530
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.656316175185893e-06,
+      "loss": 0.4554,
+      "step": 3531
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.654159020587784e-06,
+      "loss": 0.4957,
+      "step": 3532
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.652001742101745e-06,
+      "loss": 0.4936,
+      "step": 3533
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.649844340136336e-06,
+      "loss": 0.515,
+      "step": 3534
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.647686815100134e-06,
+      "loss": 0.4879,
+      "step": 3535
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.645529167401742e-06,
+      "loss": 0.4779,
+      "step": 3536
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.643371397449789e-06,
+      "loss": 0.4829,
+      "step": 3537
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6412135056529214e-06,
+      "loss": 0.4913,
+      "step": 3538
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.639055492419812e-06,
+      "loss": 0.5006,
+      "step": 3539
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.636897358159159e-06,
+      "loss": 0.4985,
+      "step": 3540
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.634739103279676e-06,
+      "loss": 0.4207,
+      "step": 3541
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.632580728190108e-06,
+      "loss": 0.4829,
+      "step": 3542
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.630422233299218e-06,
+      "loss": 0.4917,
+      "step": 3543
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6282636190157915e-06,
+      "loss": 0.4602,
+      "step": 3544
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.626104885748637e-06,
+      "loss": 0.5011,
+      "step": 3545
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.623946033906591e-06,
+      "loss": 0.5072,
+      "step": 3546
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.621787063898504e-06,
+      "loss": 0.4835,
+      "step": 3547
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.61962797613325e-06,
+      "loss": 0.4801,
+      "step": 3548
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.617468771019734e-06,
+      "loss": 0.4808,
+      "step": 3549
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.615309448966874e-06,
+      "loss": 0.513,
+      "step": 3550
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.613150010383613e-06,
+      "loss": 0.4806,
+      "step": 3551
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.610990455678916e-06,
+      "loss": 0.5114,
+      "step": 3552
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.608830785261773e-06,
+      "loss": 0.4962,
+      "step": 3553
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.606670999541189e-06,
+      "loss": 0.4777,
+      "step": 3554
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.604511098926199e-06,
+      "loss": 0.4895,
+      "step": 3555
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.602351083825854e-06,
+      "loss": 0.4815,
+      "step": 3556
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.600190954649228e-06,
+      "loss": 0.4855,
+      "step": 3557
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.598030711805419e-06,
+      "loss": 0.4815,
+      "step": 3558
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.595870355703544e-06,
+      "loss": 0.4722,
+      "step": 3559
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.593709886752742e-06,
+      "loss": 0.4889,
+      "step": 3560
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.591549305362172e-06,
+      "loss": 0.4913,
+      "step": 3561
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.589388611941018e-06,
+      "loss": 0.5174,
+      "step": 3562
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.587227806898481e-06,
+      "loss": 0.4842,
+      "step": 3563
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.585066890643787e-06,
+      "loss": 0.4623,
+      "step": 3564
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.58290586358618e-06,
+      "loss": 0.4697,
+      "step": 3565
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.580744726134927e-06,
+      "loss": 0.4534,
+      "step": 3566
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.578583478699313e-06,
+      "loss": 0.4723,
+      "step": 3567
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.576422121688649e-06,
+      "loss": 0.4558,
+      "step": 3568
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.574260655512262e-06,
+      "loss": 0.4806,
+      "step": 3569
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5720990805795014e-06,
+      "loss": 0.4632,
+      "step": 3570
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5699373972997385e-06,
+      "loss": 0.5207,
+      "step": 3571
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.567775606082362e-06,
+      "loss": 0.4405,
+      "step": 3572
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.565613707336783e-06,
+      "loss": 0.5195,
+      "step": 3573
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.563451701472433e-06,
+      "loss": 0.5244,
+      "step": 3574
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.561289588898764e-06,
+      "loss": 0.5131,
+      "step": 3575
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5591273700252465e-06,
+      "loss": 0.4632,
+      "step": 3576
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5569650452613745e-06,
+      "loss": 0.4815,
+      "step": 3577
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.554802615016659e-06,
+      "loss": 0.4819,
+      "step": 3578
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.552640079700631e-06,
+      "loss": 0.4912,
+      "step": 3579
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.550477439722842e-06,
+      "loss": 0.4884,
+      "step": 3580
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.548314695492867e-06,
+      "loss": 0.4739,
+      "step": 3581
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.546151847420293e-06,
+      "loss": 0.4777,
+      "step": 3582
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5439888959147344e-06,
+      "loss": 0.5139,
+      "step": 3583
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.54182584138582e-06,
+      "loss": 0.5045,
+      "step": 3584
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.539662684243201e-06,
+      "loss": 0.5085,
+      "step": 3585
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5374994248965475e-06,
+      "loss": 0.4586,
+      "step": 3586
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.535336063755545e-06,
+      "loss": 0.4663,
+      "step": 3587
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.533172601229906e-06,
+      "loss": 0.4995,
+      "step": 3588
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.531009037729356e-06,
+      "loss": 0.494,
+      "step": 3589
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.528845373663641e-06,
+      "loss": 0.4807,
+      "step": 3590
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.526681609442527e-06,
+      "loss": 0.5064,
+      "step": 3591
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.524517745475799e-06,
+      "loss": 0.4676,
+      "step": 3592
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.522353782173261e-06,
+      "loss": 0.4893,
+      "step": 3593
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.520189719944733e-06,
+      "loss": 0.4593,
+      "step": 3594
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.518025559200059e-06,
+      "loss": 0.4743,
+      "step": 3595
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.515861300349097e-06,
+      "loss": 0.5078,
+      "step": 3596
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.513696943801724e-06,
+      "loss": 0.504,
+      "step": 3597
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.51153248996784e-06,
+      "loss": 0.5043,
+      "step": 3598
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.509367939257357e-06,
+      "loss": 0.472,
+      "step": 3599
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.50720329208021e-06,
+      "loss": 0.486,
+      "step": 3600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5050385488463505e-06,
+      "loss": 0.4967,
+      "step": 3601
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.50287370996575e-06,
+      "loss": 0.4833,
+      "step": 3602
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.500708775848392e-06,
+      "loss": 0.4794,
+      "step": 3603
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.498543746904289e-06,
+      "loss": 0.49,
+      "step": 3604
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.496378623543461e-06,
+      "loss": 0.4606,
+      "step": 3605
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.494213406175949e-06,
+      "loss": 0.4876,
+      "step": 3606
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.492048095211817e-06,
+      "loss": 0.4954,
+      "step": 3607
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.489882691061139e-06,
+      "loss": 0.4801,
+      "step": 3608
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.487717194134011e-06,
+      "loss": 0.4648,
+      "step": 3609
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.485551604840546e-06,
+      "loss": 0.4855,
+      "step": 3610
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.483385923590875e-06,
+      "loss": 0.4821,
+      "step": 3611
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4812201507951435e-06,
+      "loss": 0.5279,
+      "step": 3612
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.479054286863518e-06,
+      "loss": 0.4786,
+      "step": 3613
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.47688833220618e-06,
+      "loss": 0.4748,
+      "step": 3614
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.474722287233329e-06,
+      "loss": 0.4929,
+      "step": 3615
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.472556152355183e-06,
+      "loss": 0.5024,
+      "step": 3616
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.470389927981975e-06,
+      "loss": 0.494,
+      "step": 3617
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.468223614523953e-06,
+      "loss": 0.5138,
+      "step": 3618
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.466057212391385e-06,
+      "loss": 0.4925,
+      "step": 3619
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.463890721994559e-06,
+      "loss": 0.5159,
+      "step": 3620
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.461724143743772e-06,
+      "loss": 0.4437,
+      "step": 3621
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.459557478049343e-06,
+      "loss": 0.4894,
+      "step": 3622
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4573907253216055e-06,
+      "loss": 0.4747,
+      "step": 3623
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.45522388597091e-06,
+      "loss": 0.4871,
+      "step": 3624
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.453056960407624e-06,
+      "loss": 0.4549,
+      "step": 3625
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.450889949042129e-06,
+      "loss": 0.4877,
+      "step": 3626
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.448722852284827e-06,
+      "loss": 0.5061,
+      "step": 3627
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.446555670546133e-06,
+      "loss": 0.4849,
+      "step": 3628
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.444388404236478e-06,
+      "loss": 0.4965,
+      "step": 3629
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.44222105376631e-06,
+      "loss": 0.4793,
+      "step": 3630
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.440053619546094e-06,
+      "loss": 0.4892,
+      "step": 3631
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4378861019863075e-06,
+      "loss": 0.5036,
+      "step": 3632
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.435718501497447e-06,
+      "loss": 0.504,
+      "step": 3633
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.433550818490023e-06,
+      "loss": 0.4892,
+      "step": 3634
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.431383053374565e-06,
+      "loss": 0.4942,
+      "step": 3635
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.42921520656161e-06,
+      "loss": 0.4588,
+      "step": 3636
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4270472784617215e-06,
+      "loss": 0.4803,
+      "step": 3637
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.424879269485469e-06,
+      "loss": 0.467,
+      "step": 3638
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.422711180043439e-06,
+      "loss": 0.4715,
+      "step": 3639
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.420543010546242e-06,
+      "loss": 0.4951,
+      "step": 3640
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.418374761404492e-06,
+      "loss": 0.4734,
+      "step": 3641
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.416206433028822e-06,
+      "loss": 0.4809,
+      "step": 3642
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.414038025829884e-06,
+      "loss": 0.455,
+      "step": 3643
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.411869540218342e-06,
+      "loss": 0.4977,
+      "step": 3644
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.409700976604871e-06,
+      "loss": 0.4726,
+      "step": 3645
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.407532335400168e-06,
+      "loss": 0.4734,
+      "step": 3646
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.40536361701494e-06,
+      "loss": 0.4909,
+      "step": 3647
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.40319482185991e-06,
+      "loss": 0.4905,
+      "step": 3648
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.401025950345815e-06,
+      "loss": 0.4722,
+      "step": 3649
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.398857002883408e-06,
+      "loss": 0.4776,
+      "step": 3650
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.396687979883453e-06,
+      "loss": 0.5048,
+      "step": 3651
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.394518881756731e-06,
+      "loss": 0.4856,
+      "step": 3652
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.392349708914039e-06,
+      "loss": 0.4888,
+      "step": 3653
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.390180461766183e-06,
+      "loss": 0.4557,
+      "step": 3654
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.388011140723988e-06,
+      "loss": 0.5177,
+      "step": 3655
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.385841746198289e-06,
+      "loss": 0.5258,
+      "step": 3656
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.383672278599939e-06,
+      "loss": 0.448,
+      "step": 3657
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.381502738339801e-06,
+      "loss": 0.4816,
+      "step": 3658
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.379333125828757e-06,
+      "loss": 0.4515,
+      "step": 3659
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.377163441477692e-06,
+      "loss": 0.5108,
+      "step": 3660
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.374993685697519e-06,
+      "loss": 0.4855,
+      "step": 3661
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.372823858899153e-06,
+      "loss": 0.4809,
+      "step": 3662
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.370653961493528e-06,
+      "loss": 0.4692,
+      "step": 3663
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.368483993891591e-06,
+      "loss": 0.4845,
+      "step": 3664
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3663139565043e-06,
+      "loss": 0.4781,
+      "step": 3665
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.364143849742628e-06,
+      "loss": 0.4584,
+      "step": 3666
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.361973674017561e-06,
+      "loss": 0.4611,
+      "step": 3667
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.359803429740097e-06,
+      "loss": 0.4663,
+      "step": 3668
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.357633117321249e-06,
+      "loss": 0.4569,
+      "step": 3669
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.355462737172041e-06,
+      "loss": 0.4654,
+      "step": 3670
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.353292289703511e-06,
+      "loss": 0.5074,
+      "step": 3671
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3511217753267084e-06,
+      "loss": 0.4981,
+      "step": 3672
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.348951194452698e-06,
+      "loss": 0.4864,
+      "step": 3673
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.346780547492554e-06,
+      "loss": 0.5096,
+      "step": 3674
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.344609834857362e-06,
+      "loss": 0.492,
+      "step": 3675
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3424390569582264e-06,
+      "loss": 0.4529,
+      "step": 3676
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.340268214206259e-06,
+      "loss": 0.5138,
+      "step": 3677
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.338097307012584e-06,
+      "loss": 0.5152,
+      "step": 3678
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.335926335788338e-06,
+      "loss": 0.5071,
+      "step": 3679
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.333755300944674e-06,
+      "loss": 0.4591,
+      "step": 3680
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.33158420289275e-06,
+      "loss": 0.4971,
+      "step": 3681
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.32941304204374e-06,
+      "loss": 0.4813,
+      "step": 3682
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.327241818808831e-06,
+      "loss": 0.4948,
+      "step": 3683
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.325070533599219e-06,
+      "loss": 0.4807,
+      "step": 3684
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.322899186826113e-06,
+      "loss": 0.4614,
+      "step": 3685
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3207277789007335e-06,
+      "loss": 0.4725,
+      "step": 3686
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.318556310234313e-06,
+      "loss": 0.5059,
+      "step": 3687
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.316384781238094e-06,
+      "loss": 0.4735,
+      "step": 3688
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.314213192323336e-06,
+      "loss": 0.4855,
+      "step": 3689
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.312041543901299e-06,
+      "loss": 0.4709,
+      "step": 3690
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.309869836383264e-06,
+      "loss": 0.4936,
+      "step": 3691
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3076980701805205e-06,
+      "loss": 0.4845,
+      "step": 3692
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.305526245704368e-06,
+      "loss": 0.4896,
+      "step": 3693
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.303354363366115e-06,
+      "loss": 0.4717,
+      "step": 3694
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.301182423577089e-06,
+      "loss": 0.4893,
+      "step": 3695
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.299010426748618e-06,
+      "loss": 0.504,
+      "step": 3696
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.296838373292047e-06,
+      "loss": 0.5112,
+      "step": 3697
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.294666263618732e-06,
+      "loss": 0.4714,
+      "step": 3698
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.292494098140035e-06,
+      "loss": 0.4786,
+      "step": 3699
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.290321877267336e-06,
+      "loss": 0.4732,
+      "step": 3700
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2881496014120155e-06,
+      "loss": 0.4734,
+      "step": 3701
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.285977270985475e-06,
+      "loss": 0.4658,
+      "step": 3702
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.28380488639912e-06,
+      "loss": 0.5057,
+      "step": 3703
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2816324480643655e-06,
+      "loss": 0.4941,
+      "step": 3704
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.279459956392643e-06,
+      "loss": 0.4813,
+      "step": 3705
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.277287411795384e-06,
+      "loss": 0.5211,
+      "step": 3706
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.275114814684041e-06,
+      "loss": 0.5008,
+      "step": 3707
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.27294216547007e-06,
+      "loss": 0.4693,
+      "step": 3708
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.270769464564937e-06,
+      "loss": 0.4795,
+      "step": 3709
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.268596712380119e-06,
+      "loss": 0.5058,
+      "step": 3710
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.266423909327107e-06,
+      "loss": 0.4913,
+      "step": 3711
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.264251055817391e-06,
+      "loss": 0.4773,
+      "step": 3712
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.262078152262481e-06,
+      "loss": 0.4941,
+      "step": 3713
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.259905199073891e-06,
+      "loss": 0.4756,
+      "step": 3714
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2577321966631455e-06,
+      "loss": 0.4711,
+      "step": 3715
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.255559145441781e-06,
+      "loss": 0.4795,
+      "step": 3716
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.253386045821338e-06,
+      "loss": 0.4958,
+      "step": 3717
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.25121289821337e-06,
+      "loss": 0.4675,
+      "step": 3718
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.24903970302944e-06,
+      "loss": 0.4868,
+      "step": 3719
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.246866460681118e-06,
+      "loss": 0.495,
+      "step": 3720
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.244693171579981e-06,
+      "loss": 0.4474,
+      "step": 3721
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.242519836137621e-06,
+      "loss": 0.4959,
+      "step": 3722
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.240346454765634e-06,
+      "loss": 0.5095,
+      "step": 3723
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.238173027875625e-06,
+      "loss": 0.5012,
+      "step": 3724
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.23599955587921e-06,
+      "loss": 0.473,
+      "step": 3725
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.233826039188009e-06,
+      "loss": 0.4743,
+      "step": 3726
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2316524782136565e-06,
+      "loss": 0.4711,
+      "step": 3727
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.229478873367794e-06,
+      "loss": 0.4734,
+      "step": 3728
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.227305225062065e-06,
+      "loss": 0.478,
+      "step": 3729
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.225131533708129e-06,
+      "loss": 0.4707,
+      "step": 3730
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.222957799717651e-06,
+      "loss": 0.4391,
+      "step": 3731
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.220784023502302e-06,
+      "loss": 0.4842,
+      "step": 3732
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.218610205473762e-06,
+      "loss": 0.4644,
+      "step": 3733
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.216436346043721e-06,
+      "loss": 0.4949,
+      "step": 3734
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.214262445623879e-06,
+      "loss": 0.5057,
+      "step": 3735
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.212088504625931e-06,
+      "loss": 0.4828,
+      "step": 3736
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.209914523461599e-06,
+      "loss": 0.4606,
+      "step": 3737
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2077405025425965e-06,
+      "loss": 0.4894,
+      "step": 3738
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2055664422806515e-06,
+      "loss": 0.4505,
+      "step": 3739
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.203392343087499e-06,
+      "loss": 0.4836,
+      "step": 3740
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.201218205374882e-06,
+      "loss": 0.4921,
+      "step": 3741
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1990440295545474e-06,
+      "loss": 0.4984,
+      "step": 3742
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.196869816038253e-06,
+      "loss": 0.4751,
+      "step": 3743
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.194695565237762e-06,
+      "loss": 0.4732,
+      "step": 3744
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.192521277564844e-06,
+      "loss": 0.4886,
+      "step": 3745
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1903469534312775e-06,
+      "loss": 0.4491,
+      "step": 3746
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.188172593248848e-06,
+      "loss": 0.4936,
+      "step": 3747
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.185998197429345e-06,
+      "loss": 0.4546,
+      "step": 3748
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.183823766384566e-06,
+      "loss": 0.4783,
+      "step": 3749
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.181649300526319e-06,
+      "loss": 0.5045,
+      "step": 3750
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1794748002664115e-06,
+      "loss": 0.523,
+      "step": 3751
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.177300266016665e-06,
+      "loss": 0.5299,
+      "step": 3752
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1751256981889e-06,
+      "loss": 0.4975,
+      "step": 3753
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.172951097194948e-06,
+      "loss": 0.5009,
+      "step": 3754
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1707764634466485e-06,
+      "loss": 0.4941,
+      "step": 3755
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.168601797355842e-06,
+      "loss": 0.4708,
+      "step": 3756
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.166427099334378e-06,
+      "loss": 0.5043,
+      "step": 3757
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.164252369794114e-06,
+      "loss": 0.4654,
+      "step": 3758
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.162077609146907e-06,
+      "loss": 0.4886,
+      "step": 3759
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.159902817804627e-06,
+      "loss": 0.4874,
+      "step": 3760
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.157727996179148e-06,
+      "loss": 0.4863,
+      "step": 3761
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.155553144682347e-06,
+      "loss": 0.5069,
+      "step": 3762
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.153378263726107e-06,
+      "loss": 0.4804,
+      "step": 3763
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1512033537223214e-06,
+      "loss": 0.4594,
+      "step": 3764
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.149028415082883e-06,
+      "loss": 0.4774,
+      "step": 3765
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.146853448219692e-06,
+      "loss": 0.4542,
+      "step": 3766
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.144678453544657e-06,
+      "loss": 0.4969,
+      "step": 3767
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.142503431469689e-06,
+      "loss": 0.4687,
+      "step": 3768
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.140328382406701e-06,
+      "loss": 0.4825,
+      "step": 3769
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.138153306767619e-06,
+      "loss": 0.4994,
+      "step": 3770
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1359782049643705e-06,
+      "loss": 0.4874,
+      "step": 3771
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.133803077408881e-06,
+      "loss": 0.4694,
+      "step": 3772
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.131627924513094e-06,
+      "loss": 0.4727,
+      "step": 3773
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.129452746688949e-06,
+      "loss": 0.4867,
+      "step": 3774
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.127277544348388e-06,
+      "loss": 0.4683,
+      "step": 3775
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1251023179033675e-06,
+      "loss": 0.4685,
+      "step": 3776
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.12292706776584e-06,
+      "loss": 0.5251,
+      "step": 3777
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.120751794347766e-06,
+      "loss": 0.5063,
+      "step": 3778
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.118576498061109e-06,
+      "loss": 0.479,
+      "step": 3779
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.116401179317838e-06,
+      "loss": 0.4959,
+      "step": 3780
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.114225838529925e-06,
+      "loss": 0.45,
+      "step": 3781
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.112050476109347e-06,
+      "loss": 0.5039,
+      "step": 3782
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.109875092468088e-06,
+      "loss": 0.4986,
+      "step": 3783
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.10769968801813e-06,
+      "loss": 0.5098,
+      "step": 3784
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.105524263171463e-06,
+      "loss": 0.4862,
+      "step": 3785
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.10334881834008e-06,
+      "loss": 0.4684,
+      "step": 3786
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.101173353935978e-06,
+      "loss": 0.4819,
+      "step": 3787
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.098997870371155e-06,
+      "loss": 0.4704,
+      "step": 3788
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.09682236805762e-06,
+      "loss": 0.4573,
+      "step": 3789
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.094646847407375e-06,
+      "loss": 0.4939,
+      "step": 3790
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0924713088324365e-06,
+      "loss": 0.4885,
+      "step": 3791
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.090295752744813e-06,
+      "loss": 0.5,
+      "step": 3792
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.088120179556526e-06,
+      "loss": 0.4625,
+      "step": 3793
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.085944589679598e-06,
+      "loss": 0.4647,
+      "step": 3794
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.083768983526048e-06,
+      "loss": 0.4749,
+      "step": 3795
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.081593361507909e-06,
+      "loss": 0.5125,
+      "step": 3796
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.079417724037208e-06,
+      "loss": 0.5011,
+      "step": 3797
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.077242071525977e-06,
+      "loss": 0.4575,
+      "step": 3798
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.075066404386254e-06,
+      "loss": 0.4982,
+      "step": 3799
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0728907230300795e-06,
+      "loss": 0.4951,
+      "step": 3800
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.07071502786949e-06,
+      "loss": 0.5007,
+      "step": 3801
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.068539319316534e-06,
+      "loss": 0.4645,
+      "step": 3802
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0663635977832584e-06,
+      "loss": 0.4919,
+      "step": 3803
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.06418786368171e-06,
+      "loss": 0.4849,
+      "step": 3804
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.062012117423938e-06,
+      "loss": 0.4758,
+      "step": 3805
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0598363594220025e-06,
+      "loss": 0.5098,
+      "step": 3806
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.057660590087955e-06,
+      "loss": 0.493,
+      "step": 3807
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0554848098338535e-06,
+      "loss": 0.4805,
+      "step": 3808
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.053309019071761e-06,
+      "loss": 0.4851,
+      "step": 3809
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0511332182137376e-06,
+      "loss": 0.5035,
+      "step": 3810
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0489574076718475e-06,
+      "loss": 0.4951,
+      "step": 3811
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.046781587858157e-06,
+      "loss": 0.4816,
+      "step": 3812
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0446057591847355e-06,
+      "loss": 0.4822,
+      "step": 3813
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0424299220636496e-06,
+      "loss": 0.4929,
+      "step": 3814
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.04025407690697e-06,
+      "loss": 0.4865,
+      "step": 3815
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.038078224126774e-06,
+      "loss": 0.485,
+      "step": 3816
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.03590236413513e-06,
+      "loss": 0.493,
+      "step": 3817
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.033726497344116e-06,
+      "loss": 0.4666,
+      "step": 3818
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.031550624165808e-06,
+      "loss": 0.4738,
+      "step": 3819
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.029374745012284e-06,
+      "loss": 0.4924,
+      "step": 3820
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.027198860295623e-06,
+      "loss": 0.4741,
+      "step": 3821
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.025022970427906e-06,
+      "loss": 0.4711,
+      "step": 3822
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.022847075821212e-06,
+      "loss": 0.4915,
+      "step": 3823
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.020671176887624e-06,
+      "loss": 0.4485,
+      "step": 3824
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.018495274039224e-06,
+      "loss": 0.4551,
+      "step": 3825
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.016319367688096e-06,
+      "loss": 0.4883,
+      "step": 3826
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.014143458246325e-06,
+      "loss": 0.4564,
+      "step": 3827
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.011967546125993e-06,
+      "loss": 0.5212,
+      "step": 3828
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.009791631739188e-06,
+      "loss": 0.4626,
+      "step": 3829
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.007615715497996e-06,
+      "loss": 0.4617,
+      "step": 3830
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.005439797814498e-06,
+      "loss": 0.4849,
+      "step": 3831
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.003263879100786e-06,
+      "loss": 0.4792,
+      "step": 3832
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.001087959768944e-06,
+      "loss": 0.4547,
+      "step": 3833
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.998912040231057e-06,
+      "loss": 0.4875,
+      "step": 3834
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.996736120899216e-06,
+      "loss": 0.4694,
+      "step": 3835
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9945602021855034e-06,
+      "loss": 0.4896,
+      "step": 3836
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.992384284502007e-06,
+      "loss": 0.485,
+      "step": 3837
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.990208368260813e-06,
+      "loss": 0.5023,
+      "step": 3838
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.988032453874008e-06,
+      "loss": 0.4978,
+      "step": 3839
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.985856541753676e-06,
+      "loss": 0.5018,
+      "step": 3840
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.983680632311905e-06,
+      "loss": 0.4794,
+      "step": 3841
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9815047259607765e-06,
+      "loss": 0.4711,
+      "step": 3842
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.979328823112377e-06,
+      "loss": 0.4664,
+      "step": 3843
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9771529241787895e-06,
+      "loss": 0.5131,
+      "step": 3844
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.974977029572095e-06,
+      "loss": 0.4902,
+      "step": 3845
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9728011397043766e-06,
+      "loss": 0.5058,
+      "step": 3846
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.970625254987717e-06,
+      "loss": 0.4551,
+      "step": 3847
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9684493758341926e-06,
+      "loss": 0.4773,
+      "step": 3848
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.966273502655885e-06,
+      "loss": 0.4734,
+      "step": 3849
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.964097635864871e-06,
+      "loss": 0.5109,
+      "step": 3850
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.961921775873227e-06,
+      "loss": 0.4706,
+      "step": 3851
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.95974592309303e-06,
+      "loss": 0.4824,
+      "step": 3852
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.957570077936352e-06,
+      "loss": 0.4695,
+      "step": 3853
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.955394240815266e-06,
+      "loss": 0.477,
+      "step": 3854
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.953218412141843e-06,
+      "loss": 0.4341,
+      "step": 3855
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.951042592328153e-06,
+      "loss": 0.4937,
+      "step": 3856
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.948866781786263e-06,
+      "loss": 0.4921,
+      "step": 3857
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.94669098092824e-06,
+      "loss": 0.4821,
+      "step": 3858
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.944515190166147e-06,
+      "loss": 0.4622,
+      "step": 3859
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.942339409912046e-06,
+      "loss": 0.4909,
+      "step": 3860
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.940163640578e-06,
+      "loss": 0.4543,
+      "step": 3861
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.937987882576063e-06,
+      "loss": 0.4732,
+      "step": 3862
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9358121363182915e-06,
+      "loss": 0.4497,
+      "step": 3863
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.933636402216744e-06,
+      "loss": 0.4566,
+      "step": 3864
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.931460680683467e-06,
+      "loss": 0.5074,
+      "step": 3865
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.92928497213051e-06,
+      "loss": 0.4965,
+      "step": 3866
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.927109276969924e-06,
+      "loss": 0.4761,
+      "step": 3867
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9249335956137474e-06,
+      "loss": 0.5075,
+      "step": 3868
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.922757928474024e-06,
+      "loss": 0.4815,
+      "step": 3869
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.920582275962795e-06,
+      "loss": 0.5217,
+      "step": 3870
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.918406638492093e-06,
+      "loss": 0.4896,
+      "step": 3871
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.916231016473952e-06,
+      "loss": 0.4725,
+      "step": 3872
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9140554103204045e-06,
+      "loss": 0.4951,
+      "step": 3873
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.911879820443475e-06,
+      "loss": 0.4548,
+      "step": 3874
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.909704247255187e-06,
+      "loss": 0.474,
+      "step": 3875
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.907528691167567e-06,
+      "loss": 0.4513,
+      "step": 3876
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9053531525926266e-06,
+      "loss": 0.449,
+      "step": 3877
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.903177631942381e-06,
+      "loss": 0.4956,
+      "step": 3878
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9010021296288455e-06,
+      "loss": 0.4919,
+      "step": 3879
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.898826646064024e-06,
+      "loss": 0.4757,
+      "step": 3880
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.89665118165992e-06,
+      "loss": 0.4814,
+      "step": 3881
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.894475736828539e-06,
+      "loss": 0.5274,
+      "step": 3882
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.892300311981871e-06,
+      "loss": 0.4697,
+      "step": 3883
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.890124907531913e-06,
+      "loss": 0.4692,
+      "step": 3884
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.887949523890654e-06,
+      "loss": 0.4728,
+      "step": 3885
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.885774161470077e-06,
+      "loss": 0.4946,
+      "step": 3886
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.883598820682165e-06,
+      "loss": 0.493,
+      "step": 3887
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.881423501938894e-06,
+      "loss": 0.4659,
+      "step": 3888
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.879248205652236e-06,
+      "loss": 0.5037,
+      "step": 3889
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.877072932234162e-06,
+      "loss": 0.4889,
+      "step": 3890
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.874897682096634e-06,
+      "loss": 0.464,
+      "step": 3891
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.872722455651612e-06,
+      "loss": 0.4752,
+      "step": 3892
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.870547253311054e-06,
+      "loss": 0.47,
+      "step": 3893
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.868372075486907e-06,
+      "loss": 0.4797,
+      "step": 3894
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8661969225911195e-06,
+      "loss": 0.4972,
+      "step": 3895
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.864021795035633e-06,
+      "loss": 0.4247,
+      "step": 3896
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.861846693232382e-06,
+      "loss": 0.4728,
+      "step": 3897
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8596716175933e-06,
+      "loss": 0.5154,
+      "step": 3898
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8574965685303145e-06,
+      "loss": 0.4751,
+      "step": 3899
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8553215464553445e-06,
+      "loss": 0.4571,
+      "step": 3900
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.853146551780309e-06,
+      "loss": 0.4911,
+      "step": 3901
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.85097158491712e-06,
+      "loss": 0.4926,
+      "step": 3902
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.84879664627768e-06,
+      "loss": 0.4832,
+      "step": 3903
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.846621736273894e-06,
+      "loss": 0.4798,
+      "step": 3904
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.844446855317655e-06,
+      "loss": 0.4907,
+      "step": 3905
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.842272003820852e-06,
+      "loss": 0.4731,
+      "step": 3906
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.840097182195373e-06,
+      "loss": 0.4967,
+      "step": 3907
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.837922390853094e-06,
+      "loss": 0.4683,
+      "step": 3908
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.835747630205888e-06,
+      "loss": 0.496,
+      "step": 3909
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.833572900665622e-06,
+      "loss": 0.4844,
+      "step": 3910
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.831398202644159e-06,
+      "loss": 0.4968,
+      "step": 3911
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.829223536553353e-06,
+      "loss": 0.4462,
+      "step": 3912
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.827048902805054e-06,
+      "loss": 0.4973,
+      "step": 3913
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.824874301811102e-06,
+      "loss": 0.4961,
+      "step": 3914
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.822699733983337e-06,
+      "loss": 0.4856,
+      "step": 3915
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.820525199733591e-06,
+      "loss": 0.4429,
+      "step": 3916
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.818350699473682e-06,
+      "loss": 0.5009,
+      "step": 3917
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.816176233615434e-06,
+      "loss": 0.4902,
+      "step": 3918
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.814001802570658e-06,
+      "loss": 0.4933,
+      "step": 3919
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.811827406751153e-06,
+      "loss": 0.4315,
+      "step": 3920
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.809653046568723e-06,
+      "loss": 0.4826,
+      "step": 3921
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.807478722435159e-06,
+      "loss": 0.4954,
+      "step": 3922
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.80530443476224e-06,
+      "loss": 0.5198,
+      "step": 3923
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.803130183961748e-06,
+      "loss": 0.4813,
+      "step": 3924
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.800955970445456e-06,
+      "loss": 0.4942,
+      "step": 3925
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.798781794625119e-06,
+      "loss": 0.454,
+      "step": 3926
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.796607656912501e-06,
+      "loss": 0.4083,
+      "step": 3927
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.794433557719351e-06,
+      "loss": 0.4857,
+      "step": 3928
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.792259497457404e-06,
+      "loss": 0.4899,
+      "step": 3929
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.790085476538402e-06,
+      "loss": 0.4789,
+      "step": 3930
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7879114953740695e-06,
+      "loss": 0.48,
+      "step": 3931
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.785737554376123e-06,
+      "loss": 0.4885,
+      "step": 3932
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.783563653956278e-06,
+      "loss": 0.4498,
+      "step": 3933
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.78138979452624e-06,
+      "loss": 0.4843,
+      "step": 3934
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.779215976497699e-06,
+      "loss": 0.4881,
+      "step": 3935
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.77704220028235e-06,
+      "loss": 0.5043,
+      "step": 3936
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.774868466291873e-06,
+      "loss": 0.4907,
+      "step": 3937
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7726947749379365e-06,
+      "loss": 0.5072,
+      "step": 3938
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7705211266322085e-06,
+      "loss": 0.4949,
+      "step": 3939
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.768347521786344e-06,
+      "loss": 0.4834,
+      "step": 3940
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.766173960811993e-06,
+      "loss": 0.4948,
+      "step": 3941
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.764000444120794e-06,
+      "loss": 0.4981,
+      "step": 3942
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.761826972124377e-06,
+      "loss": 0.4658,
+      "step": 3943
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.759653545234368e-06,
+      "loss": 0.4772,
+      "step": 3944
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.757480163862381e-06,
+      "loss": 0.4906,
+      "step": 3945
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.75530682842002e-06,
+      "loss": 0.4653,
+      "step": 3946
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.753133539318883e-06,
+      "loss": 0.4905,
+      "step": 3947
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7509602969705616e-06,
+      "loss": 0.5446,
+      "step": 3948
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.748787101786631e-06,
+      "loss": 0.4651,
+      "step": 3949
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.746613954178663e-06,
+      "loss": 0.4563,
+      "step": 3950
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.744440854558221e-06,
+      "loss": 0.4805,
+      "step": 3951
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.742267803336855e-06,
+      "loss": 0.4946,
+      "step": 3952
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.740094800926111e-06,
+      "loss": 0.485,
+      "step": 3953
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.737921847737521e-06,
+      "loss": 0.4738,
+      "step": 3954
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.735748944182611e-06,
+      "loss": 0.4521,
+      "step": 3955
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.733576090672895e-06,
+      "loss": 0.4761,
+      "step": 3956
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7314032876198815e-06,
+      "loss": 0.485,
+      "step": 3957
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.729230535435064e-06,
+      "loss": 0.4758,
+      "step": 3958
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.727057834529931e-06,
+      "loss": 0.4682,
+      "step": 3959
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7248851853159605e-06,
+      "loss": 0.4717,
+      "step": 3960
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.722712588204617e-06,
+      "loss": 0.4659,
+      "step": 3961
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.72054004360736e-06,
+      "loss": 0.4792,
+      "step": 3962
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.718367551935635e-06,
+      "loss": 0.5004,
+      "step": 3963
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.716195113600881e-06,
+      "loss": 0.4493,
+      "step": 3964
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7140227290145265e-06,
+      "loss": 0.4822,
+      "step": 3965
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.711850398587985e-06,
+      "loss": 0.4554,
+      "step": 3966
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.709678122732666e-06,
+      "loss": 0.4635,
+      "step": 3967
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.707505901859966e-06,
+      "loss": 0.4637,
+      "step": 3968
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.70533373638127e-06,
+      "loss": 0.5122,
+      "step": 3969
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.703161626707953e-06,
+      "loss": 0.4653,
+      "step": 3970
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.700989573251384e-06,
+      "loss": 0.498,
+      "step": 3971
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6988175764229126e-06,
+      "loss": 0.4954,
+      "step": 3972
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.696645636633885e-06,
+      "loss": 0.4946,
+      "step": 3973
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.694473754295634e-06,
+      "loss": 0.4536,
+      "step": 3974
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.69230192981948e-06,
+      "loss": 0.4853,
+      "step": 3975
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.690130163616737e-06,
+      "loss": 0.4805,
+      "step": 3976
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.687958456098703e-06,
+      "loss": 0.476,
+      "step": 3977
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.685786807676667e-06,
+      "loss": 0.4896,
+      "step": 3978
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.683615218761906e-06,
+      "loss": 0.487,
+      "step": 3979
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.681443689765688e-06,
+      "loss": 0.4929,
+      "step": 3980
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.679272221099268e-06,
+      "loss": 0.499,
+      "step": 3981
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.677100813173888e-06,
+      "loss": 0.4825,
+      "step": 3982
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.674929466400783e-06,
+      "loss": 0.4628,
+      "step": 3983
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6727581811911705e-06,
+      "loss": 0.4786,
+      "step": 3984
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.670586957956261e-06,
+      "loss": 0.4897,
+      "step": 3985
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.668415797107252e-06,
+      "loss": 0.5053,
+      "step": 3986
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.666244699055327e-06,
+      "loss": 0.4901,
+      "step": 3987
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.664073664211664e-06,
+      "loss": 0.4747,
+      "step": 3988
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.661902692987418e-06,
+      "loss": 0.4584,
+      "step": 3989
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.659731785793743e-06,
+      "loss": 0.4897,
+      "step": 3990
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.657560943041776e-06,
+      "loss": 0.4737,
+      "step": 3991
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.65539016514264e-06,
+      "loss": 0.4707,
+      "step": 3992
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.653219452507448e-06,
+      "loss": 0.4877,
+      "step": 3993
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.651048805547304e-06,
+      "loss": 0.4871,
+      "step": 3994
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.648878224673292e-06,
+      "loss": 0.4692,
+      "step": 3995
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.646707710296489e-06,
+      "loss": 0.4973,
+      "step": 3996
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.64453726282796e-06,
+      "loss": 0.4944,
+      "step": 3997
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.642366882678752e-06,
+      "loss": 0.4687,
+      "step": 3998
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.640196570259903e-06,
+      "loss": 0.4867,
+      "step": 3999
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.638026325982441e-06,
+      "loss": 0.493,
+      "step": 4000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.635856150257374e-06,
+      "loss": 0.4535,
+      "step": 4001
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6336860434957e-06,
+      "loss": 0.4563,
+      "step": 4002
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.631516006108411e-06,
+      "loss": 0.5008,
+      "step": 4003
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.629346038506473e-06,
+      "loss": 0.4901,
+      "step": 4004
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.627176141100847e-06,
+      "loss": 0.4863,
+      "step": 4005
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6250063143024835e-06,
+      "loss": 0.4513,
+      "step": 4006
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.622836558522309e-06,
+      "loss": 0.4644,
+      "step": 4007
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.620666874171244e-06,
+      "loss": 0.4928,
+      "step": 4008
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6184972616602e-06,
+      "loss": 0.4608,
+      "step": 4009
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.616327721400062e-06,
+      "loss": 0.4366,
+      "step": 4010
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.614158253801711e-06,
+      "loss": 0.4814,
+      "step": 4011
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6119888592760146e-06,
+      "loss": 0.4752,
+      "step": 4012
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.609819538233818e-06,
+      "loss": 0.4374,
+      "step": 4013
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.607650291085964e-06,
+      "loss": 0.51,
+      "step": 4014
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.605481118243271e-06,
+      "loss": 0.482,
+      "step": 4015
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.60331202011655e-06,
+      "loss": 0.4734,
+      "step": 4016
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.601142997116595e-06,
+      "loss": 0.4749,
+      "step": 4017
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.598974049654187e-06,
+      "loss": 0.4664,
+      "step": 4018
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.596805178140091e-06,
+      "loss": 0.4596,
+      "step": 4019
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.594636382985061e-06,
+      "loss": 0.4687,
+      "step": 4020
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.592467664599833e-06,
+      "loss": 0.4633,
+      "step": 4021
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.59029902339513e-06,
+      "loss": 0.468,
+      "step": 4022
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5881304597816604e-06,
+      "loss": 0.4822,
+      "step": 4023
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5859619741701165e-06,
+      "loss": 0.4874,
+      "step": 4024
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.583793566971179e-06,
+      "loss": 0.5059,
+      "step": 4025
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5816252385955105e-06,
+      "loss": 0.4557,
+      "step": 4026
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5794569894537595e-06,
+      "loss": 0.5076,
+      "step": 4027
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.577288819956561e-06,
+      "loss": 0.4605,
+      "step": 4028
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.575120730514534e-06,
+      "loss": 0.484,
+      "step": 4029
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.572952721538281e-06,
+      "loss": 0.4722,
+      "step": 4030
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.57078479343839e-06,
+      "loss": 0.4883,
+      "step": 4031
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.568616946625437e-06,
+      "loss": 0.4777,
+      "step": 4032
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.566449181509978e-06,
+      "loss": 0.5003,
+      "step": 4033
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.564281498502554e-06,
+      "loss": 0.432,
+      "step": 4034
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.562113898013695e-06,
+      "loss": 0.4756,
+      "step": 4035
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.559946380453908e-06,
+      "loss": 0.4561,
+      "step": 4036
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.557778946233691e-06,
+      "loss": 0.48,
+      "step": 4037
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.555611595763524e-06,
+      "loss": 0.4806,
+      "step": 4038
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.553444329453868e-06,
+      "loss": 0.4823,
+      "step": 4039
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.551277147715175e-06,
+      "loss": 0.4694,
+      "step": 4040
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.549110050957872e-06,
+      "loss": 0.4493,
+      "step": 4041
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.546943039592377e-06,
+      "loss": 0.4842,
+      "step": 4042
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.544776114029093e-06,
+      "loss": 0.4715,
+      "step": 4043
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.542609274678396e-06,
+      "loss": 0.4687,
+      "step": 4044
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.540442521950658e-06,
+      "loss": 0.5115,
+      "step": 4045
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.538275856256231e-06,
+      "loss": 0.5054,
+      "step": 4046
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.536109278005442e-06,
+      "loss": 0.5047,
+      "step": 4047
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5339427876086146e-06,
+      "loss": 0.4761,
+      "step": 4048
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.53177638547605e-06,
+      "loss": 0.5004,
+      "step": 4049
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.529610072018027e-06,
+      "loss": 0.4741,
+      "step": 4050
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.527443847644818e-06,
+      "loss": 0.4724,
+      "step": 4051
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5252777127666725e-06,
+      "loss": 0.4939,
+      "step": 4052
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.523111667793821e-06,
+      "loss": 0.4485,
+      "step": 4053
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.520945713136483e-06,
+      "loss": 0.4768,
+      "step": 4054
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.518779849204859e-06,
+      "loss": 0.4727,
+      "step": 4055
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.516614076409126e-06,
+      "loss": 0.5082,
+      "step": 4056
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.514448395159455e-06,
+      "loss": 0.4786,
+      "step": 4057
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.512282805865991e-06,
+      "loss": 0.5029,
+      "step": 4058
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.510117308938861e-06,
+      "loss": 0.4993,
+      "step": 4059
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5079519047881835e-06,
+      "loss": 0.445,
+      "step": 4060
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.505786593824052e-06,
+      "loss": 0.4722,
+      "step": 4061
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.50362137645654e-06,
+      "loss": 0.4318,
+      "step": 4062
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.501456253095711e-06,
+      "loss": 0.5144,
+      "step": 4063
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4992912241516085e-06,
+      "loss": 0.4394,
+      "step": 4064
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4971262900342514e-06,
+      "loss": 0.4813,
+      "step": 4065
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.494961451153652e-06,
+      "loss": 0.5111,
+      "step": 4066
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.492796707919792e-06,
+      "loss": 0.4839,
+      "step": 4067
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.490632060742644e-06,
+      "loss": 0.4614,
+      "step": 4068
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.488467510032163e-06,
+      "loss": 0.4308,
+      "step": 4069
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4863030561982775e-06,
+      "loss": 0.4643,
+      "step": 4070
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.484138699650905e-06,
+      "loss": 0.4758,
+      "step": 4071
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.481974440799943e-06,
+      "loss": 0.4668,
+      "step": 4072
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.479810280055269e-06,
+      "loss": 0.4723,
+      "step": 4073
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.477646217826741e-06,
+      "loss": 0.4869,
+      "step": 4074
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.475482254524203e-06,
+      "loss": 0.4715,
+      "step": 4075
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.473318390557475e-06,
+      "loss": 0.4568,
+      "step": 4076
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.471154626336361e-06,
+      "loss": 0.4747,
+      "step": 4077
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4689909622706464e-06,
+      "loss": 0.4757,
+      "step": 4078
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.466827398770096e-06,
+      "loss": 0.4601,
+      "step": 4079
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.464663936244456e-06,
+      "loss": 0.4626,
+      "step": 4080
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.462500575103456e-06,
+      "loss": 0.4733,
+      "step": 4081
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4603373157568005e-06,
+      "loss": 0.4969,
+      "step": 4082
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.458174158614181e-06,
+      "loss": 0.4976,
+      "step": 4083
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.456011104085267e-06,
+      "loss": 0.4776,
+      "step": 4084
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4538481525797074e-06,
+      "loss": 0.4993,
+      "step": 4085
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.451685304507134e-06,
+      "loss": 0.4762,
+      "step": 4086
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4495225602771585e-06,
+      "loss": 0.4959,
+      "step": 4087
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4473599202993705e-06,
+      "loss": 0.5135,
+      "step": 4088
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.445197384983343e-06,
+      "loss": 0.5122,
+      "step": 4089
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.443034954738627e-06,
+      "loss": 0.453,
+      "step": 4090
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.440872629974754e-06,
+      "loss": 0.5147,
+      "step": 4091
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.438710411101238e-06,
+      "loss": 0.4984,
+      "step": 4092
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.436548298527569e-06,
+      "loss": 0.5161,
+      "step": 4093
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.434386292663218e-06,
+      "loss": 0.4509,
+      "step": 4094
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.43222439391764e-06,
+      "loss": 0.4702,
+      "step": 4095
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.430062602700263e-06,
+      "loss": 0.4738,
+      "step": 4096
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4279009194204985e-06,
+      "loss": 0.49,
+      "step": 4097
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.425739344487739e-06,
+      "loss": 0.5122,
+      "step": 4098
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.423577878311351e-06,
+      "loss": 0.4906,
+      "step": 4099
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.421416521300687e-06,
+      "loss": 0.4637,
+      "step": 4100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.419255273865075e-06,
+      "loss": 0.4668,
+      "step": 4101
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.417094136413822e-06,
+      "loss": 0.4859,
+      "step": 4102
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.414933109356214e-06,
+      "loss": 0.4658,
+      "step": 4103
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.412772193101521e-06,
+      "loss": 0.4839,
+      "step": 4104
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.410611388058984e-06,
+      "loss": 0.4922,
+      "step": 4105
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.408450694637829e-06,
+      "loss": 0.458,
+      "step": 4106
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4062901132472604e-06,
+      "loss": 0.4926,
+      "step": 4107
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.404129644296457e-06,
+      "loss": 0.4614,
+      "step": 4108
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.401969288194581e-06,
+      "loss": 0.4787,
+      "step": 4109
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.399809045350773e-06,
+      "loss": 0.4815,
+      "step": 4110
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.397648916174147e-06,
+      "loss": 0.4778,
+      "step": 4111
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.395488901073802e-06,
+      "loss": 0.472,
+      "step": 4112
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.393329000458812e-06,
+      "loss": 0.4584,
+      "step": 4113
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.391169214738229e-06,
+      "loss": 0.4487,
+      "step": 4114
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.389009544321084e-06,
+      "loss": 0.475,
+      "step": 4115
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3868499896163885e-06,
+      "loss": 0.4757,
+      "step": 4116
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3846905510331265e-06,
+      "loss": 0.4359,
+      "step": 4117
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.382531228980268e-06,
+      "loss": 0.4709,
+      "step": 4118
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3803720238667505e-06,
+      "loss": 0.5215,
+      "step": 4119
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.378212936101498e-06,
+      "loss": 0.4747,
+      "step": 4120
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.376053966093411e-06,
+      "loss": 0.5014,
+      "step": 4121
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.373895114251363e-06,
+      "loss": 0.4678,
+      "step": 4122
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.371736380984209e-06,
+      "loss": 0.5127,
+      "step": 4123
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.369577766700785e-06,
+      "loss": 0.503,
+      "step": 4124
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.367419271809894e-06,
+      "loss": 0.5063,
+      "step": 4125
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.365260896720324e-06,
+      "loss": 0.4843,
+      "step": 4126
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.363102641840844e-06,
+      "loss": 0.4765,
+      "step": 4127
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3609445075801894e-06,
+      "loss": 0.5106,
+      "step": 4128
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.358786494347079e-06,
+      "loss": 0.4983,
+      "step": 4129
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.356628602550213e-06,
+      "loss": 0.4365,
+      "step": 4130
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.354470832598259e-06,
+      "loss": 0.4816,
+      "step": 4131
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.352313184899867e-06,
+      "loss": 0.445,
+      "step": 4132
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.350155659863667e-06,
+      "loss": 0.4829,
+      "step": 4133
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.347998257898256e-06,
+      "loss": 0.4588,
+      "step": 4134
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.345840979412217e-06,
+      "loss": 0.5022,
+      "step": 4135
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.343683824814109e-06,
+      "loss": 0.5017,
+      "step": 4136
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.341526794512459e-06,
+      "loss": 0.4832,
+      "step": 4137
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3393698889157785e-06,
+      "loss": 0.4758,
+      "step": 4138
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.337213108432556e-06,
+      "loss": 0.4975,
+      "step": 4139
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.33505645347125e-06,
+      "loss": 0.4964,
+      "step": 4140
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.332899924440298e-06,
+      "loss": 0.5239,
+      "step": 4141
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3307435217481195e-06,
+      "loss": 0.4702,
+      "step": 4142
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3285872458031e-06,
+      "loss": 0.4721,
+      "step": 4143
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.326431097013608e-06,
+      "loss": 0.4714,
+      "step": 4144
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3242750757879835e-06,
+      "loss": 0.46,
+      "step": 4145
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.322119182534547e-06,
+      "loss": 0.4426,
+      "step": 4146
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.319963417661593e-06,
+      "loss": 0.4965,
+      "step": 4147
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.317807781577388e-06,
+      "loss": 0.487,
+      "step": 4148
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3156522746901806e-06,
+      "loss": 0.4856,
+      "step": 4149
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.313496897408189e-06,
+      "loss": 0.4929,
+      "step": 4150
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.311341650139611e-06,
+      "loss": 0.4859,
+      "step": 4151
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.309186533292617e-06,
+      "loss": 0.495,
+      "step": 4152
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.307031547275356e-06,
+      "loss": 0.4553,
+      "step": 4153
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.304876692495947e-06,
+      "loss": 0.4702,
+      "step": 4154
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3027219693624885e-06,
+      "loss": 0.4843,
+      "step": 4155
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.300567378283055e-06,
+      "loss": 0.4456,
+      "step": 4156
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.29841291966569e-06,
+      "loss": 0.4851,
+      "step": 4157
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.296258593918419e-06,
+      "loss": 0.5071,
+      "step": 4158
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.294104401449237e-06,
+      "loss": 0.5006,
+      "step": 4159
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.291950342666117e-06,
+      "loss": 0.4423,
+      "step": 4160
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.289796417977004e-06,
+      "loss": 0.5123,
+      "step": 4161
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.287642627789822e-06,
+      "loss": 0.4582,
+      "step": 4162
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2854889725124635e-06,
+      "loss": 0.4719,
+      "step": 4163
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2833354525527995e-06,
+      "loss": 0.4313,
+      "step": 4164
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.281182068318675e-06,
+      "loss": 0.4505,
+      "step": 4165
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2790288202179085e-06,
+      "loss": 0.4735,
+      "step": 4166
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.276875708658294e-06,
+      "loss": 0.4838,
+      "step": 4167
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.274722734047597e-06,
+      "loss": 0.4746,
+      "step": 4168
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.272569896793558e-06,
+      "loss": 0.5076,
+      "step": 4169
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.270417197303895e-06,
+      "loss": 0.4752,
+      "step": 4170
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.268264635986295e-06,
+      "loss": 0.4792,
+      "step": 4171
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.26611221324842e-06,
+      "loss": 0.4626,
+      "step": 4172
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.263959929497911e-06,
+      "loss": 0.4679,
+      "step": 4173
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.261807785142373e-06,
+      "loss": 0.4807,
+      "step": 4174
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.259655780589392e-06,
+      "loss": 0.499,
+      "step": 4175
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2575039162465295e-06,
+      "loss": 0.4743,
+      "step": 4176
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2553521925213074e-06,
+      "loss": 0.4883,
+      "step": 4177
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.253200609821238e-06,
+      "loss": 0.4688,
+      "step": 4178
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.251049168553797e-06,
+      "loss": 0.4874,
+      "step": 4179
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.248897869126431e-06,
+      "loss": 0.4918,
+      "step": 4180
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.246746711946569e-06,
+      "loss": 0.4913,
+      "step": 4181
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.244595697421608e-06,
+      "loss": 0.504,
+      "step": 4182
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.242444825958912e-06,
+      "loss": 0.4861,
+      "step": 4183
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.24029409796583e-06,
+      "loss": 0.4845,
+      "step": 4184
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.238143513849678e-06,
+      "loss": 0.4858,
+      "step": 4185
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.235993074017738e-06,
+      "loss": 0.4715,
+      "step": 4186
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.233842778877277e-06,
+      "loss": 0.4693,
+      "step": 4187
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.231692628835529e-06,
+      "loss": 0.4573,
+      "step": 4188
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.229542624299694e-06,
+      "loss": 0.4776,
+      "step": 4189
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.227392765676957e-06,
+      "loss": 0.4495,
+      "step": 4190
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.225243053374469e-06,
+      "loss": 0.4951,
+      "step": 4191
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.223093487799348e-06,
+      "loss": 0.4845,
+      "step": 4192
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.220944069358693e-06,
+      "loss": 0.5019,
+      "step": 4193
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.218794798459575e-06,
+      "loss": 0.4771,
+      "step": 4194
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2166456755090255e-06,
+      "loss": 0.4725,
+      "step": 4195
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.214496700914065e-06,
+      "loss": 0.4502,
+      "step": 4196
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2123478750816705e-06,
+      "loss": 0.4815,
+      "step": 4197
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2101991984188e-06,
+      "loss": 0.5149,
+      "step": 4198
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.208050671332382e-06,
+      "loss": 0.4584,
+      "step": 4199
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.205902294229313e-06,
+      "loss": 0.4522,
+      "step": 4200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.203754067516464e-06,
+      "loss": 0.5109,
+      "step": 4201
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.201605991600679e-06,
+      "loss": 0.4741,
+      "step": 4202
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.199458066888769e-06,
+      "loss": 0.4809,
+      "step": 4203
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.19731029378752e-06,
+      "loss": 0.4953,
+      "step": 4204
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.195162672703689e-06,
+      "loss": 0.4739,
+      "step": 4205
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.193015204044e-06,
+      "loss": 0.4836,
+      "step": 4206
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.190867888215154e-06,
+      "loss": 0.5039,
+      "step": 4207
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.188720725623822e-06,
+      "loss": 0.4889,
+      "step": 4208
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.18657371667664e-06,
+      "loss": 0.472,
+      "step": 4209
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.184426861780223e-06,
+      "loss": 0.4291,
+      "step": 4210
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.182280161341152e-06,
+      "loss": 0.4669,
+      "step": 4211
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.18013361576598e-06,
+      "loss": 0.4792,
+      "step": 4212
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.17798722546123e-06,
+      "loss": 0.514,
+      "step": 4213
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.175840990833398e-06,
+      "loss": 0.4793,
+      "step": 4214
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.173694912288946e-06,
+      "loss": 0.4696,
+      "step": 4215
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.171548990234311e-06,
+      "loss": 0.4756,
+      "step": 4216
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1694032250759e-06,
+      "loss": 0.4819,
+      "step": 4217
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.167257617220085e-06,
+      "loss": 0.4933,
+      "step": 4218
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.165112167073215e-06,
+      "loss": 0.4918,
+      "step": 4219
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1629668750416046e-06,
+      "loss": 0.4362,
+      "step": 4220
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.16082174153154e-06,
+      "loss": 0.4839,
+      "step": 4221
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1586767669492794e-06,
+      "loss": 0.4447,
+      "step": 4222
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.156531951701046e-06,
+      "loss": 0.4747,
+      "step": 4223
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.154387296193037e-06,
+      "loss": 0.4545,
+      "step": 4224
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.15224280083142e-06,
+      "loss": 0.4767,
+      "step": 4225
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.150098466022327e-06,
+      "loss": 0.4582,
+      "step": 4226
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.147954292171864e-06,
+      "loss": 0.5026,
+      "step": 4227
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.145810279686108e-06,
+      "loss": 0.4984,
+      "step": 4228
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.143666428971098e-06,
+      "loss": 0.486,
+      "step": 4229
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.141522740432852e-06,
+      "loss": 0.4687,
+      "step": 4230
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1393792144773496e-06,
+      "loss": 0.496,
+      "step": 4231
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.137235851510544e-06,
+      "loss": 0.4624,
+      "step": 4232
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.135092651938354e-06,
+      "loss": 0.4743,
+      "step": 4233
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.132949616166673e-06,
+      "loss": 0.485,
+      "step": 4234
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.130806744601356e-06,
+      "loss": 0.5068,
+      "step": 4235
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.128664037648233e-06,
+      "loss": 0.4798,
+      "step": 4236
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.126521495713103e-06,
+      "loss": 0.4827,
+      "step": 4237
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.124379119201726e-06,
+      "loss": 0.4723,
+      "step": 4238
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.122236908519839e-06,
+      "loss": 0.4509,
+      "step": 4239
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.120094864073146e-06,
+      "loss": 0.4907,
+      "step": 4240
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.117952986267315e-06,
+      "loss": 0.4719,
+      "step": 4241
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.115811275507987e-06,
+      "loss": 0.4707,
+      "step": 4242
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.113669732200772e-06,
+      "loss": 0.4662,
+      "step": 4243
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.111528356751242e-06,
+      "loss": 0.4992,
+      "step": 4244
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.109387149564947e-06,
+      "loss": 0.4574,
+      "step": 4245
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1072461110473935e-06,
+      "loss": 0.4787,
+      "step": 4246
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1051052416040655e-06,
+      "loss": 0.4922,
+      "step": 4247
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.102964541640413e-06,
+      "loss": 0.4597,
+      "step": 4248
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.100824011561849e-06,
+      "loss": 0.4968,
+      "step": 4249
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.098683651773759e-06,
+      "loss": 0.5024,
+      "step": 4250
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.096543462681499e-06,
+      "loss": 0.4733,
+      "step": 4251
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.094403444690383e-06,
+      "loss": 0.4747,
+      "step": 4252
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0922635982057e-06,
+      "loss": 0.4734,
+      "step": 4253
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.090123923632709e-06,
+      "loss": 0.4944,
+      "step": 4254
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.087984421376627e-06,
+      "loss": 0.4988,
+      "step": 4255
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.085845091842644e-06,
+      "loss": 0.4858,
+      "step": 4256
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0837059354359225e-06,
+      "loss": 0.48,
+      "step": 4257
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0815669525615816e-06,
+      "loss": 0.4661,
+      "step": 4258
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.079428143624712e-06,
+      "loss": 0.4754,
+      "step": 4259
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.077289509030378e-06,
+      "loss": 0.4827,
+      "step": 4260
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.075151049183599e-06,
+      "loss": 0.436,
+      "step": 4261
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.073012764489368e-06,
+      "loss": 0.4649,
+      "step": 4262
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.070874655352649e-06,
+      "loss": 0.5109,
+      "step": 4263
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.068736722178363e-06,
+      "loss": 0.4843,
+      "step": 4264
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.066598965371402e-06,
+      "loss": 0.4507,
+      "step": 4265
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.06446138533663e-06,
+      "loss": 0.507,
+      "step": 4266
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.062323982478868e-06,
+      "loss": 0.454,
+      "step": 4267
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.060186757202909e-06,
+      "loss": 0.466,
+      "step": 4268
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0580497099135155e-06,
+      "loss": 0.457,
+      "step": 4269
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.055912841015407e-06,
+      "loss": 0.4833,
+      "step": 4270
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.053776150913279e-06,
+      "loss": 0.4711,
+      "step": 4271
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.051639640011783e-06,
+      "loss": 0.4662,
+      "step": 4272
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.049503308715547e-06,
+      "loss": 0.4995,
+      "step": 4273
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.047367157429159e-06,
+      "loss": 0.4874,
+      "step": 4274
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.045231186557173e-06,
+      "loss": 0.4645,
+      "step": 4275
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.043095396504111e-06,
+      "loss": 0.4767,
+      "step": 4276
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.04095978767446e-06,
+      "loss": 0.5075,
+      "step": 4277
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.03882436047267e-06,
+      "loss": 0.4857,
+      "step": 4278
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.036689115303162e-06,
+      "loss": 0.4487,
+      "step": 4279
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.034554052570318e-06,
+      "loss": 0.4628,
+      "step": 4280
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.032419172678486e-06,
+      "loss": 0.4858,
+      "step": 4281
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.030284476031982e-06,
+      "loss": 0.5062,
+      "step": 4282
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0281499630350855e-06,
+      "loss": 0.4564,
+      "step": 4283
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.026015634092039e-06,
+      "loss": 0.497,
+      "step": 4284
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.023881489607056e-06,
+      "loss": 0.5275,
+      "step": 4285
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.021747529984309e-06,
+      "loss": 0.4763,
+      "step": 4286
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.019613755627938e-06,
+      "loss": 0.4454,
+      "step": 4287
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.017480166942049e-06,
+      "loss": 0.4594,
+      "step": 4288
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.015346764330712e-06,
+      "loss": 0.4604,
+      "step": 4289
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0132135481979605e-06,
+      "loss": 0.4999,
+      "step": 4290
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.011080518947795e-06,
+      "loss": 0.4701,
+      "step": 4291
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.008947676984178e-06,
+      "loss": 0.4489,
+      "step": 4292
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0068150227110375e-06,
+      "loss": 0.4539,
+      "step": 4293
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.004682556532267e-06,
+      "loss": 0.4634,
+      "step": 4294
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.002550278851725e-06,
+      "loss": 0.4619,
+      "step": 4295
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.000418190073231e-06,
+      "loss": 0.4408,
+      "step": 4296
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.998286290600573e-06,
+      "loss": 0.4878,
+      "step": 4297
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.996154580837497e-06,
+      "loss": 0.4413,
+      "step": 4298
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.994023061187721e-06,
+      "loss": 0.5042,
+      "step": 4299
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.991891732054921e-06,
+      "loss": 0.4475,
+      "step": 4300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.989760593842738e-06,
+      "loss": 0.4739,
+      "step": 4301
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.987629646954779e-06,
+      "loss": 0.4664,
+      "step": 4302
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.985498891794615e-06,
+      "loss": 0.4728,
+      "step": 4303
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.983368328765774e-06,
+      "loss": 0.4905,
+      "step": 4304
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.981237958271757e-06,
+      "loss": 0.4672,
+      "step": 4305
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.979107780716025e-06,
+      "loss": 0.4272,
+      "step": 4306
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.976977796501999e-06,
+      "loss": 0.4888,
+      "step": 4307
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.974848006033066e-06,
+      "loss": 0.5214,
+      "step": 4308
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.97271840971258e-06,
+      "loss": 0.4922,
+      "step": 4309
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.97058900794385e-06,
+      "loss": 0.4955,
+      "step": 4310
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.968459801130156e-06,
+      "loss": 0.4463,
+      "step": 4311
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.966330789674739e-06,
+      "loss": 0.4824,
+      "step": 4312
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.964201973980797e-06,
+      "loss": 0.4908,
+      "step": 4313
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9620733544515e-06,
+      "loss": 0.5033,
+      "step": 4314
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.959944931489977e-06,
+      "loss": 0.4713,
+      "step": 4315
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9578167054993145e-06,
+      "loss": 0.458,
+      "step": 4316
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.955688676882572e-06,
+      "loss": 0.492,
+      "step": 4317
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.953560846042765e-06,
+      "loss": 0.4661,
+      "step": 4318
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.95143321338287e-06,
+      "loss": 0.4648,
+      "step": 4319
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.949305779305833e-06,
+      "loss": 0.4651,
+      "step": 4320
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.947178544214557e-06,
+      "loss": 0.4514,
+      "step": 4321
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.945051508511904e-06,
+      "loss": 0.464,
+      "step": 4322
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.94292467260071e-06,
+      "loss": 0.4648,
+      "step": 4323
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.940798036883761e-06,
+      "loss": 0.4787,
+      "step": 4324
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.938671601763809e-06,
+      "loss": 0.4949,
+      "step": 4325
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.936545367643574e-06,
+      "loss": 0.4342,
+      "step": 4326
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.934419334925729e-06,
+      "loss": 0.4862,
+      "step": 4327
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.932293504012913e-06,
+      "loss": 0.5081,
+      "step": 4328
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9301678753077285e-06,
+      "loss": 0.4801,
+      "step": 4329
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.928042449212736e-06,
+      "loss": 0.4624,
+      "step": 4330
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9259172261304604e-06,
+      "loss": 0.4549,
+      "step": 4331
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9237922064633875e-06,
+      "loss": 0.4809,
+      "step": 4332
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.921667390613962e-06,
+      "loss": 0.4713,
+      "step": 4333
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.919542778984595e-06,
+      "loss": 0.4626,
+      "step": 4334
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9174183719776545e-06,
+      "loss": 0.4876,
+      "step": 4335
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.915294169995472e-06,
+      "loss": 0.4713,
+      "step": 4336
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.913170173440339e-06,
+      "loss": 0.4958,
+      "step": 4337
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.91104638271451e-06,
+      "loss": 0.4835,
+      "step": 4338
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.908922798220196e-06,
+      "loss": 0.4515,
+      "step": 4339
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.906799420359576e-06,
+      "loss": 0.4672,
+      "step": 4340
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.904676249534784e-06,
+      "loss": 0.4536,
+      "step": 4341
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.902553286147917e-06,
+      "loss": 0.4701,
+      "step": 4342
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.900430530601033e-06,
+      "loss": 0.4929,
+      "step": 4343
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.89830798329615e-06,
+      "loss": 0.4984,
+      "step": 4344
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.896185644635245e-06,
+      "loss": 0.4666,
+      "step": 4345
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.89406351502026e-06,
+      "loss": 0.4626,
+      "step": 4346
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8919415948530925e-06,
+      "loss": 0.4937,
+      "step": 4347
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.889819884535603e-06,
+      "loss": 0.4992,
+      "step": 4348
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.887698384469613e-06,
+      "loss": 0.4584,
+      "step": 4349
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8855770950569e-06,
+      "loss": 0.4577,
+      "step": 4350
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.883456016699206e-06,
+      "loss": 0.4637,
+      "step": 4351
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.881335149798233e-06,
+      "loss": 0.4662,
+      "step": 4352
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8792144947556384e-06,
+      "loss": 0.4621,
+      "step": 4353
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.877094051973045e-06,
+      "loss": 0.4936,
+      "step": 4354
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.874973821852032e-06,
+      "loss": 0.4897,
+      "step": 4355
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8728538047941395e-06,
+      "loss": 0.4796,
+      "step": 4356
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.870734001200866e-06,
+      "loss": 0.4541,
+      "step": 4357
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.868614411473673e-06,
+      "loss": 0.4586,
+      "step": 4358
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.866495036013976e-06,
+      "loss": 0.4737,
+      "step": 4359
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8643758752231555e-06,
+      "loss": 0.4588,
+      "step": 4360
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.862256929502547e-06,
+      "loss": 0.4901,
+      "step": 4361
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.860138199253449e-06,
+      "loss": 0.4578,
+      "step": 4362
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8580196848771145e-06,
+      "loss": 0.4875,
+      "step": 4363
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.855901386774763e-06,
+      "loss": 0.4609,
+      "step": 4364
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.853783305347562e-06,
+      "loss": 0.4739,
+      "step": 4365
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.85166544099665e-06,
+      "loss": 0.469,
+      "step": 4366
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.849547794123118e-06,
+      "loss": 0.4742,
+      "step": 4367
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8474303651280135e-06,
+      "loss": 0.4742,
+      "step": 4368
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.845313154412348e-06,
+      "loss": 0.4785,
+      "step": 4369
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.84319616237709e-06,
+      "loss": 0.4703,
+      "step": 4370
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.841079389423165e-06,
+      "loss": 0.4855,
+      "step": 4371
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8389628359514585e-06,
+      "loss": 0.4391,
+      "step": 4372
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8368465023628145e-06,
+      "loss": 0.4631,
+      "step": 4373
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.834730389058034e-06,
+      "loss": 0.4897,
+      "step": 4374
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.83261449643788e-06,
+      "loss": 0.4427,
+      "step": 4375
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.830498824903066e-06,
+      "loss": 0.5111,
+      "step": 4376
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8283833748542705e-06,
+      "loss": 0.4557,
+      "step": 4377
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.826268146692132e-06,
+      "loss": 0.4709,
+      "step": 4378
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.824153140817238e-06,
+      "loss": 0.4689,
+      "step": 4379
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.822038357630139e-06,
+      "loss": 0.4558,
+      "step": 4380
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.819923797531348e-06,
+      "loss": 0.5041,
+      "step": 4381
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.817809460921326e-06,
+      "loss": 0.4666,
+      "step": 4382
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.815695348200498e-06,
+      "loss": 0.4322,
+      "step": 4383
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.813581459769249e-06,
+      "loss": 0.4676,
+      "step": 4384
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.811467796027913e-06,
+      "loss": 0.4924,
+      "step": 4385
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8093543573767868e-06,
+      "loss": 0.4715,
+      "step": 4386
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8072411442161284e-06,
+      "loss": 0.4671,
+      "step": 4387
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.805128156946144e-06,
+      "loss": 0.4703,
+      "step": 4388
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8030153959670023e-06,
+      "loss": 0.4742,
+      "step": 4389
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8009028616788334e-06,
+      "loss": 0.4665,
+      "step": 4390
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7987905544817138e-06,
+      "loss": 0.4996,
+      "step": 4391
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.796678474775684e-06,
+      "loss": 0.5088,
+      "step": 4392
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.794566622960745e-06,
+      "loss": 0.4455,
+      "step": 4393
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7924549994368437e-06,
+      "loss": 0.4699,
+      "step": 4394
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7903436046038923e-06,
+      "loss": 0.4793,
+      "step": 4395
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7882324388617596e-06,
+      "loss": 0.4666,
+      "step": 4396
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.786121502610266e-06,
+      "loss": 0.4559,
+      "step": 4397
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.784010796249191e-06,
+      "loss": 0.4861,
+      "step": 4398
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.781900320178275e-06,
+      "loss": 0.4958,
+      "step": 4399
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7797900747972054e-06,
+      "loss": 0.4978,
+      "step": 4400
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7776800605056346e-06,
+      "loss": 0.4899,
+      "step": 4401
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.775570277703165e-06,
+      "loss": 0.4989,
+      "step": 4402
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7734607267893597e-06,
+      "loss": 0.4858,
+      "step": 4403
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7713514081637365e-06,
+      "loss": 0.477,
+      "step": 4404
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7692423222257667e-06,
+      "loss": 0.4625,
+      "step": 4405
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7671334693748808e-06,
+      "loss": 0.4497,
+      "step": 4406
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.765024850010465e-06,
+      "loss": 0.4649,
+      "step": 4407
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7629164645318584e-06,
+      "loss": 0.4854,
+      "step": 4408
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7608083133383584e-06,
+      "loss": 0.4783,
+      "step": 4409
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7587003968292184e-06,
+      "loss": 0.4808,
+      "step": 4410
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.756592715403644e-06,
+      "loss": 0.4819,
+      "step": 4411
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7544852694608004e-06,
+      "loss": 0.4996,
+      "step": 4412
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.752378059399806e-06,
+      "loss": 0.4816,
+      "step": 4413
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7502710856197337e-06,
+      "loss": 0.4845,
+      "step": 4414
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.748164348519614e-06,
+      "loss": 0.4731,
+      "step": 4415
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.746057848498432e-06,
+      "loss": 0.5104,
+      "step": 4416
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7439515859551246e-06,
+      "loss": 0.475,
+      "step": 4417
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7418455612885883e-06,
+      "loss": 0.4936,
+      "step": 4418
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7397397748976737e-06,
+      "loss": 0.4763,
+      "step": 4419
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7376342271811826e-06,
+      "loss": 0.4788,
+      "step": 4420
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.735528918537876e-06,
+      "loss": 0.4707,
+      "step": 4421
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7334238493664677e-06,
+      "loss": 0.4603,
+      "step": 4422
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7313190200656257e-06,
+      "loss": 0.4813,
+      "step": 4423
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.729214431033973e-06,
+      "loss": 0.4431,
+      "step": 4424
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.727110082670089e-06,
+      "loss": 0.4536,
+      "step": 4425
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.725005975372503e-06,
+      "loss": 0.4975,
+      "step": 4426
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.722902109539705e-06,
+      "loss": 0.4661,
+      "step": 4427
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.720798485570132e-06,
+      "loss": 0.4932,
+      "step": 4428
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.718695103862181e-06,
+      "loss": 0.474,
+      "step": 4429
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7165919648142013e-06,
+      "loss": 0.4683,
+      "step": 4430
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7144890688244937e-06,
+      "loss": 0.499,
+      "step": 4431
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.712386416291317e-06,
+      "loss": 0.4804,
+      "step": 4432
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7102840076128842e-06,
+      "loss": 0.4627,
+      "step": 4433
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.708181843187355e-06,
+      "loss": 0.4786,
+      "step": 4434
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7060799234128507e-06,
+      "loss": 0.4747,
+      "step": 4435
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7039782486874463e-06,
+      "loss": 0.4734,
+      "step": 4436
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.701876819409161e-06,
+      "loss": 0.5061,
+      "step": 4437
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6997756359759784e-06,
+      "loss": 0.4842,
+      "step": 4438
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6976746987858332e-06,
+      "loss": 0.4887,
+      "step": 4439
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6955740082366047e-06,
+      "loss": 0.4617,
+      "step": 4440
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6934735647261377e-06,
+      "loss": 0.4582,
+      "step": 4441
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.691373368652226e-06,
+      "loss": 0.4686,
+      "step": 4442
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.689273420412609e-06,
+      "loss": 0.4999,
+      "step": 4443
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6871737204049897e-06,
+      "loss": 0.4651,
+      "step": 4444
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6850742690270214e-06,
+      "loss": 0.4873,
+      "step": 4445
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6829750666763043e-06,
+      "loss": 0.4906,
+      "step": 4446
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6808761137503986e-06,
+      "loss": 0.507,
+      "step": 4447
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6787774106468162e-06,
+      "loss": 0.4921,
+      "step": 4448
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6766789577630146e-06,
+      "loss": 0.471,
+      "step": 4449
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6745807554964148e-06,
+      "loss": 0.4781,
+      "step": 4450
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6724828042443845e-06,
+      "loss": 0.4643,
+      "step": 4451
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6703851044042395e-06,
+      "loss": 0.4984,
+      "step": 4452
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.668287656373259e-06,
+      "loss": 0.4755,
+      "step": 4453
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6661904605486633e-06,
+      "loss": 0.4829,
+      "step": 4454
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6640935173276314e-06,
+      "loss": 0.4672,
+      "step": 4455
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.661996827107295e-06,
+      "loss": 0.4636,
+      "step": 4456
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.659900390284733e-06,
+      "loss": 0.4785,
+      "step": 4457
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6578042072569797e-06,
+      "loss": 0.4661,
+      "step": 4458
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.655708278421023e-06,
+      "loss": 0.4982,
+      "step": 4459
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.653612604173798e-06,
+      "loss": 0.4712,
+      "step": 4460
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.651517184912195e-06,
+      "loss": 0.4595,
+      "step": 4461
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6494220210330554e-06,
+      "loss": 0.4254,
+      "step": 4462
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.647327112933171e-06,
+      "loss": 0.5013,
+      "step": 4463
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.645232461009287e-06,
+      "loss": 0.4681,
+      "step": 4464
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.643138065658099e-06,
+      "loss": 0.4647,
+      "step": 4465
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6410439272762528e-06,
+      "loss": 0.4703,
+      "step": 4466
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.638950046260348e-06,
+      "loss": 0.4903,
+      "step": 4467
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.636856423006935e-06,
+      "loss": 0.4907,
+      "step": 4468
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6347630579125125e-06,
+      "loss": 0.4505,
+      "step": 4469
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6326699513735342e-06,
+      "loss": 0.4624,
+      "step": 4470
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6305771037864045e-06,
+      "loss": 0.5045,
+      "step": 4471
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.628484515547474e-06,
+      "loss": 0.4999,
+      "step": 4472
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6263921870530495e-06,
+      "loss": 0.5033,
+      "step": 4473
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6243001186993875e-06,
+      "loss": 0.4864,
+      "step": 4474
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6222083108826927e-06,
+      "loss": 0.4952,
+      "step": 4475
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6201167639991243e-06,
+      "loss": 0.4996,
+      "step": 4476
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6180254784447865e-06,
+      "loss": 0.4585,
+      "step": 4477
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.615934454615741e-06,
+      "loss": 0.5067,
+      "step": 4478
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6138436929079957e-06,
+      "loss": 0.4524,
+      "step": 4479
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.611753193717508e-06,
+      "loss": 0.4428,
+      "step": 4480
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.609662957440187e-06,
+      "loss": 0.4588,
+      "step": 4481
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6075729844718954e-06,
+      "loss": 0.4571,
+      "step": 4482
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.605483275208439e-06,
+      "loss": 0.4779,
+      "step": 4483
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6033938300455796e-06,
+      "loss": 0.4978,
+      "step": 4484
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6013046493790273e-06,
+      "loss": 0.5048,
+      "step": 4485
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5992157336044397e-06,
+      "loss": 0.4893,
+      "step": 4486
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5971270831174266e-06,
+      "loss": 0.4813,
+      "step": 4487
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.59503869831355e-06,
+      "loss": 0.458,
+      "step": 4488
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5929505795883145e-06,
+      "loss": 0.4432,
+      "step": 4489
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5908627273371803e-06,
+      "loss": 0.4599,
+      "step": 4490
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.588775141955557e-06,
+      "loss": 0.4948,
+      "step": 4491
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5866878238387997e-06,
+      "loss": 0.4352,
+      "step": 4492
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5846007733822165e-06,
+      "loss": 0.4748,
+      "step": 4493
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.582513990981063e-06,
+      "loss": 0.4762,
+      "step": 4494
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5804274770305445e-06,
+      "loss": 0.4523,
+      "step": 4495
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5783412319258153e-06,
+      "loss": 0.4831,
+      "step": 4496
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5762552560619806e-06,
+      "loss": 0.454,
+      "step": 4497
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.574169549834091e-06,
+      "loss": 0.4795,
+      "step": 4498
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5720841136371487e-06,
+      "loss": 0.4774,
+      "step": 4499
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.569998947866105e-06,
+      "loss": 0.4663,
+      "step": 4500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5679140529158572e-06,
+      "loss": 0.4977,
+      "step": 4501
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5658294291812567e-06,
+      "loss": 0.4789,
+      "step": 4502
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.563745077057095e-06,
+      "loss": 0.4948,
+      "step": 4503
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5616609969381204e-06,
+      "loss": 0.468,
+      "step": 4504
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.559577189219028e-06,
+      "loss": 0.4662,
+      "step": 4505
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.557493654294455e-06,
+      "loss": 0.4687,
+      "step": 4506
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5554103925589955e-06,
+      "loss": 0.4833,
+      "step": 4507
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5533274044071885e-06,
+      "loss": 0.4876,
+      "step": 4508
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5512446902335178e-06,
+      "loss": 0.5004,
+      "step": 4509
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5491622504324187e-06,
+      "loss": 0.4886,
+      "step": 4510
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.547080085398279e-06,
+      "loss": 0.4653,
+      "step": 4511
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5449981955254233e-06,
+      "loss": 0.4672,
+      "step": 4512
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.542916581208131e-06,
+      "loss": 0.4545,
+      "step": 4513
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.540835242840635e-06,
+      "loss": 0.4537,
+      "step": 4514
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5387541808171027e-06,
+      "loss": 0.4844,
+      "step": 4515
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5366733955316577e-06,
+      "loss": 0.438,
+      "step": 4516
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.534592887378374e-06,
+      "loss": 0.5008,
+      "step": 4517
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.532512656751263e-06,
+      "loss": 0.4757,
+      "step": 4518
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5304327040442902e-06,
+      "loss": 0.4954,
+      "step": 4519
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5283530296513723e-06,
+      "loss": 0.468,
+      "step": 4520
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5262736339663634e-06,
+      "loss": 0.4608,
+      "step": 4521
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5241945173830698e-06,
+      "loss": 0.478,
+      "step": 4522
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5221156802952505e-06,
+      "loss": 0.4665,
+      "step": 4523
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5200371230966e-06,
+      "loss": 0.4948,
+      "step": 4524
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5179588461807675e-06,
+      "loss": 0.4727,
+      "step": 4525
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5158808499413517e-06,
+      "loss": 0.5003,
+      "step": 4526
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5138031347718883e-06,
+      "loss": 0.4731,
+      "step": 4527
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.511725701065869e-06,
+      "loss": 0.481,
+      "step": 4528
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5096485492167266e-06,
+      "loss": 0.4764,
+      "step": 4529
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.507571679617844e-06,
+      "loss": 0.4725,
+      "step": 4530
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5054950926625485e-06,
+      "loss": 0.4814,
+      "step": 4531
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5034187887441152e-06,
+      "loss": 0.4566,
+      "step": 4532
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5013427682557633e-06,
+      "loss": 0.4906,
+      "step": 4533
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.499267031590663e-06,
+      "loss": 0.514,
+      "step": 4534
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4971915791419252e-06,
+      "loss": 0.4512,
+      "step": 4535
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4951164113026103e-06,
+      "loss": 0.4717,
+      "step": 4536
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4930415284657247e-06,
+      "loss": 0.4809,
+      "step": 4537
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.490966931024219e-06,
+      "loss": 0.4612,
+      "step": 4538
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4888926193709917e-06,
+      "loss": 0.4876,
+      "step": 4539
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.486818593898887e-06,
+      "loss": 0.4647,
+      "step": 4540
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4847448550006925e-06,
+      "loss": 0.4769,
+      "step": 4541
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.482671403069144e-06,
+      "loss": 0.4623,
+      "step": 4542
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4805982384969232e-06,
+      "loss": 0.4626,
+      "step": 4543
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4785253616766555e-06,
+      "loss": 0.4997,
+      "step": 4544
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4764527730009124e-06,
+      "loss": 0.4844,
+      "step": 4545
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.474380472862213e-06,
+      "loss": 0.468,
+      "step": 4546
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4723084616530182e-06,
+      "loss": 0.4771,
+      "step": 4547
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4702367397657366e-06,
+      "loss": 0.4464,
+      "step": 4548
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.468165307592721e-06,
+      "loss": 0.4707,
+      "step": 4549
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4660941655262697e-06,
+      "loss": 0.4865,
+      "step": 4550
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.464023313958626e-06,
+      "loss": 0.4568,
+      "step": 4551
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.461952753281979e-06,
+      "loss": 0.471,
+      "step": 4552
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4598824838884603e-06,
+      "loss": 0.4694,
+      "step": 4553
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.45781250617015e-06,
+      "loss": 0.475,
+      "step": 4554
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.455742820519069e-06,
+      "loss": 0.4646,
+      "step": 4555
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.453673427327185e-06,
+      "loss": 0.4924,
+      "step": 4556
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4516043269864117e-06,
+      "loss": 0.4867,
+      "step": 4557
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.449535519888604e-06,
+      "loss": 0.486,
+      "step": 4558
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4474670064255633e-06,
+      "loss": 0.5068,
+      "step": 4559
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.445398786989038e-06,
+      "loss": 0.483,
+      "step": 4560
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4433308619707118e-06,
+      "loss": 0.4675,
+      "step": 4561
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.441263231762224e-06,
+      "loss": 0.4787,
+      "step": 4562
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4391958967551527e-06,
+      "loss": 0.4693,
+      "step": 4563
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4371288573410155e-06,
+      "loss": 0.4745,
+      "step": 4564
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4350621139112826e-06,
+      "loss": 0.4587,
+      "step": 4565
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.432995666857366e-06,
+      "loss": 0.4887,
+      "step": 4566
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4309295165706147e-06,
+      "loss": 0.4588,
+      "step": 4567
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.42886366344233e-06,
+      "loss": 0.4828,
+      "step": 4568
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4267981078637546e-06,
+      "loss": 0.4774,
+      "step": 4569
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4247328502260697e-06,
+      "loss": 0.4649,
+      "step": 4570
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.422667890920407e-06,
+      "loss": 0.4631,
+      "step": 4571
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.420603230337841e-06,
+      "loss": 0.4652,
+      "step": 4572
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4185388688693823e-06,
+      "loss": 0.4635,
+      "step": 4573
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.416474806905993e-06,
+      "loss": 0.4706,
+      "step": 4574
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.414411044838577e-06,
+      "loss": 0.4417,
+      "step": 4575
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4123475830579763e-06,
+      "loss": 0.5018,
+      "step": 4576
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4102844219549817e-06,
+      "loss": 0.4919,
+      "step": 4577
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4082215619203276e-06,
+      "loss": 0.4802,
+      "step": 4578
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4061590033446833e-06,
+      "loss": 0.4675,
+      "step": 4579
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.404096746618672e-06,
+      "loss": 0.5005,
+      "step": 4580
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.40203479213285e-06,
+      "loss": 0.4917,
+      "step": 4581
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.39997314027772e-06,
+      "loss": 0.4757,
+      "step": 4582
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3979117914437354e-06,
+      "loss": 0.4729,
+      "step": 4583
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3958507460212764e-06,
+      "loss": 0.4121,
+      "step": 4584
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3937900044006777e-06,
+      "loss": 0.4492,
+      "step": 4585
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3917295669722138e-06,
+      "loss": 0.491,
+      "step": 4586
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3896694341260984e-06,
+      "loss": 0.4853,
+      "step": 4587
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.387609606252491e-06,
+      "loss": 0.4854,
+      "step": 4588
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3855500837414936e-06,
+      "loss": 0.4569,
+      "step": 4589
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3834908669831456e-06,
+      "loss": 0.5035,
+      "step": 4590
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3814319563674347e-06,
+      "loss": 0.4821,
+      "step": 4591
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3793733522842877e-06,
+      "loss": 0.4703,
+      "step": 4592
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.377315055123571e-06,
+      "loss": 0.4564,
+      "step": 4593
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3752570652750978e-06,
+      "loss": 0.4782,
+      "step": 4594
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3731993831286196e-06,
+      "loss": 0.4645,
+      "step": 4595
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.37114200907383e-06,
+      "loss": 0.4746,
+      "step": 4596
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.369084943500366e-06,
+      "loss": 0.4838,
+      "step": 4597
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3670281867978043e-06,
+      "loss": 0.4881,
+      "step": 4598
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3649717393556637e-06,
+      "loss": 0.4914,
+      "step": 4599
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3629156015634046e-06,
+      "loss": 0.4883,
+      "step": 4600
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3608597738104305e-06,
+      "loss": 0.4967,
+      "step": 4601
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3588042564860814e-06,
+      "loss": 0.4792,
+      "step": 4602
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.356749049979643e-06,
+      "loss": 0.4967,
+      "step": 4603
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.354694154680342e-06,
+      "loss": 0.5018,
+      "step": 4604
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3526395709773423e-06,
+      "loss": 0.4854,
+      "step": 4605
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.350585299259753e-06,
+      "loss": 0.4475,
+      "step": 4606
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.348531339916622e-06,
+      "loss": 0.4962,
+      "step": 4607
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.346477693336938e-06,
+      "loss": 0.4628,
+      "step": 4608
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.344424359909632e-06,
+      "loss": 0.4749,
+      "step": 4609
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.342371340023573e-06,
+      "loss": 0.4532,
+      "step": 4610
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.340318634067573e-06,
+      "loss": 0.4883,
+      "step": 4611
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.338266242430386e-06,
+      "loss": 0.4814,
+      "step": 4612
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3362141655006998e-06,
+      "loss": 0.4468,
+      "step": 4613
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.33416240366715e-06,
+      "loss": 0.4679,
+      "step": 4614
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3321109573183096e-06,
+      "loss": 0.4769,
+      "step": 4615
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.330059826842691e-06,
+      "loss": 0.4485,
+      "step": 4616
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3280090126287477e-06,
+      "loss": 0.4759,
+      "step": 4617
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3259585150648742e-06,
+      "loss": 0.4729,
+      "step": 4618
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.323908334539403e-06,
+      "loss": 0.4608,
+      "step": 4619
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3218584714406068e-06,
+      "loss": 0.4685,
+      "step": 4620
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3198089261567015e-06,
+      "loss": 0.4805,
+      "step": 4621
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3177596990758377e-06,
+      "loss": 0.4637,
+      "step": 4622
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3157107905861098e-06,
+      "loss": 0.4832,
+      "step": 4623
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3136622010755517e-06,
+      "loss": 0.4546,
+      "step": 4624
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3116139309321315e-06,
+      "loss": 0.4466,
+      "step": 4625
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.309565980543764e-06,
+      "loss": 0.4904,
+      "step": 4626
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3075183502983017e-06,
+      "loss": 0.4784,
+      "step": 4627
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.305471040583531e-06,
+      "loss": 0.4775,
+      "step": 4628
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3034240517871847e-06,
+      "loss": 0.489,
+      "step": 4629
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3013773842969316e-06,
+      "loss": 0.4586,
+      "step": 4630
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2993310385003784e-06,
+      "loss": 0.4613,
+      "step": 4631
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2972850147850753e-06,
+      "loss": 0.4876,
+      "step": 4632
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2952393135385037e-06,
+      "loss": 0.5115,
+      "step": 4633
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2931939351480925e-06,
+      "loss": 0.4921,
+      "step": 4634
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.291148880001207e-06,
+      "loss": 0.4584,
+      "step": 4635
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.289104148485146e-06,
+      "loss": 0.4867,
+      "step": 4636
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.287059740987152e-06,
+      "loss": 0.4693,
+      "step": 4637
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2850156578944094e-06,
+      "loss": 0.4699,
+      "step": 4638
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2829718995940305e-06,
+      "loss": 0.4518,
+      "step": 4639
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2809284664730763e-06,
+      "loss": 0.4608,
+      "step": 4640
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2788853589185442e-06,
+      "loss": 0.4571,
+      "step": 4641
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.276842577317364e-06,
+      "loss": 0.4898,
+      "step": 4642
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.274800122056409e-06,
+      "loss": 0.4672,
+      "step": 4643
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2727579935224933e-06,
+      "loss": 0.4396,
+      "step": 4644
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.270716192102361e-06,
+      "loss": 0.4645,
+      "step": 4645
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2686747181826996e-06,
+      "loss": 0.4844,
+      "step": 4646
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2666335721501374e-06,
+      "loss": 0.4648,
+      "step": 4647
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2645927543912316e-06,
+      "loss": 0.4977,
+      "step": 4648
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2625522652924846e-06,
+      "loss": 0.4559,
+      "step": 4649
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2605121052403377e-06,
+      "loss": 0.5029,
+      "step": 4650
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2584722746211617e-06,
+      "loss": 0.4837,
+      "step": 4651
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.256432773821271e-06,
+      "loss": 0.4852,
+      "step": 4652
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.254393603226921e-06,
+      "loss": 0.4544,
+      "step": 4653
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2523547632242946e-06,
+      "loss": 0.4723,
+      "step": 4654
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2503162541995194e-06,
+      "loss": 0.4663,
+      "step": 4655
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2482780765386613e-06,
+      "loss": 0.4741,
+      "step": 4656
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2462402306277158e-06,
+      "loss": 0.4853,
+      "step": 4657
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2442027168526245e-06,
+      "loss": 0.4577,
+      "step": 4658
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2421655355992585e-06,
+      "loss": 0.4856,
+      "step": 4659
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.240128687253432e-06,
+      "loss": 0.4487,
+      "step": 4660
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.238092172200894e-06,
+      "loss": 0.4839,
+      "step": 4661
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.236055990827328e-06,
+      "loss": 0.4733,
+      "step": 4662
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.234020143518356e-06,
+      "loss": 0.475,
+      "step": 4663
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2319846306595406e-06,
+      "loss": 0.503,
+      "step": 4664
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.229949452636373e-06,
+      "loss": 0.4851,
+      "step": 4665
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2279146098342872e-06,
+      "loss": 0.4695,
+      "step": 4666
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2258801026386543e-06,
+      "loss": 0.4804,
+      "step": 4667
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.223845931434776e-06,
+      "loss": 0.4864,
+      "step": 4668
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.221812096607895e-06,
+      "loss": 0.4546,
+      "step": 4669
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2197785985431905e-06,
+      "loss": 0.4987,
+      "step": 4670
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.217745437625775e-06,
+      "loss": 0.4726,
+      "step": 4671
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.215712614240699e-06,
+      "loss": 0.4679,
+      "step": 4672
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2136801287729503e-06,
+      "loss": 0.4817,
+      "step": 4673
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.211647981607449e-06,
+      "loss": 0.4663,
+      "step": 4674
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2096161731290546e-06,
+      "loss": 0.4801,
+      "step": 4675
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.207584703722562e-06,
+      "loss": 0.4309,
+      "step": 4676
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2055535737726993e-06,
+      "loss": 0.4826,
+      "step": 4677
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.203522783664133e-06,
+      "loss": 0.524,
+      "step": 4678
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.201492333781465e-06,
+      "loss": 0.4934,
+      "step": 4679
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.199462224509231e-06,
+      "loss": 0.4763,
+      "step": 4680
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1974324562319037e-06,
+      "loss": 0.4763,
+      "step": 4681
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1954030293338917e-06,
+      "loss": 0.5018,
+      "step": 4682
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.193373944199536e-06,
+      "loss": 0.4607,
+      "step": 4683
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1913452012131175e-06,
+      "loss": 0.4403,
+      "step": 4684
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.189316800758848e-06,
+      "loss": 0.4798,
+      "step": 4685
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.187288743220876e-06,
+      "loss": 0.4711,
+      "step": 4686
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1852610289832874e-06,
+      "loss": 0.4528,
+      "step": 4687
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.183233658430098e-06,
+      "loss": 0.4267,
+      "step": 4688
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1812066319452627e-06,
+      "loss": 0.4788,
+      "step": 4689
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.179179949912673e-06,
+      "loss": 0.445,
+      "step": 4690
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.177153612716146e-06,
+      "loss": 0.4427,
+      "step": 4691
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1751276207394442e-06,
+      "loss": 0.4772,
+      "step": 4692
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1731019743662603e-06,
+      "loss": 0.4282,
+      "step": 4693
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1710766739802174e-06,
+      "loss": 0.4625,
+      "step": 4694
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.169051719964881e-06,
+      "loss": 0.4626,
+      "step": 4695
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.167027112703747e-06,
+      "loss": 0.5125,
+      "step": 4696
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1650028525802425e-06,
+      "loss": 0.4985,
+      "step": 4697
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1629789399777343e-06,
+      "loss": 0.4952,
+      "step": 4698
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.160955375279523e-06,
+      "loss": 0.4969,
+      "step": 4699
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.158932158868837e-06,
+      "loss": 0.4574,
+      "step": 4700
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1569092911288457e-06,
+      "loss": 0.4521,
+      "step": 4701
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1548867724426526e-06,
+      "loss": 0.4567,
+      "step": 4702
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1528646031932865e-06,
+      "loss": 0.4632,
+      "step": 4703
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.15084278376372e-06,
+      "loss": 0.4585,
+      "step": 4704
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1488213145368574e-06,
+      "loss": 0.4631,
+      "step": 4705
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1468001958955286e-06,
+      "loss": 0.4789,
+      "step": 4706
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.14477942822251e-06,
+      "loss": 0.4673,
+      "step": 4707
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1427590119004984e-06,
+      "loss": 0.4664,
+      "step": 4708
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.140738947312133e-06,
+      "loss": 0.4846,
+      "step": 4709
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1387192348399876e-06,
+      "loss": 0.5049,
+      "step": 4710
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.136699874866559e-06,
+      "loss": 0.4818,
+      "step": 4711
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1346808677742868e-06,
+      "loss": 0.4785,
+      "step": 4712
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.132662213945543e-06,
+      "loss": 0.4901,
+      "step": 4713
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.130643913762627e-06,
+      "loss": 0.4626,
+      "step": 4714
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1286259676077756e-06,
+      "loss": 0.477,
+      "step": 4715
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1266083758631594e-06,
+      "loss": 0.4846,
+      "step": 4716
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1245911389108763e-06,
+      "loss": 0.4721,
+      "step": 4717
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1225742571329643e-06,
+      "loss": 0.483,
+      "step": 4718
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1205577309113892e-06,
+      "loss": 0.499,
+      "step": 4719
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1185415606280506e-06,
+      "loss": 0.4654,
+      "step": 4720
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1165257466647813e-06,
+      "loss": 0.451,
+      "step": 4721
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1145102894033473e-06,
+      "loss": 0.421,
+      "step": 4722
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1124951892254447e-06,
+      "loss": 0.4655,
+      "step": 4723
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1104804465127032e-06,
+      "loss": 0.4886,
+      "step": 4724
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1084660616466868e-06,
+      "loss": 0.5001,
+      "step": 4725
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.106452035008887e-06,
+      "loss": 0.4947,
+      "step": 4726
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.104438366980732e-06,
+      "loss": 0.4472,
+      "step": 4727
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1024250579435823e-06,
+      "loss": 0.4722,
+      "step": 4728
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.100412108278724e-06,
+      "loss": 0.456,
+      "step": 4729
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0983995183673828e-06,
+      "loss": 0.4999,
+      "step": 4730
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0963872885907133e-06,
+      "loss": 0.4957,
+      "step": 4731
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0943754193297993e-06,
+      "loss": 0.4531,
+      "step": 4732
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.092363910965661e-06,
+      "loss": 0.4725,
+      "step": 4733
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.090352763879247e-06,
+      "loss": 0.4838,
+      "step": 4734
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.088341978451438e-06,
+      "loss": 0.4691,
+      "step": 4735
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0863315550630487e-06,
+      "loss": 0.4408,
+      "step": 4736
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0843214940948208e-06,
+      "loss": 0.4626,
+      "step": 4737
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.08231179592743e-06,
+      "loss": 0.4881,
+      "step": 4738
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0803024609414856e-06,
+      "loss": 0.4641,
+      "step": 4739
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.078293489517522e-06,
+      "loss": 0.4648,
+      "step": 4740
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0762848820360113e-06,
+      "loss": 0.4856,
+      "step": 4741
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0742766388773526e-06,
+      "loss": 0.4789,
+      "step": 4742
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0722687604218757e-06,
+      "loss": 0.4752,
+      "step": 4743
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.070261247049845e-06,
+      "loss": 0.4735,
+      "step": 4744
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.068254099141453e-06,
+      "loss": 0.4778,
+      "step": 4745
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.066247317076822e-06,
+      "loss": 0.4642,
+      "step": 4746
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0642409012360074e-06,
+      "loss": 0.4492,
+      "step": 4747
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0622348519989957e-06,
+      "loss": 0.4591,
+      "step": 4748
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0602291697457e-06,
+      "loss": 0.4661,
+      "step": 4749
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0582238548559684e-06,
+      "loss": 0.4826,
+      "step": 4750
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.056218907709577e-06,
+      "loss": 0.4254,
+      "step": 4751
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0542143286862324e-06,
+      "loss": 0.4784,
+      "step": 4752
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0522101181655715e-06,
+      "loss": 0.4697,
+      "step": 4753
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0502062765271644e-06,
+      "loss": 0.4641,
+      "step": 4754
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0482028041505054e-06,
+      "loss": 0.4678,
+      "step": 4755
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.046199701415023e-06,
+      "loss": 0.4738,
+      "step": 4756
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.044196968700077e-06,
+      "loss": 0.4493,
+      "step": 4757
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0421946063849515e-06,
+      "loss": 0.4652,
+      "step": 4758
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.040192614848869e-06,
+      "loss": 0.5046,
+      "step": 4759
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0381909944709708e-06,
+      "loss": 0.4858,
+      "step": 4760
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0361897456303367e-06,
+      "loss": 0.4535,
+      "step": 4761
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.034188868705976e-06,
+      "loss": 0.4814,
+      "step": 4762
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.032188364076819e-06,
+      "loss": 0.4602,
+      "step": 4763
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0301882321217346e-06,
+      "loss": 0.4618,
+      "step": 4764
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0281884732195205e-06,
+      "loss": 0.4806,
+      "step": 4765
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0261890877488964e-06,
+      "loss": 0.4868,
+      "step": 4766
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0241900760885164e-06,
+      "loss": 0.4824,
+      "step": 4767
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0221914386169683e-06,
+      "loss": 0.5062,
+      "step": 4768
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0201931757127577e-06,
+      "loss": 0.5189,
+      "step": 4769
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0181952877543287e-06,
+      "loss": 0.4598,
+      "step": 4770
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.016197775120054e-06,
+      "loss": 0.4764,
+      "step": 4771
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0142006381882275e-06,
+      "loss": 0.4526,
+      "step": 4772
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0122038773370794e-06,
+      "loss": 0.4744,
+      "step": 4773
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0102074929447688e-06,
+      "loss": 0.4614,
+      "step": 4774
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0082114853893767e-06,
+      "loss": 0.442,
+      "step": 4775
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0062158550489184e-06,
+      "loss": 0.4779,
+      "step": 4776
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.00422060230134e-06,
+      "loss": 0.4379,
+      "step": 4777
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.002225727524507e-06,
+      "loss": 0.5106,
+      "step": 4778
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.00023123109622e-06,
+      "loss": 0.4972,
+      "step": 4779
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.998237113394212e-06,
+      "loss": 0.4772,
+      "step": 4780
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.996243374796133e-06,
+      "loss": 0.4685,
+      "step": 4781
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9942500156795677e-06,
+      "loss": 0.503,
+      "step": 4782
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.992257036422034e-06,
+      "loss": 0.4635,
+      "step": 4783
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.990264437400966e-06,
+      "loss": 0.4818,
+      "step": 4784
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.988272218993736e-06,
+      "loss": 0.468,
+      "step": 4785
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9862803815776388e-06,
+      "loss": 0.4983,
+      "step": 4786
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.984288925529898e-06,
+      "loss": 0.4963,
+      "step": 4787
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.982297851227667e-06,
+      "loss": 0.4811,
+      "step": 4788
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.980307159048024e-06,
+      "loss": 0.4795,
+      "step": 4789
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9783168493679777e-06,
+      "loss": 0.4806,
+      "step": 4790
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.976326922564463e-06,
+      "loss": 0.469,
+      "step": 4791
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.974337379014341e-06,
+      "loss": 0.4709,
+      "step": 4792
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.972348219094402e-06,
+      "loss": 0.4935,
+      "step": 4793
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.970359443181364e-06,
+      "loss": 0.4845,
+      "step": 4794
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9683710516518706e-06,
+      "loss": 0.4756,
+      "step": 4795
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9663830448824927e-06,
+      "loss": 0.4778,
+      "step": 4796
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9643954232497307e-06,
+      "loss": 0.4521,
+      "step": 4797
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9624081871300093e-06,
+      "loss": 0.4918,
+      "step": 4798
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9604213368996802e-06,
+      "loss": 0.4964,
+      "step": 4799
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9584348729350263e-06,
+      "loss": 0.4604,
+      "step": 4800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9564487956122514e-06,
+      "loss": 0.4872,
+      "step": 4801
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.954463105307489e-06,
+      "loss": 0.4755,
+      "step": 4802
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.952477802396801e-06,
+      "loss": 0.4766,
+      "step": 4803
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9504928872561717e-06,
+      "loss": 0.496,
+      "step": 4804
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.948508360261515e-06,
+      "loss": 0.4854,
+      "step": 4805
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.946524221788673e-06,
+      "loss": 0.5181,
+      "step": 4806
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9445404722134085e-06,
+      "loss": 0.5047,
+      "step": 4807
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.942557111911415e-06,
+      "loss": 0.4786,
+      "step": 4808
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9405741412583132e-06,
+      "loss": 0.5068,
+      "step": 4809
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9385915606296445e-06,
+      "loss": 0.4552,
+      "step": 4810
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9366093704008836e-06,
+      "loss": 0.4831,
+      "step": 4811
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9346275709474255e-06,
+      "loss": 0.5078,
+      "step": 4812
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9326461626445934e-06,
+      "loss": 0.4705,
+      "step": 4813
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9306651458676377e-06,
+      "loss": 0.4921,
+      "step": 4814
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9286845209917313e-06,
+      "loss": 0.4447,
+      "step": 4815
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9267042883919774e-06,
+      "loss": 0.45,
+      "step": 4816
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9247244484434013e-06,
+      "loss": 0.483,
+      "step": 4817
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9227450015209537e-06,
+      "loss": 0.4977,
+      "step": 4818
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.920765947999513e-06,
+      "loss": 0.4699,
+      "step": 4819
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9187872882538854e-06,
+      "loss": 0.4757,
+      "step": 4820
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9168090226587944e-06,
+      "loss": 0.4714,
+      "step": 4821
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.914831151588896e-06,
+      "loss": 0.4917,
+      "step": 4822
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9128536754187704e-06,
+      "loss": 0.4548,
+      "step": 4823
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9108765945229213e-06,
+      "loss": 0.5199,
+      "step": 4824
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9088999092757776e-06,
+      "loss": 0.4459,
+      "step": 4825
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.906923620051696e-06,
+      "loss": 0.4904,
+      "step": 4826
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9049477272249533e-06,
+      "loss": 0.4641,
+      "step": 4827
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.902972231169755e-06,
+      "loss": 0.4285,
+      "step": 4828
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.900997132260231e-06,
+      "loss": 0.4968,
+      "step": 4829
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.899022430870435e-06,
+      "loss": 0.4543,
+      "step": 4830
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.897048127374347e-06,
+      "loss": 0.4843,
+      "step": 4831
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8950742221458704e-06,
+      "loss": 0.4429,
+      "step": 4832
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.893100715558831e-06,
+      "loss": 0.4765,
+      "step": 4833
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.891127607986983e-06,
+      "loss": 0.5,
+      "step": 4834
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8891548998040035e-06,
+      "loss": 0.5003,
+      "step": 4835
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.887182591383494e-06,
+      "loss": 0.4963,
+      "step": 4836
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.885210683098982e-06,
+      "loss": 0.4345,
+      "step": 4837
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.883239175323913e-06,
+      "loss": 0.4359,
+      "step": 4838
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.881268068431664e-06,
+      "loss": 0.4918,
+      "step": 4839
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.879297362795532e-06,
+      "loss": 0.4502,
+      "step": 4840
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8773270587887402e-06,
+      "loss": 0.473,
+      "step": 4841
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8753571567844347e-06,
+      "loss": 0.4716,
+      "step": 4842
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8733876571556867e-06,
+      "loss": 0.4998,
+      "step": 4843
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.871418560275486e-06,
+      "loss": 0.4503,
+      "step": 4844
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.869449866516753e-06,
+      "loss": 0.4953,
+      "step": 4845
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.867481576252328e-06,
+      "loss": 0.4542,
+      "step": 4846
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8655136898549763e-06,
+      "loss": 0.4653,
+      "step": 4847
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8635462076973853e-06,
+      "loss": 0.4562,
+      "step": 4848
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8615791301521688e-06,
+      "loss": 0.4746,
+      "step": 4849
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8596124575918586e-06,
+      "loss": 0.4811,
+      "step": 4850
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8576461903889148e-06,
+      "loss": 0.4767,
+      "step": 4851
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8556803289157187e-06,
+      "loss": 0.4756,
+      "step": 4852
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8537148735445752e-06,
+      "loss": 0.4811,
+      "step": 4853
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8517498246477117e-06,
+      "loss": 0.4752,
+      "step": 4854
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8497851825972815e-06,
+      "loss": 0.4678,
+      "step": 4855
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.847820947765355e-06,
+      "loss": 0.4707,
+      "step": 4856
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8458571205239293e-06,
+      "loss": 0.4775,
+      "step": 4857
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8438937012449275e-06,
+      "loss": 0.4791,
+      "step": 4858
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8419306903001855e-06,
+      "loss": 0.4758,
+      "step": 4859
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8399680880614728e-06,
+      "loss": 0.4682,
+      "step": 4860
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8380058949004784e-06,
+      "loss": 0.4814,
+      "step": 4861
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8360441111888084e-06,
+      "loss": 0.4769,
+      "step": 4862
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8340827372979973e-06,
+      "loss": 0.4823,
+      "step": 4863
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8321217735994986e-06,
+      "loss": 0.4538,
+      "step": 4864
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8301612204646877e-06,
+      "loss": 0.4725,
+      "step": 4865
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8282010782648704e-06,
+      "loss": 0.4843,
+      "step": 4866
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.826241347371263e-06,
+      "loss": 0.478,
+      "step": 4867
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8242820281550102e-06,
+      "loss": 0.4687,
+      "step": 4868
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8223231209871806e-06,
+      "loss": 0.4931,
+      "step": 4869
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8203646262387573e-06,
+      "loss": 0.5087,
+      "step": 4870
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8184065442806504e-06,
+      "loss": 0.4778,
+      "step": 4871
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8164488754836967e-06,
+      "loss": 0.4658,
+      "step": 4872
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8144916202186433e-06,
+      "loss": 0.5046,
+      "step": 4873
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.812534778856168e-06,
+      "loss": 0.468,
+      "step": 4874
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8105783517668682e-06,
+      "loss": 0.4409,
+      "step": 4875
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.808622339321259e-06,
+      "loss": 0.4704,
+      "step": 4876
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.806666741889779e-06,
+      "loss": 0.4845,
+      "step": 4877
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8047115598427954e-06,
+      "loss": 0.4529,
+      "step": 4878
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8027567935505845e-06,
+      "loss": 0.4727,
+      "step": 4879
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8008024433833514e-06,
+      "loss": 0.468,
+      "step": 4880
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7988485097112234e-06,
+      "loss": 0.4861,
+      "step": 4881
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7968949929042414e-06,
+      "loss": 0.467,
+      "step": 4882
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7949418933323757e-06,
+      "loss": 0.496,
+      "step": 4883
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.792989211365512e-06,
+      "loss": 0.4524,
+      "step": 4884
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.791036947373461e-06,
+      "loss": 0.4565,
+      "step": 4885
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.789085101725952e-06,
+      "loss": 0.4757,
+      "step": 4886
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.787133674792636e-06,
+      "loss": 0.5051,
+      "step": 4887
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7851826669430814e-06,
+      "loss": 0.5012,
+      "step": 4888
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7832320785467814e-06,
+      "loss": 0.4847,
+      "step": 4889
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.781281909973148e-06,
+      "loss": 0.4632,
+      "step": 4890
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7793321615915143e-06,
+      "loss": 0.4823,
+      "step": 4891
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.777382833771135e-06,
+      "loss": 0.4768,
+      "step": 4892
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.775433926881181e-06,
+      "loss": 0.4608,
+      "step": 4893
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7734854412907464e-06,
+      "loss": 0.4882,
+      "step": 4894
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.771537377368846e-06,
+      "loss": 0.4698,
+      "step": 4895
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7695897354844138e-06,
+      "loss": 0.4534,
+      "step": 4896
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.767642516006304e-06,
+      "loss": 0.498,
+      "step": 4897
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.765695719303293e-06,
+      "loss": 0.489,
+      "step": 4898
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.763749345744071e-06,
+      "loss": 0.4597,
+      "step": 4899
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.761803395697254e-06,
+      "loss": 0.4721,
+      "step": 4900
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7598578695313755e-06,
+      "loss": 0.4999,
+      "step": 4901
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7579127676148893e-06,
+      "loss": 0.4578,
+      "step": 4902
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.755968090316168e-06,
+      "loss": 0.4912,
+      "step": 4903
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7540238380035077e-06,
+      "loss": 0.4766,
+      "step": 4904
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.752080011045115e-06,
+      "loss": 0.4872,
+      "step": 4905
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7501366098091255e-06,
+      "loss": 0.4877,
+      "step": 4906
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.748193634663589e-06,
+      "loss": 0.4615,
+      "step": 4907
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.746251085976476e-06,
+      "loss": 0.4861,
+      "step": 4908
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7443089641156764e-06,
+      "loss": 0.4534,
+      "step": 4909
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.742367269449001e-06,
+      "loss": 0.4641,
+      "step": 4910
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.740426002344174e-06,
+      "loss": 0.4345,
+      "step": 4911
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7384851631688435e-06,
+      "loss": 0.487,
+      "step": 4912
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7365447522905765e-06,
+      "loss": 0.4858,
+      "step": 4913
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7346047700768574e-06,
+      "loss": 0.4845,
+      "step": 4914
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7326652168950908e-06,
+      "loss": 0.4677,
+      "step": 4915
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7307260931125966e-06,
+      "loss": 0.4464,
+      "step": 4916
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.728787399096616e-06,
+      "loss": 0.494,
+      "step": 4917
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7268491352143132e-06,
+      "loss": 0.4577,
+      "step": 4918
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7249113018327577e-06,
+      "loss": 0.4687,
+      "step": 4919
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7229738993189544e-06,
+      "loss": 0.4736,
+      "step": 4920
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7210369280398165e-06,
+      "loss": 0.4843,
+      "step": 4921
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.719100388362174e-06,
+      "loss": 0.4765,
+      "step": 4922
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.717164280652781e-06,
+      "loss": 0.4958,
+      "step": 4923
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7152286052783094e-06,
+      "loss": 0.4423,
+      "step": 4924
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7132933626053403e-06,
+      "loss": 0.4913,
+      "step": 4925
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.711358553000386e-06,
+      "loss": 0.4684,
+      "step": 4926
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7094241768298702e-06,
+      "loss": 0.4795,
+      "step": 4927
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.707490234460132e-06,
+      "loss": 0.4959,
+      "step": 4928
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7055567262574323e-06,
+      "loss": 0.4652,
+      "step": 4929
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.70362365258795e-06,
+      "loss": 0.4987,
+      "step": 4930
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7016910138177754e-06,
+      "loss": 0.4682,
+      "step": 4931
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6997588103129273e-06,
+      "loss": 0.4813,
+      "step": 4932
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6978270424393355e-06,
+      "loss": 0.4781,
+      "step": 4933
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6958957105628443e-06,
+      "loss": 0.5024,
+      "step": 4934
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6939648150492206e-06,
+      "loss": 0.4996,
+      "step": 4935
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6920343562641493e-06,
+      "loss": 0.4842,
+      "step": 4936
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.690104334573227e-06,
+      "loss": 0.4533,
+      "step": 4937
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.688174750341971e-06,
+      "loss": 0.5215,
+      "step": 4938
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.68624560393582e-06,
+      "loss": 0.4634,
+      "step": 4939
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6843168957201214e-06,
+      "loss": 0.4634,
+      "step": 4940
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.682388626060146e-06,
+      "loss": 0.483,
+      "step": 4941
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.680460795321077e-06,
+      "loss": 0.4697,
+      "step": 4942
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6785334038680178e-06,
+      "loss": 0.4476,
+      "step": 4943
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6766064520659873e-06,
+      "loss": 0.4472,
+      "step": 4944
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6746799402799207e-06,
+      "loss": 0.4576,
+      "step": 4945
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6727538688746725e-06,
+      "loss": 0.4636,
+      "step": 4946
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6708282382150118e-06,
+      "loss": 0.4348,
+      "step": 4947
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6689030486656214e-06,
+      "loss": 0.4653,
+      "step": 4948
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.666978300591105e-06,
+      "loss": 0.4713,
+      "step": 4949
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.665053994355981e-06,
+      "loss": 0.4658,
+      "step": 4950
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6631301303246845e-06,
+      "loss": 0.481,
+      "step": 4951
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.661206708861566e-06,
+      "loss": 0.4985,
+      "step": 4952
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.659283730330895e-06,
+      "loss": 0.4563,
+      "step": 4953
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6573611950968516e-06,
+      "loss": 0.5041,
+      "step": 4954
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.655439103523536e-06,
+      "loss": 0.4791,
+      "step": 4955
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6535174559749646e-06,
+      "loss": 0.4487,
+      "step": 4956
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6515962528150673e-06,
+      "loss": 0.4546,
+      "step": 4957
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.649675494407693e-06,
+      "loss": 0.4529,
+      "step": 4958
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6477551811166058e-06,
+      "loss": 0.4857,
+      "step": 4959
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6458353133054797e-06,
+      "loss": 0.485,
+      "step": 4960
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6439158913379115e-06,
+      "loss": 0.4545,
+      "step": 4961
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.641996915577411e-06,
+      "loss": 0.4532,
+      "step": 4962
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6400783863874037e-06,
+      "loss": 0.4811,
+      "step": 4963
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.638160304131232e-06,
+      "loss": 0.4346,
+      "step": 4964
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6362426691721475e-06,
+      "loss": 0.5144,
+      "step": 4965
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6343254818733243e-06,
+      "loss": 0.4563,
+      "step": 4966
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6324087425978496e-06,
+      "loss": 0.4606,
+      "step": 4967
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6304924517087238e-06,
+      "loss": 0.475,
+      "step": 4968
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.628576609568864e-06,
+      "loss": 0.4599,
+      "step": 4969
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.626661216541104e-06,
+      "loss": 0.4482,
+      "step": 4970
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.624746272988187e-06,
+      "loss": 0.482,
+      "step": 4971
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6228317792727762e-06,
+      "loss": 0.4924,
+      "step": 4972
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.620917735757449e-06,
+      "loss": 0.485,
+      "step": 4973
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.619004142804695e-06,
+      "loss": 0.4602,
+      "step": 4974
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.617091000776922e-06,
+      "loss": 0.4408,
+      "step": 4975
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6151783100364506e-06,
+      "loss": 0.485,
+      "step": 4976
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6132660709455136e-06,
+      "loss": 0.4701,
+      "step": 4977
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.611354283866261e-06,
+      "loss": 0.4652,
+      "step": 4978
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6094429491607565e-06,
+      "loss": 0.493,
+      "step": 4979
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6075320671909797e-06,
+      "loss": 0.4912,
+      "step": 4980
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.605621638318823e-06,
+      "loss": 0.4515,
+      "step": 4981
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6037116629060926e-06,
+      "loss": 0.4228,
+      "step": 4982
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.601802141314508e-06,
+      "loss": 0.4781,
+      "step": 4983
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.599893073905705e-06,
+      "loss": 0.4962,
+      "step": 4984
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.597984461041232e-06,
+      "loss": 0.5098,
+      "step": 4985
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.596076303082552e-06,
+      "loss": 0.465,
+      "step": 4986
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5941686003910426e-06,
+      "loss": 0.481,
+      "step": 4987
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5922613533279945e-06,
+      "loss": 0.4674,
+      "step": 4988
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5903545622546083e-06,
+      "loss": 0.4806,
+      "step": 4989
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.588448227532006e-06,
+      "loss": 0.4648,
+      "step": 4990
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.586542349521215e-06,
+      "loss": 0.4519,
+      "step": 4991
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5846369285831796e-06,
+      "loss": 0.4539,
+      "step": 4992
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.582731965078764e-06,
+      "loss": 0.4379,
+      "step": 4993
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5808274593687343e-06,
+      "loss": 0.4582,
+      "step": 4994
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5789234118137775e-06,
+      "loss": 0.4536,
+      "step": 4995
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5770198227744924e-06,
+      "loss": 0.4553,
+      "step": 4996
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.575116692611387e-06,
+      "loss": 0.4586,
+      "step": 4997
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.573214021684887e-06,
+      "loss": 0.4654,
+      "step": 4998
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.571311810355333e-06,
+      "loss": 0.4931,
+      "step": 4999
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5694100589829718e-06,
+      "loss": 0.5083,
+      "step": 5000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5675087679279675e-06,
+      "loss": 0.4828,
+      "step": 5001
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.565607937550399e-06,
+      "loss": 0.5049,
+      "step": 5002
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.56370756821025e-06,
+      "loss": 0.4542,
+      "step": 5003
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.561807660267423e-06,
+      "loss": 0.4625,
+      "step": 5004
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5599082140817376e-06,
+      "loss": 0.4495,
+      "step": 5005
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5580092300129144e-06,
+      "loss": 0.4817,
+      "step": 5006
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.556110708420595e-06,
+      "loss": 0.506,
+      "step": 5007
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5542126496643327e-06,
+      "loss": 0.4319,
+      "step": 5008
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5523150541035878e-06,
+      "loss": 0.4817,
+      "step": 5009
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5504179220977376e-06,
+      "loss": 0.4689,
+      "step": 5010
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.548521254006071e-06,
+      "loss": 0.4666,
+      "step": 5011
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5466250501877897e-06,
+      "loss": 0.4663,
+      "step": 5012
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.544729311002004e-06,
+      "loss": 0.4437,
+      "step": 5013
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5428340368077427e-06,
+      "loss": 0.4923,
+      "step": 5014
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.540939227963937e-06,
+      "loss": 0.4581,
+      "step": 5015
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5390448848294375e-06,
+      "loss": 0.4732,
+      "step": 5016
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.537151007763005e-06,
+      "loss": 0.4573,
+      "step": 5017
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5352575971233114e-06,
+      "loss": 0.4878,
+      "step": 5018
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5333646532689416e-06,
+      "loss": 0.515,
+      "step": 5019
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5314721765583884e-06,
+      "loss": 0.4968,
+      "step": 5020
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5295801673500584e-06,
+      "loss": 0.4875,
+      "step": 5021
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5276886260022713e-06,
+      "loss": 0.4898,
+      "step": 5022
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5257975528732563e-06,
+      "loss": 0.5119,
+      "step": 5023
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.523906948321155e-06,
+      "loss": 0.4858,
+      "step": 5024
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.52201681270402e-06,
+      "loss": 0.4555,
+      "step": 5025
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5201271463798128e-06,
+      "loss": 0.4952,
+      "step": 5026
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.518237949706408e-06,
+      "loss": 0.4364,
+      "step": 5027
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5163492230415926e-06,
+      "loss": 0.4597,
+      "step": 5028
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5144609667430625e-06,
+      "loss": 0.4641,
+      "step": 5029
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.512573181168425e-06,
+      "loss": 0.4573,
+      "step": 5030
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.510685866675201e-06,
+      "loss": 0.4634,
+      "step": 5031
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.508799023620815e-06,
+      "loss": 0.4609,
+      "step": 5032
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.506912652362609e-06,
+      "loss": 0.4627,
+      "step": 5033
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5050267532578344e-06,
+      "loss": 0.4793,
+      "step": 5034
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5031413266636507e-06,
+      "loss": 0.4938,
+      "step": 5035
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.501256372937131e-06,
+      "loss": 0.4788,
+      "step": 5036
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.499371892435258e-06,
+      "loss": 0.4475,
+      "step": 5037
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.497487885514921e-06,
+      "loss": 0.4722,
+      "step": 5038
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4956043525329243e-06,
+      "loss": 0.5056,
+      "step": 5039
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.493721293845981e-06,
+      "loss": 0.4458,
+      "step": 5040
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4918387098107144e-06,
+      "loss": 0.4693,
+      "step": 5041
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4899566007836596e-06,
+      "loss": 0.4756,
+      "step": 5042
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.488074967121255e-06,
+      "loss": 0.4572,
+      "step": 5043
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4861938091798576e-06,
+      "loss": 0.4941,
+      "step": 5044
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.484313127315731e-06,
+      "loss": 0.4504,
+      "step": 5045
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4824329218850436e-06,
+      "loss": 0.453,
+      "step": 5046
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.480553193243883e-06,
+      "loss": 0.516,
+      "step": 5047
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.478673941748242e-06,
+      "loss": 0.4707,
+      "step": 5048
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.476795167754019e-06,
+      "loss": 0.4416,
+      "step": 5049
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4749168716170276e-06,
+      "loss": 0.4856,
+      "step": 5050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4730390536929907e-06,
+      "loss": 0.4563,
+      "step": 5051
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.471161714337533e-06,
+      "loss": 0.4756,
+      "step": 5052
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4692848539062e-06,
+      "loss": 0.4742,
+      "step": 5053
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4674084727544418e-06,
+      "loss": 0.4808,
+      "step": 5054
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.465532571237613e-06,
+      "loss": 0.5048,
+      "step": 5055
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.463657149710983e-06,
+      "loss": 0.4773,
+      "step": 5056
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.46178220852973e-06,
+      "loss": 0.4763,
+      "step": 5057
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.459907748048934e-06,
+      "loss": 0.4625,
+      "step": 5058
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4580337686235968e-06,
+      "loss": 0.4417,
+      "step": 5059
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4561602706086207e-06,
+      "loss": 0.4694,
+      "step": 5060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4542872543588147e-06,
+      "loss": 0.5034,
+      "step": 5061
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4524147202289022e-06,
+      "loss": 0.4703,
+      "step": 5062
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4505426685735153e-06,
+      "loss": 0.4671,
+      "step": 5063
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.448671099747188e-06,
+      "loss": 0.4523,
+      "step": 5064
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.446800014104368e-06,
+      "loss": 0.4961,
+      "step": 5065
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4449294119994164e-06,
+      "loss": 0.4643,
+      "step": 5066
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4430592937865916e-06,
+      "loss": 0.458,
+      "step": 5067
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.44118965982007e-06,
+      "loss": 0.4637,
+      "step": 5068
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.439320510453928e-06,
+      "loss": 0.4449,
+      "step": 5069
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4374518460421564e-06,
+      "loss": 0.4775,
+      "step": 5070
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4355836669386534e-06,
+      "loss": 0.4316,
+      "step": 5071
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4337159734972226e-06,
+      "loss": 0.474,
+      "step": 5072
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4318487660715783e-06,
+      "loss": 0.4685,
+      "step": 5073
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4299820450153427e-06,
+      "loss": 0.4547,
+      "step": 5074
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.428115810682042e-06,
+      "loss": 0.4631,
+      "step": 5075
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.426250063425114e-06,
+      "loss": 0.4575,
+      "step": 5076
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.424384803597904e-06,
+      "loss": 0.4601,
+      "step": 5077
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.422520031553664e-06,
+      "loss": 0.4729,
+      "step": 5078
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.420655747645554e-06,
+      "loss": 0.4837,
+      "step": 5079
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4187919522266436e-06,
+      "loss": 0.4886,
+      "step": 5080
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.416928645649903e-06,
+      "loss": 0.4495,
+      "step": 5081
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4150658282682177e-06,
+      "loss": 0.4781,
+      "step": 5082
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4132035004343763e-06,
+      "loss": 0.4643,
+      "step": 5083
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.411341662501077e-06,
+      "loss": 0.4581,
+      "step": 5084
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4094803148209235e-06,
+      "loss": 0.4621,
+      "step": 5085
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4076194577464285e-06,
+      "loss": 0.4548,
+      "step": 5086
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4057590916300073e-06,
+      "loss": 0.4768,
+      "step": 5087
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4038992168239874e-06,
+      "loss": 0.4813,
+      "step": 5088
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.402039833680601e-06,
+      "loss": 0.4714,
+      "step": 5089
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4001809425519877e-06,
+      "loss": 0.4442,
+      "step": 5090
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.398322543790193e-06,
+      "loss": 0.4639,
+      "step": 5091
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.396464637747173e-06,
+      "loss": 0.4508,
+      "step": 5092
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.394607224774782e-06,
+      "loss": 0.4521,
+      "step": 5093
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3927503052247892e-06,
+      "loss": 0.4905,
+      "step": 5094
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3908938794488662e-06,
+      "loss": 0.4907,
+      "step": 5095
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3890379477985937e-06,
+      "loss": 0.4841,
+      "step": 5096
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3871825106254585e-06,
+      "loss": 0.4989,
+      "step": 5097
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.385327568280849e-06,
+      "loss": 0.4695,
+      "step": 5098
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.383473121116065e-06,
+      "loss": 0.4727,
+      "step": 5099
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3816191694823113e-06,
+      "loss": 0.4734,
+      "step": 5100
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3797657137306994e-06,
+      "loss": 0.4733,
+      "step": 5101
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.377912754212245e-06,
+      "loss": 0.4773,
+      "step": 5102
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.376060291277873e-06,
+      "loss": 0.476,
+      "step": 5103
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.374208325278409e-06,
+      "loss": 0.4694,
+      "step": 5104
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3723568565645882e-06,
+      "loss": 0.457,
+      "step": 5105
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.370505885487053e-06,
+      "loss": 0.4649,
+      "step": 5106
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.368655412396348e-06,
+      "loss": 0.4711,
+      "step": 5107
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.366805437642925e-06,
+      "loss": 0.4663,
+      "step": 5108
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.364955961577145e-06,
+      "loss": 0.4514,
+      "step": 5109
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.363106984549266e-06,
+      "loss": 0.4989,
+      "step": 5110
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.361258506909459e-06,
+      "loss": 0.4776,
+      "step": 5111
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.359410529007797e-06,
+      "loss": 0.5112,
+      "step": 5112
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3575630511942615e-06,
+      "loss": 0.45,
+      "step": 5113
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.355716073818735e-06,
+      "loss": 0.481,
+      "step": 5114
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3538695972310106e-06,
+      "loss": 0.5126,
+      "step": 5115
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3520236217807786e-06,
+      "loss": 0.4526,
+      "step": 5116
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3501781478176417e-06,
+      "loss": 0.4742,
+      "step": 5117
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3483331756911067e-06,
+      "loss": 0.5092,
+      "step": 5118
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3464887057505782e-06,
+      "loss": 0.4551,
+      "step": 5119
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3446447383453784e-06,
+      "loss": 0.473,
+      "step": 5120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3428012738247218e-06,
+      "loss": 0.4659,
+      "step": 5121
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3409583125377343e-06,
+      "loss": 0.4572,
+      "step": 5122
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.339115854833448e-06,
+      "loss": 0.4681,
+      "step": 5123
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3372739010607925e-06,
+      "loss": 0.479,
+      "step": 5124
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3354324515686062e-06,
+      "loss": 0.4804,
+      "step": 5125
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.333591506705638e-06,
+      "loss": 0.4731,
+      "step": 5126
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3317510668205297e-06,
+      "loss": 0.4829,
+      "step": 5127
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3299111322618345e-06,
+      "loss": 0.4604,
+      "step": 5128
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.32807170337801e-06,
+      "loss": 0.474,
+      "step": 5129
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3262327805174145e-06,
+      "loss": 0.4767,
+      "step": 5130
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.324394364028311e-06,
+      "loss": 0.4608,
+      "step": 5131
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3225564542588734e-06,
+      "loss": 0.4774,
+      "step": 5132
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3207190515571696e-06,
+      "loss": 0.4508,
+      "step": 5133
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3188821562711785e-06,
+      "loss": 0.456,
+      "step": 5134
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3170457687487805e-06,
+      "loss": 0.4567,
+      "step": 5135
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3152098893377583e-06,
+      "loss": 0.4757,
+      "step": 5136
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3133745183857988e-06,
+      "loss": 0.48,
+      "step": 5137
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.311539656240499e-06,
+      "loss": 0.4841,
+      "step": 5138
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.30970530324935e-06,
+      "loss": 0.4779,
+      "step": 5139
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.307871459759752e-06,
+      "loss": 0.4504,
+      "step": 5140
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3060381261190095e-06,
+      "loss": 0.462,
+      "step": 5141
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3042053026743245e-06,
+      "loss": 0.5248,
+      "step": 5142
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3023729897728082e-06,
+      "loss": 0.4795,
+      "step": 5143
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3005411877614736e-06,
+      "loss": 0.4851,
+      "step": 5144
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2987098969872364e-06,
+      "loss": 0.4502,
+      "step": 5145
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.296879117796917e-06,
+      "loss": 0.4449,
+      "step": 5146
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.295048850537235e-06,
+      "loss": 0.4758,
+      "step": 5147
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.293219095554816e-06,
+      "loss": 0.4553,
+      "step": 5148
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.29138985319619e-06,
+      "loss": 0.4566,
+      "step": 5149
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2895611238077863e-06,
+      "loss": 0.4643,
+      "step": 5150
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.28773290773594e-06,
+      "loss": 0.4659,
+      "step": 5151
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.28590520532689e-06,
+      "loss": 0.4703,
+      "step": 5152
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2840780169267707e-06,
+      "loss": 0.458,
+      "step": 5153
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2822513428816275e-06,
+      "loss": 0.4762,
+      "step": 5154
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.280425183537404e-06,
+      "loss": 0.4885,
+      "step": 5155
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2785995392399486e-06,
+      "loss": 0.4753,
+      "step": 5156
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2767744103350097e-06,
+      "loss": 0.4364,
+      "step": 5157
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.274949797168243e-06,
+      "loss": 0.4766,
+      "step": 5158
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2731257000851975e-06,
+      "loss": 0.5096,
+      "step": 5159
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2713021194313324e-06,
+      "loss": 0.4587,
+      "step": 5160
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2694790555520063e-06,
+      "loss": 0.4487,
+      "step": 5161
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2676565087924807e-06,
+      "loss": 0.4696,
+      "step": 5162
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.265834479497919e-06,
+      "loss": 0.4678,
+      "step": 5163
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2640129680133872e-06,
+      "loss": 0.4769,
+      "step": 5164
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2621919746838487e-06,
+      "loss": 0.4727,
+      "step": 5165
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.260371499854175e-06,
+      "loss": 0.4698,
+      "step": 5166
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2585515438691362e-06,
+      "loss": 0.4708,
+      "step": 5167
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2567321070734048e-06,
+      "loss": 0.4928,
+      "step": 5168
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.254913189811555e-06,
+      "loss": 0.4776,
+      "step": 5169
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.253094792428065e-06,
+      "loss": 0.444,
+      "step": 5170
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.251276915267307e-06,
+      "loss": 0.4629,
+      "step": 5171
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.249459558673564e-06,
+      "loss": 0.5114,
+      "step": 5172
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.247642722991011e-06,
+      "loss": 0.4581,
+      "step": 5173
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2458264085637346e-06,
+      "loss": 0.4817,
+      "step": 5174
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2440106157357172e-06,
+      "loss": 0.4949,
+      "step": 5175
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2421953448508403e-06,
+      "loss": 0.4823,
+      "step": 5176
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2403805962528894e-06,
+      "loss": 0.4536,
+      "step": 5177
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.238566370285554e-06,
+      "loss": 0.4702,
+      "step": 5178
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2367526672924155e-06,
+      "loss": 0.4729,
+      "step": 5179
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2349394876169667e-06,
+      "loss": 0.4601,
+      "step": 5180
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2331268316025973e-06,
+      "loss": 0.4422,
+      "step": 5181
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2313146995925933e-06,
+      "loss": 0.5247,
+      "step": 5182
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.229503091930148e-06,
+      "loss": 0.4514,
+      "step": 5183
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2276920089583535e-06,
+      "loss": 0.5013,
+      "step": 5184
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.225881451020197e-06,
+      "loss": 0.4604,
+      "step": 5185
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2240714184585767e-06,
+      "loss": 0.4765,
+      "step": 5186
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2222619116162854e-06,
+      "loss": 0.4608,
+      "step": 5187
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2204529308360125e-06,
+      "loss": 0.4579,
+      "step": 5188
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.218644476460354e-06,
+      "loss": 0.493,
+      "step": 5189
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2168365488318072e-06,
+      "loss": 0.4678,
+      "step": 5190
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.215029148292759e-06,
+      "loss": 0.4467,
+      "step": 5191
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.213222275185511e-06,
+      "loss": 0.4898,
+      "step": 5192
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.211415929852257e-06,
+      "loss": 0.4743,
+      "step": 5193
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.209610112635088e-06,
+      "loss": 0.4429,
+      "step": 5194
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.207804823876002e-06,
+      "loss": 0.4663,
+      "step": 5195
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2060000639168937e-06,
+      "loss": 0.4508,
+      "step": 5196
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2041958330995556e-06,
+      "loss": 0.4839,
+      "step": 5197
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.202392131765683e-06,
+      "loss": 0.4632,
+      "step": 5198
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.20058896025687e-06,
+      "loss": 0.4915,
+      "step": 5199
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.19878631891461e-06,
+      "loss": 0.4916,
+      "step": 5200
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1969842080802977e-06,
+      "loss": 0.458,
+      "step": 5201
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.195182628095224e-06,
+      "loss": 0.4908,
+      "step": 5202
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1933815793005817e-06,
+      "loss": 0.4816,
+      "step": 5203
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1915810620374626e-06,
+      "loss": 0.4962,
+      "step": 5204
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1897810766468585e-06,
+      "loss": 0.4528,
+      "step": 5205
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.187981623469659e-06,
+      "loss": 0.4866,
+      "step": 5206
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.186182702846655e-06,
+      "loss": 0.4824,
+      "step": 5207
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1843843151185323e-06,
+      "loss": 0.4791,
+      "step": 5208
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.18258646062588e-06,
+      "loss": 0.4845,
+      "step": 5209
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.180789139709185e-06,
+      "loss": 0.4663,
+      "step": 5210
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1789923527088336e-06,
+      "loss": 0.4541,
+      "step": 5211
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1771960999651094e-06,
+      "loss": 0.4691,
+      "step": 5212
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1754003818181974e-06,
+      "loss": 0.4399,
+      "step": 5213
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1736051986081767e-06,
+      "loss": 0.4575,
+      "step": 5214
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1718105506750307e-06,
+      "loss": 0.4774,
+      "step": 5215
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1700164383586377e-06,
+      "loss": 0.4443,
+      "step": 5216
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.168222861998776e-06,
+      "loss": 0.4688,
+      "step": 5217
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.166429821935122e-06,
+      "loss": 0.4826,
+      "step": 5218
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.164637318507253e-06,
+      "loss": 0.4669,
+      "step": 5219
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1628453520546383e-06,
+      "loss": 0.5002,
+      "step": 5220
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.161053922916651e-06,
+      "loss": 0.4485,
+      "step": 5221
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.159263031432562e-06,
+      "loss": 0.4916,
+      "step": 5222
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1574726779415385e-06,
+      "loss": 0.4541,
+      "step": 5223
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1556828627826485e-06,
+      "loss": 0.4909,
+      "step": 5224
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.153893586294853e-06,
+      "loss": 0.4902,
+      "step": 5225
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1521048488170153e-06,
+      "loss": 0.4403,
+      "step": 5226
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1503166506878962e-06,
+      "loss": 0.4723,
+      "step": 5227
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1485289922461537e-06,
+      "loss": 0.4525,
+      "step": 5228
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.146741873830343e-06,
+      "loss": 0.4983,
+      "step": 5229
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.14495529577892e-06,
+      "loss": 0.4833,
+      "step": 5230
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1431692584302315e-06,
+      "loss": 0.4939,
+      "step": 5231
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1413837621225285e-06,
+      "loss": 0.4816,
+      "step": 5232
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1395988071939577e-06,
+      "loss": 0.4754,
+      "step": 5233
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1378143939825617e-06,
+      "loss": 0.4511,
+      "step": 5234
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1360305228262827e-06,
+      "loss": 0.4654,
+      "step": 5235
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.134247194062961e-06,
+      "loss": 0.4937,
+      "step": 5236
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.132464408030328e-06,
+      "loss": 0.4795,
+      "step": 5237
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.130682165066019e-06,
+      "loss": 0.4787,
+      "step": 5238
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.128900465507564e-06,
+      "loss": 0.4555,
+      "step": 5239
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1271193096923897e-06,
+      "loss": 0.4761,
+      "step": 5240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1253386979578218e-06,
+      "loss": 0.4297,
+      "step": 5241
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1235586306410823e-06,
+      "loss": 0.4946,
+      "step": 5242
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1217791080792854e-06,
+      "loss": 0.4535,
+      "step": 5243
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1200001306094485e-06,
+      "loss": 0.4837,
+      "step": 5244
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.118221698568485e-06,
+      "loss": 0.4838,
+      "step": 5245
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1164438122931978e-06,
+      "loss": 0.4614,
+      "step": 5246
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.114666472120299e-06,
+      "loss": 0.4601,
+      "step": 5247
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.112889678386385e-06,
+      "loss": 0.4339,
+      "step": 5248
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.111113431427956e-06,
+      "loss": 0.4682,
+      "step": 5249
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1093377315814077e-06,
+      "loss": 0.4775,
+      "step": 5250
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1075625791830284e-06,
+      "loss": 0.4697,
+      "step": 5251
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1057879745690046e-06,
+      "loss": 0.4551,
+      "step": 5252
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1040139180754253e-06,
+      "loss": 0.4726,
+      "step": 5253
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1022404100382654e-06,
+      "loss": 0.4635,
+      "step": 5254
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1004674507934026e-06,
+      "loss": 0.4856,
+      "step": 5255
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.09869504067661e-06,
+      "loss": 0.456,
+      "step": 5256
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.096923180023552e-06,
+      "loss": 0.4682,
+      "step": 5257
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.095151869169793e-06,
+      "loss": 0.4786,
+      "step": 5258
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0933811084507975e-06,
+      "loss": 0.481,
+      "step": 5259
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.091610898201917e-06,
+      "loss": 0.495,
+      "step": 5260
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0898412387584034e-06,
+      "loss": 0.477,
+      "step": 5261
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0880721304554054e-06,
+      "loss": 0.4562,
+      "step": 5262
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0863035736279623e-06,
+      "loss": 0.4779,
+      "step": 5263
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.084535568611013e-06,
+      "loss": 0.4576,
+      "step": 5264
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0827681157393965e-06,
+      "loss": 0.4666,
+      "step": 5265
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0810012153478353e-06,
+      "loss": 0.4502,
+      "step": 5266
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0792348677709566e-06,
+      "loss": 0.4466,
+      "step": 5267
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0774690733432825e-06,
+      "loss": 0.4749,
+      "step": 5268
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.075703832399223e-06,
+      "loss": 0.4922,
+      "step": 5269
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0739391452730912e-06,
+      "loss": 0.4893,
+      "step": 5270
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.072175012299092e-06,
+      "loss": 0.4557,
+      "step": 5271
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.070411433811326e-06,
+      "loss": 0.4987,
+      "step": 5272
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0686484101437902e-06,
+      "loss": 0.4624,
+      "step": 5273
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0668859416303714e-06,
+      "loss": 0.4794,
+      "step": 5274
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0651240286048568e-06,
+      "loss": 0.4594,
+      "step": 5275
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.063362671400926e-06,
+      "loss": 0.4713,
+      "step": 5276
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.061601870352154e-06,
+      "loss": 0.4717,
+      "step": 5277
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0598416257920095e-06,
+      "loss": 0.505,
+      "step": 5278
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.058081938053859e-06,
+      "loss": 0.4647,
+      "step": 5279
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0563228074709575e-06,
+      "loss": 0.485,
+      "step": 5280
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.054564234376459e-06,
+      "loss": 0.4837,
+      "step": 5281
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0528062191034116e-06,
+      "loss": 0.4596,
+      "step": 5282
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0510487619847564e-06,
+      "loss": 0.4952,
+      "step": 5283
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0492918633533303e-06,
+      "loss": 0.5019,
+      "step": 5284
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0475355235418653e-06,
+      "loss": 0.4824,
+      "step": 5285
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.045779742882982e-06,
+      "loss": 0.488,
+      "step": 5286
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0440245217092e-06,
+      "loss": 0.4385,
+      "step": 5287
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.042269860352934e-06,
+      "loss": 0.4605,
+      "step": 5288
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0405157591464887e-06,
+      "loss": 0.4743,
+      "step": 5289
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0387622184220653e-06,
+      "loss": 0.4347,
+      "step": 5290
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0370092385117595e-06,
+      "loss": 0.4513,
+      "step": 5291
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.035256819747557e-06,
+      "loss": 0.4475,
+      "step": 5292
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.03350496246134e-06,
+      "loss": 0.462,
+      "step": 5293
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0317536669848855e-06,
+      "loss": 0.5181,
+      "step": 5294
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.030002933649862e-06,
+      "loss": 0.4729,
+      "step": 5295
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0282527627878314e-06,
+      "loss": 0.4462,
+      "step": 5296
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0265031547302527e-06,
+      "loss": 0.4305,
+      "step": 5297
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0247541098084717e-06,
+      "loss": 0.4572,
+      "step": 5298
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0230056283537345e-06,
+      "loss": 0.4738,
+      "step": 5299
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0212577106971726e-06,
+      "loss": 0.4739,
+      "step": 5300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0195103571698205e-06,
+      "loss": 0.4469,
+      "step": 5301
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.017763568102601e-06,
+      "loss": 0.4726,
+      "step": 5302
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0160173438263257e-06,
+      "loss": 0.4888,
+      "step": 5303
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0142716846717053e-06,
+      "loss": 0.4471,
+      "step": 5304
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0125265909693436e-06,
+      "loss": 0.4626,
+      "step": 5305
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.01078206304973e-06,
+      "loss": 0.4525,
+      "step": 5306
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.009038101243257e-06,
+      "loss": 0.4547,
+      "step": 5307
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0072947058802044e-06,
+      "loss": 0.4568,
+      "step": 5308
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.005551877290743e-06,
+      "loss": 0.4632,
+      "step": 5309
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0038096158049384e-06,
+      "loss": 0.4628,
+      "step": 5310
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.002067921752752e-06,
+      "loss": 0.4509,
+      "step": 5311
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0003267954640293e-06,
+      "loss": 0.4899,
+      "step": 5312
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9985862372685185e-06,
+      "loss": 0.473,
+      "step": 5313
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9968462474958546e-06,
+      "loss": 0.4539,
+      "step": 5314
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.995106826475563e-06,
+      "loss": 0.5062,
+      "step": 5315
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.993367974537065e-06,
+      "loss": 0.4578,
+      "step": 5316
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9916296920096763e-06,
+      "loss": 0.4565,
+      "step": 5317
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9898919792225948e-06,
+      "loss": 0.4573,
+      "step": 5318
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.988154836504923e-06,
+      "loss": 0.4703,
+      "step": 5319
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.98641826418565e-06,
+      "loss": 0.4618,
+      "step": 5320
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.984682262593653e-06,
+      "loss": 0.4655,
+      "step": 5321
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.982946832057707e-06,
+      "loss": 0.4519,
+      "step": 5322
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9812119729064764e-06,
+      "loss": 0.476,
+      "step": 5323
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9794776854685165e-06,
+      "loss": 0.4774,
+      "step": 5324
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9777439700722762e-06,
+      "loss": 0.4767,
+      "step": 5325
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.976010827046095e-06,
+      "loss": 0.4428,
+      "step": 5326
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.974278256718204e-06,
+      "loss": 0.4393,
+      "step": 5327
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9725462594167283e-06,
+      "loss": 0.4652,
+      "step": 5328
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.970814835469679e-06,
+      "loss": 0.4625,
+      "step": 5329
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9690839852049627e-06,
+      "loss": 0.4632,
+      "step": 5330
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9673537089503775e-06,
+      "loss": 0.484,
+      "step": 5331
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9656240070336114e-06,
+      "loss": 0.4604,
+      "step": 5332
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.963894879782244e-06,
+      "loss": 0.4366,
+      "step": 5333
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.962166327523748e-06,
+      "loss": 0.4504,
+      "step": 5334
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9604383505854818e-06,
+      "loss": 0.4559,
+      "step": 5335
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9587109492947003e-06,
+      "loss": 0.4559,
+      "step": 5336
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9569841239785477e-06,
+      "loss": 0.4941,
+      "step": 5337
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9552578749640578e-06,
+      "loss": 0.474,
+      "step": 5338
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9535322025781573e-06,
+      "loss": 0.4603,
+      "step": 5339
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9518071071476643e-06,
+      "loss": 0.4674,
+      "step": 5340
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9500825889992817e-06,
+      "loss": 0.4732,
+      "step": 5341
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.948358648459611e-06,
+      "loss": 0.4731,
+      "step": 5342
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9466352858551383e-06,
+      "loss": 0.4707,
+      "step": 5343
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.944912501512245e-06,
+      "loss": 0.4709,
+      "step": 5344
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9431902957571995e-06,
+      "loss": 0.486,
+      "step": 5345
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9414686689161638e-06,
+      "loss": 0.5071,
+      "step": 5346
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9397476213151844e-06,
+      "loss": 0.4872,
+      "step": 5347
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.938027153280204e-06,
+      "loss": 0.4971,
+      "step": 5348
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.936307265137054e-06,
+      "loss": 0.4674,
+      "step": 5349
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9345879572114545e-06,
+      "loss": 0.4772,
+      "step": 5350
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9328692298290193e-06,
+      "loss": 0.4829,
+      "step": 5351
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9311510833152453e-06,
+      "loss": 0.4888,
+      "step": 5352
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.929433517995526e-06,
+      "loss": 0.4595,
+      "step": 5353
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9277165341951422e-06,
+      "loss": 0.4775,
+      "step": 5354
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9260001322392655e-06,
+      "loss": 0.501,
+      "step": 5355
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.924284312452957e-06,
+      "loss": 0.486,
+      "step": 5356
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9225690751611677e-06,
+      "loss": 0.4764,
+      "step": 5357
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.920854420688736e-06,
+      "loss": 0.4629,
+      "step": 5358
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9191403493603923e-06,
+      "loss": 0.5125,
+      "step": 5359
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.917426861500756e-06,
+      "loss": 0.476,
+      "step": 5360
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.915713957434337e-06,
+      "loss": 0.4636,
+      "step": 5361
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.914001637485533e-06,
+      "loss": 0.4677,
+      "step": 5362
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.912289901978634e-06,
+      "loss": 0.4511,
+      "step": 5363
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.910578751237814e-06,
+      "loss": 0.447,
+      "step": 5364
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9088681855871393e-06,
+      "loss": 0.4634,
+      "step": 5365
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.907158205350567e-06,
+      "loss": 0.4216,
+      "step": 5366
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9054488108519415e-06,
+      "loss": 0.4671,
+      "step": 5367
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.903740002414996e-06,
+      "loss": 0.4813,
+      "step": 5368
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9020317803633554e-06,
+      "loss": 0.4617,
+      "step": 5369
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.900324145020528e-06,
+      "loss": 0.4723,
+      "step": 5370
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.898617096709916e-06,
+      "loss": 0.4963,
+      "step": 5371
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8969106357548084e-06,
+      "loss": 0.4863,
+      "step": 5372
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8952047624783843e-06,
+      "loss": 0.4636,
+      "step": 5373
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8934994772037096e-06,
+      "loss": 0.4678,
+      "step": 5374
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8917947802537422e-06,
+      "loss": 0.4714,
+      "step": 5375
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8900906719513222e-06,
+      "loss": 0.4897,
+      "step": 5376
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8883871526191861e-06,
+      "loss": 0.4854,
+      "step": 5377
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8866842225799514e-06,
+      "loss": 0.4585,
+      "step": 5378
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8849818821561277e-06,
+      "loss": 0.4603,
+      "step": 5379
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.883280131670117e-06,
+      "loss": 0.4621,
+      "step": 5380
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8815789714442018e-06,
+      "loss": 0.4803,
+      "step": 5381
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8798784018005572e-06,
+      "loss": 0.5054,
+      "step": 5382
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.878178423061247e-06,
+      "loss": 0.5,
+      "step": 5383
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8764790355482192e-06,
+      "loss": 0.4402,
+      "step": 5384
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8747802395833114e-06,
+      "loss": 0.4807,
+      "step": 5385
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8730820354882561e-06,
+      "loss": 0.4408,
+      "step": 5386
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8713844235846623e-06,
+      "loss": 0.4357,
+      "step": 5387
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8696874041940338e-06,
+      "loss": 0.4756,
+      "step": 5388
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.867990977637763e-06,
+      "loss": 0.4603,
+      "step": 5389
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8662951442371235e-06,
+      "loss": 0.4828,
+      "step": 5390
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8645999043132813e-06,
+      "loss": 0.4739,
+      "step": 5391
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8629052581872942e-06,
+      "loss": 0.4479,
+      "step": 5392
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.861211206180098e-06,
+      "loss": 0.4506,
+      "step": 5393
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.859517748612522e-06,
+      "loss": 0.4636,
+      "step": 5394
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8578248858052844e-06,
+      "loss": 0.4844,
+      "step": 5395
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.856132618078984e-06,
+      "loss": 0.4643,
+      "step": 5396
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8544409457541106e-06,
+      "loss": 0.4563,
+      "step": 5397
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8527498691510476e-06,
+      "loss": 0.4531,
+      "step": 5398
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8510593885900542e-06,
+      "loss": 0.4908,
+      "step": 5399
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8493695043912834e-06,
+      "loss": 0.4746,
+      "step": 5400
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8476802168747766e-06,
+      "loss": 0.5005,
+      "step": 5401
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.845991526360455e-06,
+      "loss": 0.4852,
+      "step": 5402
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8443034331681336e-06,
+      "loss": 0.4836,
+      "step": 5403
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8426159376175118e-06,
+      "loss": 0.4731,
+      "step": 5404
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8409290400281765e-06,
+      "loss": 0.4691,
+      "step": 5405
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8392427407196016e-06,
+      "loss": 0.4563,
+      "step": 5406
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8375570400111448e-06,
+      "loss": 0.4905,
+      "step": 5407
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8358719382220536e-06,
+      "loss": 0.4859,
+      "step": 5408
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8341874356714612e-06,
+      "loss": 0.4635,
+      "step": 5409
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.832503532678387e-06,
+      "loss": 0.4772,
+      "step": 5410
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8308202295617367e-06,
+      "loss": 0.4526,
+      "step": 5411
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.829137526640305e-06,
+      "loss": 0.475,
+      "step": 5412
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8274554242327674e-06,
+      "loss": 0.447,
+      "step": 5413
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8257739226576898e-06,
+      "loss": 0.4356,
+      "step": 5414
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.824093022233524e-06,
+      "loss": 0.473,
+      "step": 5415
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8224127232786076e-06,
+      "loss": 0.4474,
+      "step": 5416
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.820733026111164e-06,
+      "loss": 0.4449,
+      "step": 5417
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8190539310493038e-06,
+      "loss": 0.5066,
+      "step": 5418
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8173754384110194e-06,
+      "loss": 0.4822,
+      "step": 5419
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8156975485141937e-06,
+      "loss": 0.4677,
+      "step": 5420
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8140202616765945e-06,
+      "loss": 0.4519,
+      "step": 5421
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.812343578215875e-06,
+      "loss": 0.4758,
+      "step": 5422
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8106674984495726e-06,
+      "loss": 0.473,
+      "step": 5423
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8089920226951147e-06,
+      "loss": 0.4561,
+      "step": 5424
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8073171512698074e-06,
+      "loss": 0.4889,
+      "step": 5425
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.805642884490848e-06,
+      "loss": 0.4374,
+      "step": 5426
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.803969222675318e-06,
+      "loss": 0.4755,
+      "step": 5427
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8022961661401833e-06,
+      "loss": 0.4619,
+      "step": 5428
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8006237152022976e-06,
+      "loss": 0.4763,
+      "step": 5429
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7989518701783953e-06,
+      "loss": 0.4724,
+      "step": 5430
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7972806313850994e-06,
+      "loss": 0.4609,
+      "step": 5431
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7956099991389204e-06,
+      "loss": 0.4699,
+      "step": 5432
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7939399737562451e-06,
+      "loss": 0.4667,
+      "step": 5433
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7922705555533565e-06,
+      "loss": 0.4815,
+      "step": 5434
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.790601744846418e-06,
+      "loss": 0.4792,
+      "step": 5435
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7889335419514735e-06,
+      "loss": 0.4797,
+      "step": 5436
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7872659471844572e-06,
+      "loss": 0.4537,
+      "step": 5437
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.785598960861189e-06,
+      "loss": 0.4933,
+      "step": 5438
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7839325832973652e-06,
+      "loss": 0.4759,
+      "step": 5439
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.782266814808578e-06,
+      "loss": 0.4557,
+      "step": 5440
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7806016557103001e-06,
+      "loss": 0.435,
+      "step": 5441
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.778937106317883e-06,
+      "loss": 0.4886,
+      "step": 5442
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7772731669465699e-06,
+      "loss": 0.4508,
+      "step": 5443
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7756098379114873e-06,
+      "loss": 0.4645,
+      "step": 5444
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7739471195276397e-06,
+      "loss": 0.4872,
+      "step": 5445
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7722850121099267e-06,
+      "loss": 0.4352,
+      "step": 5446
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7706235159731262e-06,
+      "loss": 0.461,
+      "step": 5447
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.768962631431897e-06,
+      "loss": 0.463,
+      "step": 5448
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7673023588007882e-06,
+      "loss": 0.4826,
+      "step": 5449
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7656426983942315e-06,
+      "loss": 0.47,
+      "step": 5450
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.763983650526539e-06,
+      "loss": 0.462,
+      "step": 5451
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7623252155119113e-06,
+      "loss": 0.4353,
+      "step": 5452
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7606673936644303e-06,
+      "loss": 0.4847,
+      "step": 5453
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7590101852980635e-06,
+      "loss": 0.4396,
+      "step": 5454
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7573535907266625e-06,
+      "loss": 0.4848,
+      "step": 5455
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7556976102639583e-06,
+      "loss": 0.4774,
+      "step": 5456
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7540422442235712e-06,
+      "loss": 0.477,
+      "step": 5457
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7523874929190016e-06,
+      "loss": 0.4423,
+      "step": 5458
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7507333566636354e-06,
+      "loss": 0.4469,
+      "step": 5459
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7490798357707406e-06,
+      "loss": 0.4848,
+      "step": 5460
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.747426930553472e-06,
+      "loss": 0.4772,
+      "step": 5461
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7457746413248612e-06,
+      "loss": 0.4826,
+      "step": 5462
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7441229683978277e-06,
+      "loss": 0.4967,
+      "step": 5463
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7424719120851757e-06,
+      "loss": 0.4735,
+      "step": 5464
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7408214726995892e-06,
+      "loss": 0.4705,
+      "step": 5465
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7391716505536365e-06,
+      "loss": 0.4688,
+      "step": 5466
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7375224459597716e-06,
+      "loss": 0.4724,
+      "step": 5467
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.735873859230326e-06,
+      "loss": 0.4763,
+      "step": 5468
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7342258906775183e-06,
+      "loss": 0.4452,
+      "step": 5469
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7325785406134499e-06,
+      "loss": 0.4793,
+      "step": 5470
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7309318093501032e-06,
+      "loss": 0.4533,
+      "step": 5471
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7292856971993454e-06,
+      "loss": 0.4866,
+      "step": 5472
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7276402044729262e-06,
+      "loss": 0.4625,
+      "step": 5473
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.725995331482475e-06,
+      "loss": 0.4999,
+      "step": 5474
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7243510785395074e-06,
+      "loss": 0.4747,
+      "step": 5475
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7227074459554206e-06,
+      "loss": 0.4766,
+      "step": 5476
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7210644340414928e-06,
+      "loss": 0.4557,
+      "step": 5477
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7194220431088887e-06,
+      "loss": 0.4891,
+      "step": 5478
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7177802734686494e-06,
+      "loss": 0.4756,
+      "step": 5479
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7161391254317023e-06,
+      "loss": 0.4237,
+      "step": 5480
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7144985993088563e-06,
+      "loss": 0.4726,
+      "step": 5481
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7128586954108035e-06,
+      "loss": 0.459,
+      "step": 5482
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7112194140481159e-06,
+      "loss": 0.4779,
+      "step": 5483
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7095807555312515e-06,
+      "loss": 0.467,
+      "step": 5484
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7079427201705439e-06,
+      "loss": 0.4658,
+      "step": 5485
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7063053082762143e-06,
+      "loss": 0.4853,
+      "step": 5486
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.704668520158364e-06,
+      "loss": 0.4914,
+      "step": 5487
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7030323561269763e-06,
+      "loss": 0.4528,
+      "step": 5488
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7013968164919153e-06,
+      "loss": 0.4362,
+      "step": 5489
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6997619015629307e-06,
+      "loss": 0.4623,
+      "step": 5490
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.698127611649647e-06,
+      "loss": 0.4968,
+      "step": 5491
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6964939470615754e-06,
+      "loss": 0.4568,
+      "step": 5492
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6948609081081087e-06,
+      "loss": 0.485,
+      "step": 5493
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.693228495098519e-06,
+      "loss": 0.458,
+      "step": 5494
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.691596708341961e-06,
+      "loss": 0.4863,
+      "step": 5495
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.689965548147473e-06,
+      "loss": 0.4734,
+      "step": 5496
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6883350148239686e-06,
+      "loss": 0.4917,
+      "step": 5497
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6867051086802477e-06,
+      "loss": 0.4612,
+      "step": 5498
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6850758300249908e-06,
+      "loss": 0.462,
+      "step": 5499
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6834471791667583e-06,
+      "loss": 0.482,
+      "step": 5500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.681819156413993e-06,
+      "loss": 0.444,
+      "step": 5501
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6801917620750192e-06,
+      "loss": 0.4964,
+      "step": 5502
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6785649964580375e-06,
+      "loss": 0.4557,
+      "step": 5503
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6769388598711368e-06,
+      "loss": 0.4666,
+      "step": 5504
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6753133526222791e-06,
+      "loss": 0.4604,
+      "step": 5505
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6736884750193121e-06,
+      "loss": 0.445,
+      "step": 5506
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6720642273699677e-06,
+      "loss": 0.4824,
+      "step": 5507
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6704406099818494e-06,
+      "loss": 0.4676,
+      "step": 5508
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.668817623162447e-06,
+      "loss": 0.4867,
+      "step": 5509
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6671952672191316e-06,
+      "loss": 0.49,
+      "step": 5510
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6655735424591507e-06,
+      "loss": 0.4445,
+      "step": 5511
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6639524491896342e-06,
+      "loss": 0.4725,
+      "step": 5512
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.662331987717598e-06,
+      "loss": 0.4915,
+      "step": 5513
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.660712158349928e-06,
+      "loss": 0.4672,
+      "step": 5514
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6590929613933976e-06,
+      "loss": 0.4791,
+      "step": 5515
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6574743971546597e-06,
+      "loss": 0.4738,
+      "step": 5516
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6558564659402437e-06,
+      "loss": 0.4919,
+      "step": 5517
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6542391680565606e-06,
+      "loss": 0.4519,
+      "step": 5518
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6526225038099076e-06,
+      "loss": 0.5007,
+      "step": 5519
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6510064735064525e-06,
+      "loss": 0.4852,
+      "step": 5520
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6493910774522476e-06,
+      "loss": 0.4526,
+      "step": 5521
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6477763159532272e-06,
+      "loss": 0.4728,
+      "step": 5522
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6461621893152002e-06,
+      "loss": 0.4918,
+      "step": 5523
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6445486978438568e-06,
+      "loss": 0.4524,
+      "step": 5524
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6429358418447738e-06,
+      "loss": 0.4951,
+      "step": 5525
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6413236216233975e-06,
+      "loss": 0.4656,
+      "step": 5526
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6397120374850584e-06,
+      "loss": 0.4416,
+      "step": 5527
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6381010897349697e-06,
+      "loss": 0.4683,
+      "step": 5528
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6364907786782165e-06,
+      "loss": 0.4649,
+      "step": 5529
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6348811046197694e-06,
+      "loss": 0.4616,
+      "step": 5530
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.633272067864477e-06,
+      "loss": 0.4773,
+      "step": 5531
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6316636687170667e-06,
+      "loss": 0.4887,
+      "step": 5532
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6300559074821465e-06,
+      "loss": 0.466,
+      "step": 5533
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.628448784464199e-06,
+      "loss": 0.4727,
+      "step": 5534
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6268422999675914e-06,
+      "loss": 0.4687,
+      "step": 5535
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6252364542965676e-06,
+      "loss": 0.4609,
+      "step": 5536
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6236312477552501e-06,
+      "loss": 0.4656,
+      "step": 5537
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.622026680647642e-06,
+      "loss": 0.4722,
+      "step": 5538
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6204227532776257e-06,
+      "loss": 0.4633,
+      "step": 5539
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6188194659489575e-06,
+      "loss": 0.4492,
+      "step": 5540
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6172168189652777e-06,
+      "loss": 0.488,
+      "step": 5541
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.615614812630103e-06,
+      "loss": 0.4618,
+      "step": 5542
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.614013447246831e-06,
+      "loss": 0.462,
+      "step": 5543
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6124127231187349e-06,
+      "loss": 0.4368,
+      "step": 5544
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6108126405489705e-06,
+      "loss": 0.4855,
+      "step": 5545
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6092131998405653e-06,
+      "loss": 0.4517,
+      "step": 5546
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6076144012964318e-06,
+      "loss": 0.4901,
+      "step": 5547
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.606016245219358e-06,
+      "loss": 0.4343,
+      "step": 5548
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6044187319120113e-06,
+      "loss": 0.4455,
+      "step": 5549
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6028218616769364e-06,
+      "loss": 0.4847,
+      "step": 5550
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6012256348165584e-06,
+      "loss": 0.4748,
+      "step": 5551
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5996300516331753e-06,
+      "loss": 0.4392,
+      "step": 5552
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.598035112428969e-06,
+      "loss": 0.4282,
+      "step": 5553
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.596440817505996e-06,
+      "loss": 0.4384,
+      "step": 5554
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.594847167166193e-06,
+      "loss": 0.5078,
+      "step": 5555
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5932541617113744e-06,
+      "loss": 0.4779,
+      "step": 5556
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5916618014432284e-06,
+      "loss": 0.4765,
+      "step": 5557
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5900700866633267e-06,
+      "loss": 0.4869,
+      "step": 5558
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5884790176731163e-06,
+      "loss": 0.4519,
+      "step": 5559
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5868885947739187e-06,
+      "loss": 0.4609,
+      "step": 5560
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.58529881826694e-06,
+      "loss": 0.4653,
+      "step": 5561
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.58370968845326e-06,
+      "loss": 0.4799,
+      "step": 5562
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.582121205633833e-06,
+      "loss": 0.4741,
+      "step": 5563
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5805333701094967e-06,
+      "loss": 0.442,
+      "step": 5564
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5789461821809631e-06,
+      "loss": 0.4575,
+      "step": 5565
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.577359642148818e-06,
+      "loss": 0.471,
+      "step": 5566
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5757737503135334e-06,
+      "loss": 0.4299,
+      "step": 5567
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5741885069754526e-06,
+      "loss": 0.4583,
+      "step": 5568
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5726039124347947e-06,
+      "loss": 0.5232,
+      "step": 5569
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5710199669916592e-06,
+      "loss": 0.4667,
+      "step": 5570
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5694366709460236e-06,
+      "loss": 0.4837,
+      "step": 5571
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5678540245977352e-06,
+      "loss": 0.4599,
+      "step": 5572
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5662720282465288e-06,
+      "loss": 0.4867,
+      "step": 5573
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5646906821920104e-06,
+      "loss": 0.4561,
+      "step": 5574
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5631099867336607e-06,
+      "loss": 0.471,
+      "step": 5575
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5615299421708412e-06,
+      "loss": 0.4536,
+      "step": 5576
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.55995054880279e-06,
+      "loss": 0.4533,
+      "step": 5577
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5583718069286151e-06,
+      "loss": 0.4792,
+      "step": 5578
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.556793716847313e-06,
+      "loss": 0.4723,
+      "step": 5579
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.555216278857749e-06,
+      "loss": 0.4795,
+      "step": 5580
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.553639493258664e-06,
+      "loss": 0.4649,
+      "step": 5581
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.55206336034868e-06,
+      "loss": 0.4777,
+      "step": 5582
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5504878804262902e-06,
+      "loss": 0.4651,
+      "step": 5583
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5489130537898683e-06,
+      "loss": 0.4616,
+      "step": 5584
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5473388807376632e-06,
+      "loss": 0.4971,
+      "step": 5585
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5457653615677991e-06,
+      "loss": 0.4241,
+      "step": 5586
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.544192496578278e-06,
+      "loss": 0.4496,
+      "step": 5587
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5426202860669776e-06,
+      "loss": 0.461,
+      "step": 5588
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5410487303316484e-06,
+      "loss": 0.5144,
+      "step": 5589
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5394778296699215e-06,
+      "loss": 0.4638,
+      "step": 5590
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5379075843793006e-06,
+      "loss": 0.4623,
+      "step": 5591
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5363379947571678e-06,
+      "loss": 0.5175,
+      "step": 5592
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5347690611007798e-06,
+      "loss": 0.4863,
+      "step": 5593
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5332007837072704e-06,
+      "loss": 0.4602,
+      "step": 5594
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5316331628736447e-06,
+      "loss": 0.4885,
+      "step": 5595
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5300661988967885e-06,
+      "loss": 0.429,
+      "step": 5596
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5284998920734611e-06,
+      "loss": 0.4588,
+      "step": 5597
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.526934242700298e-06,
+      "loss": 0.4561,
+      "step": 5598
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5253692510738087e-06,
+      "loss": 0.4814,
+      "step": 5599
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5238049174903824e-06,
+      "loss": 0.4502,
+      "step": 5600
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5222412422462756e-06,
+      "loss": 0.4675,
+      "step": 5601
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5206782256376273e-06,
+      "loss": 0.4534,
+      "step": 5602
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.51911586796045e-06,
+      "loss": 0.4643,
+      "step": 5603
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.51755416951063e-06,
+      "loss": 0.4751,
+      "step": 5604
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5159931305839304e-06,
+      "loss": 0.5096,
+      "step": 5605
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5144327514759893e-06,
+      "loss": 0.4631,
+      "step": 5606
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5128730324823166e-06,
+      "loss": 0.485,
+      "step": 5607
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5113139738983013e-06,
+      "loss": 0.4923,
+      "step": 5608
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5097555760192056e-06,
+      "loss": 0.4305,
+      "step": 5609
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5081978391401663e-06,
+      "loss": 0.472,
+      "step": 5610
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5066407635561981e-06,
+      "loss": 0.453,
+      "step": 5611
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5050843495621837e-06,
+      "loss": 0.4637,
+      "step": 5612
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5035285974528869e-06,
+      "loss": 0.4956,
+      "step": 5613
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.501973507522943e-06,
+      "loss": 0.4796,
+      "step": 5614
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5004190800668633e-06,
+      "loss": 0.4745,
+      "step": 5615
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4988653153790333e-06,
+      "loss": 0.465,
+      "step": 5616
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4973122137537145e-06,
+      "loss": 0.4747,
+      "step": 5617
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4957597754850367e-06,
+      "loss": 0.4761,
+      "step": 5618
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4942080008670113e-06,
+      "loss": 0.4955,
+      "step": 5619
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4926568901935212e-06,
+      "loss": 0.4731,
+      "step": 5620
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4911064437583227e-06,
+      "loss": 0.4694,
+      "step": 5621
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4895566618550477e-06,
+      "loss": 0.4828,
+      "step": 5622
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.488007544777203e-06,
+      "loss": 0.4105,
+      "step": 5623
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4864590928181656e-06,
+      "loss": 0.4612,
+      "step": 5624
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4849113062711896e-06,
+      "loss": 0.497,
+      "step": 5625
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.483364185429404e-06,
+      "loss": 0.453,
+      "step": 5626
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4818177305858096e-06,
+      "loss": 0.4845,
+      "step": 5627
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.480271942033281e-06,
+      "loss": 0.444,
+      "step": 5628
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.478726820064571e-06,
+      "loss": 0.452,
+      "step": 5629
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4771823649722977e-06,
+      "loss": 0.5115,
+      "step": 5630
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4756385770489602e-06,
+      "loss": 0.4565,
+      "step": 5631
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.474095456586928e-06,
+      "loss": 0.4427,
+      "step": 5632
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4725530038784458e-06,
+      "loss": 0.46,
+      "step": 5633
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.471011219215633e-06,
+      "loss": 0.4638,
+      "step": 5634
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4694701028904762e-06,
+      "loss": 0.4439,
+      "step": 5635
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4679296551948418e-06,
+      "loss": 0.4709,
+      "step": 5636
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4663898764204699e-06,
+      "loss": 0.4867,
+      "step": 5637
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4648507668589669e-06,
+      "loss": 0.4614,
+      "step": 5638
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4633123268018184e-06,
+      "loss": 0.4718,
+      "step": 5639
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4617745565403862e-06,
+      "loss": 0.4705,
+      "step": 5640
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4602374563658966e-06,
+      "loss": 0.4813,
+      "step": 5641
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.458701026569454e-06,
+      "loss": 0.4879,
+      "step": 5642
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4571652674420378e-06,
+      "loss": 0.4627,
+      "step": 5643
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4556301792744941e-06,
+      "loss": 0.4861,
+      "step": 5644
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4540957623575458e-06,
+      "loss": 0.4535,
+      "step": 5645
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4525620169817933e-06,
+      "loss": 0.4656,
+      "step": 5646
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4510289434376994e-06,
+      "loss": 0.4886,
+      "step": 5647
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4494965420156087e-06,
+      "loss": 0.476,
+      "step": 5648
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.447964813005735e-06,
+      "loss": 0.4753,
+      "step": 5649
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4464337566981623e-06,
+      "loss": 0.4734,
+      "step": 5650
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.44490337338285e-06,
+      "loss": 0.4749,
+      "step": 5651
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4433736633496343e-06,
+      "loss": 0.4765,
+      "step": 5652
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4418446268882146e-06,
+      "loss": 0.4585,
+      "step": 5653
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.440316264288169e-06,
+      "loss": 0.4625,
+      "step": 5654
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.438788575838948e-06,
+      "loss": 0.4716,
+      "step": 5655
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4372615618298701e-06,
+      "loss": 0.4764,
+      "step": 5656
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.435735222550131e-06,
+      "loss": 0.4593,
+      "step": 5657
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.434209558288795e-06,
+      "loss": 0.4873,
+      "step": 5658
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4326845693348013e-06,
+      "loss": 0.4688,
+      "step": 5659
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4311602559769615e-06,
+      "loss": 0.5018,
+      "step": 5660
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.429636618503954e-06,
+      "loss": 0.4808,
+      "step": 5661
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4281136572043347e-06,
+      "loss": 0.4565,
+      "step": 5662
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4265913723665299e-06,
+      "loss": 0.4588,
+      "step": 5663
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4250697642788375e-06,
+      "loss": 0.4756,
+      "step": 5664
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4235488332294274e-06,
+      "loss": 0.4427,
+      "step": 5665
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.422028579506342e-06,
+      "loss": 0.4981,
+      "step": 5666
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4205090033974928e-06,
+      "loss": 0.4671,
+      "step": 5667
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.418990105190665e-06,
+      "loss": 0.4675,
+      "step": 5668
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.417471885173516e-06,
+      "loss": 0.4375,
+      "step": 5669
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4159543436335738e-06,
+      "loss": 0.4623,
+      "step": 5670
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4144374808582379e-06,
+      "loss": 0.4889,
+      "step": 5671
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.412921297134781e-06,
+      "loss": 0.4643,
+      "step": 5672
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4114057927503426e-06,
+      "loss": 0.4455,
+      "step": 5673
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.409890967991938e-06,
+      "loss": 0.4688,
+      "step": 5674
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4083768231464529e-06,
+      "loss": 0.4722,
+      "step": 5675
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4068633585006425e-06,
+      "loss": 0.4673,
+      "step": 5676
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4053505743411355e-06,
+      "loss": 0.4967,
+      "step": 5677
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4038384709544312e-06,
+      "loss": 0.4577,
+      "step": 5678
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4023270486268969e-06,
+      "loss": 0.464,
+      "step": 5679
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.400816307644775e-06,
+      "loss": 0.4702,
+      "step": 5680
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3993062482941766e-06,
+      "loss": 0.4656,
+      "step": 5681
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3977968708610844e-06,
+      "loss": 0.4418,
+      "step": 5682
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3962881756313528e-06,
+      "loss": 0.4926,
+      "step": 5683
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3947801628907066e-06,
+      "loss": 0.4453,
+      "step": 5684
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.393272832924738e-06,
+      "loss": 0.481,
+      "step": 5685
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3917661860189163e-06,
+      "loss": 0.4749,
+      "step": 5686
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3902602224585732e-06,
+      "loss": 0.4737,
+      "step": 5687
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3887549425289197e-06,
+      "loss": 0.4543,
+      "step": 5688
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3872503465150339e-06,
+      "loss": 0.4754,
+      "step": 5689
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3857464347018602e-06,
+      "loss": 0.4665,
+      "step": 5690
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3842432073742185e-06,
+      "loss": 0.4632,
+      "step": 5691
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3827406648167996e-06,
+      "loss": 0.4592,
+      "step": 5692
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3812388073141576e-06,
+      "loss": 0.4698,
+      "step": 5693
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3797376351507263e-06,
+      "loss": 0.4647,
+      "step": 5694
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.378237148610806e-06,
+      "loss": 0.4663,
+      "step": 5695
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3767373479785623e-06,
+      "loss": 0.4676,
+      "step": 5696
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.375238233538037e-06,
+      "loss": 0.4758,
+      "step": 5697
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.373739805573142e-06,
+      "loss": 0.4596,
+      "step": 5698
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3722420643676516e-06,
+      "loss": 0.4939,
+      "step": 5699
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3707450102052211e-06,
+      "loss": 0.4605,
+      "step": 5700
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3692486433693692e-06,
+      "loss": 0.4633,
+      "step": 5701
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3677529641434828e-06,
+      "loss": 0.503,
+      "step": 5702
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3662579728108227e-06,
+      "loss": 0.4934,
+      "step": 5703
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3647636696545192e-06,
+      "loss": 0.4774,
+      "step": 5704
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3632700549575667e-06,
+      "loss": 0.477,
+      "step": 5705
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3617771290028382e-06,
+      "loss": 0.48,
+      "step": 5706
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3602848920730705e-06,
+      "loss": 0.4637,
+      "step": 5707
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3587933444508689e-06,
+      "loss": 0.4463,
+      "step": 5708
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3573024864187134e-06,
+      "loss": 0.4747,
+      "step": 5709
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.355812318258946e-06,
+      "loss": 0.4498,
+      "step": 5710
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.354322840253785e-06,
+      "loss": 0.4609,
+      "step": 5711
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3528340526853145e-06,
+      "loss": 0.4329,
+      "step": 5712
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3513459558354892e-06,
+      "loss": 0.4933,
+      "step": 5713
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3498585499861317e-06,
+      "loss": 0.4544,
+      "step": 5714
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3483718354189356e-06,
+      "loss": 0.4812,
+      "step": 5715
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3468858124154605e-06,
+      "loss": 0.473,
+      "step": 5716
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.345400481257138e-06,
+      "loss": 0.477,
+      "step": 5717
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3439158422252674e-06,
+      "loss": 0.4858,
+      "step": 5718
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.342431895601017e-06,
+      "loss": 0.4676,
+      "step": 5719
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3409486416654243e-06,
+      "loss": 0.4493,
+      "step": 5720
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.339466080699397e-06,
+      "loss": 0.5007,
+      "step": 5721
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3379842129837066e-06,
+      "loss": 0.4656,
+      "step": 5722
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3365030387989986e-06,
+      "loss": 0.4859,
+      "step": 5723
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3350225584257853e-06,
+      "loss": 0.4464,
+      "step": 5724
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3335427721444466e-06,
+      "loss": 0.4468,
+      "step": 5725
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3320636802352328e-06,
+      "loss": 0.4763,
+      "step": 5726
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.330585282978263e-06,
+      "loss": 0.475,
+      "step": 5727
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3291075806535203e-06,
+      "loss": 0.4406,
+      "step": 5728
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3276305735408613e-06,
+      "loss": 0.4901,
+      "step": 5729
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.326154261920009e-06,
+      "loss": 0.4716,
+      "step": 5730
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3246786460705547e-06,
+      "loss": 0.4736,
+      "step": 5731
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3232037262719571e-06,
+      "loss": 0.4533,
+      "step": 5732
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3217295028035466e-06,
+      "loss": 0.4409,
+      "step": 5733
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3202559759445156e-06,
+      "loss": 0.4644,
+      "step": 5734
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3187831459739298e-06,
+      "loss": 0.5116,
+      "step": 5735
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.31731101317072e-06,
+      "loss": 0.4672,
+      "step": 5736
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3158395778136879e-06,
+      "loss": 0.4542,
+      "step": 5737
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.314368840181502e-06,
+      "loss": 0.4472,
+      "step": 5738
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3128988005526943e-06,
+      "loss": 0.4796,
+      "step": 5739
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.311429459205671e-06,
+      "loss": 0.4364,
+      "step": 5740
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3099608164187027e-06,
+      "loss": 0.4562,
+      "step": 5741
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.308492872469928e-06,
+      "loss": 0.4721,
+      "step": 5742
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.307025627637355e-06,
+      "loss": 0.4537,
+      "step": 5743
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3055590821988573e-06,
+      "loss": 0.4472,
+      "step": 5744
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3040932364321757e-06,
+      "loss": 0.4793,
+      "step": 5745
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3026280906149197e-06,
+      "loss": 0.48,
+      "step": 5746
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3011636450245657e-06,
+      "loss": 0.4781,
+      "step": 5747
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2996998999384591e-06,
+      "loss": 0.5289,
+      "step": 5748
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2982368556338105e-06,
+      "loss": 0.4637,
+      "step": 5749
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2967745123877002e-06,
+      "loss": 0.4658,
+      "step": 5750
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2953128704770706e-06,
+      "loss": 0.4796,
+      "step": 5751
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2938519301787372e-06,
+      "loss": 0.4619,
+      "step": 5752
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.29239169176938e-06,
+      "loss": 0.4855,
+      "step": 5753
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2909321555255461e-06,
+      "loss": 0.4502,
+      "step": 5754
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2894733217236495e-06,
+      "loss": 0.4583,
+      "step": 5755
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2880151906399735e-06,
+      "loss": 0.4072,
+      "step": 5756
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.286557762550663e-06,
+      "loss": 0.4405,
+      "step": 5757
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2851010377317353e-06,
+      "loss": 0.4637,
+      "step": 5758
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2836450164590714e-06,
+      "loss": 0.4607,
+      "step": 5759
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2821896990084203e-06,
+      "loss": 0.4799,
+      "step": 5760
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2807350856553985e-06,
+      "loss": 0.4554,
+      "step": 5761
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2792811766754854e-06,
+      "loss": 0.4542,
+      "step": 5762
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2778279723440312e-06,
+      "loss": 0.4626,
+      "step": 5763
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2763754729362521e-06,
+      "loss": 0.4472,
+      "step": 5764
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2749236787272267e-06,
+      "loss": 0.5004,
+      "step": 5765
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2734725899919037e-06,
+      "loss": 0.4485,
+      "step": 5766
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.272022207005102e-06,
+      "loss": 0.501,
+      "step": 5767
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2705725300414972e-06,
+      "loss": 0.4524,
+      "step": 5768
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.269123559375639e-06,
+      "loss": 0.4642,
+      "step": 5769
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2676752952819415e-06,
+      "loss": 0.465,
+      "step": 5770
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.266227738034681e-06,
+      "loss": 0.4732,
+      "step": 5771
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2647808879080042e-06,
+      "loss": 0.472,
+      "step": 5772
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.263334745175926e-06,
+      "loss": 0.4729,
+      "step": 5773
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.261889310112321e-06,
+      "loss": 0.489,
+      "step": 5774
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.260444582990934e-06,
+      "loss": 0.4422,
+      "step": 5775
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.259000564085376e-06,
+      "loss": 0.491,
+      "step": 5776
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2575572536691195e-06,
+      "loss": 0.4355,
+      "step": 5777
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.256114652015506e-06,
+      "loss": 0.5042,
+      "step": 5778
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.254672759397747e-06,
+      "loss": 0.4788,
+      "step": 5779
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2532315760889113e-06,
+      "loss": 0.4509,
+      "step": 5780
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2517911023619394e-06,
+      "loss": 0.4767,
+      "step": 5781
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2503513384896354e-06,
+      "loss": 0.4631,
+      "step": 5782
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2489122847446676e-06,
+      "loss": 0.4809,
+      "step": 5783
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2474739413995702e-06,
+      "loss": 0.4829,
+      "step": 5784
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.246036308726749e-06,
+      "loss": 0.4499,
+      "step": 5785
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2445993869984657e-06,
+      "loss": 0.4774,
+      "step": 5786
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.243163176486854e-06,
+      "loss": 0.4923,
+      "step": 5787
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2417276774639081e-06,
+      "loss": 0.4696,
+      "step": 5788
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.240292890201492e-06,
+      "loss": 0.5015,
+      "step": 5789
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.238858814971332e-06,
+      "loss": 0.4391,
+      "step": 5790
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2374254520450213e-06,
+      "loss": 0.4369,
+      "step": 5791
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.235992801694017e-06,
+      "loss": 0.4659,
+      "step": 5792
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2345608641896434e-06,
+      "loss": 0.4548,
+      "step": 5793
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.233129639803085e-06,
+      "loss": 0.4335,
+      "step": 5794
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2316991288053954e-06,
+      "loss": 0.4626,
+      "step": 5795
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2302693314674924e-06,
+      "loss": 0.4619,
+      "step": 5796
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2288402480601574e-06,
+      "loss": 0.4868,
+      "step": 5797
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2274118788540385e-06,
+      "loss": 0.462,
+      "step": 5798
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2259842241196484e-06,
+      "loss": 0.4773,
+      "step": 5799
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2245572841273607e-06,
+      "loss": 0.434,
+      "step": 5800
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2231310591474177e-06,
+      "loss": 0.4814,
+      "step": 5801
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.221705549449926e-06,
+      "loss": 0.4415,
+      "step": 5802
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.220280755304854e-06,
+      "loss": 0.4542,
+      "step": 5803
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2188566769820382e-06,
+      "loss": 0.4863,
+      "step": 5804
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.217433314751178e-06,
+      "loss": 0.4678,
+      "step": 5805
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2160106688818346e-06,
+      "loss": 0.4834,
+      "step": 5806
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2145887396434368e-06,
+      "loss": 0.4483,
+      "step": 5807
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2131675273052778e-06,
+      "loss": 0.441,
+      "step": 5808
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2117470321365126e-06,
+      "loss": 0.4755,
+      "step": 5809
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2103272544061623e-06,
+      "loss": 0.4536,
+      "step": 5810
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2089081943831133e-06,
+      "loss": 0.466,
+      "step": 5811
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.207489852336111e-06,
+      "loss": 0.4506,
+      "step": 5812
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2060722285337695e-06,
+      "loss": 0.4575,
+      "step": 5813
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2046553232445663e-06,
+      "loss": 0.4895,
+      "step": 5814
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2032391367368406e-06,
+      "loss": 0.448,
+      "step": 5815
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2018236692788005e-06,
+      "loss": 0.462,
+      "step": 5816
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2004089211385094e-06,
+      "loss": 0.4859,
+      "step": 5817
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1989948925839017e-06,
+      "loss": 0.4625,
+      "step": 5818
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1975815838827754e-06,
+      "loss": 0.4956,
+      "step": 5819
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1961689953027844e-06,
+      "loss": 0.5015,
+      "step": 5820
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1947571271114572e-06,
+      "loss": 0.505,
+      "step": 5821
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1933459795761799e-06,
+      "loss": 0.4681,
+      "step": 5822
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1919355529641996e-06,
+      "loss": 0.4888,
+      "step": 5823
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1905258475426323e-06,
+      "loss": 0.5005,
+      "step": 5824
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1891168635784562e-06,
+      "loss": 0.489,
+      "step": 5825
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.187708601338507e-06,
+      "loss": 0.4653,
+      "step": 5826
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1863010610894932e-06,
+      "loss": 0.4606,
+      "step": 5827
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.184894243097982e-06,
+      "loss": 0.4667,
+      "step": 5828
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1834881476304005e-06,
+      "loss": 0.4638,
+      "step": 5829
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1820827749530438e-06,
+      "loss": 0.413,
+      "step": 5830
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1806781253320692e-06,
+      "loss": 0.4487,
+      "step": 5831
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1792741990334934e-06,
+      "loss": 0.4924,
+      "step": 5832
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1778709963232022e-06,
+      "loss": 0.4423,
+      "step": 5833
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1764685174669416e-06,
+      "loss": 0.4354,
+      "step": 5834
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.175066762730317e-06,
+      "loss": 0.4316,
+      "step": 5835
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1736657323788015e-06,
+      "loss": 0.4732,
+      "step": 5836
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.17226542667773e-06,
+      "loss": 0.4929,
+      "step": 5837
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.170865845892296e-06,
+      "loss": 0.4695,
+      "step": 5838
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.169466990287565e-06,
+      "loss": 0.5035,
+      "step": 5839
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1680688601284544e-06,
+      "loss": 0.5005,
+      "step": 5840
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.166671455679751e-06,
+      "loss": 0.4439,
+      "step": 5841
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.165274777206103e-06,
+      "loss": 0.4986,
+      "step": 5842
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1638788249720184e-06,
+      "loss": 0.4918,
+      "step": 5843
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1624835992418705e-06,
+      "loss": 0.436,
+      "step": 5844
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.161089100279894e-06,
+      "loss": 0.474,
+      "step": 5845
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1596953283501872e-06,
+      "loss": 0.4549,
+      "step": 5846
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1583022837167086e-06,
+      "loss": 0.4673,
+      "step": 5847
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.156909966643282e-06,
+      "loss": 0.4916,
+      "step": 5848
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1555183773935879e-06,
+      "loss": 0.457,
+      "step": 5849
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.154127516231175e-06,
+      "loss": 0.4713,
+      "step": 5850
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1527373834194505e-06,
+      "loss": 0.4716,
+      "step": 5851
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1513479792216852e-06,
+      "loss": 0.4688,
+      "step": 5852
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.149959303901012e-06,
+      "loss": 0.4618,
+      "step": 5853
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1485713577204255e-06,
+      "loss": 0.4946,
+      "step": 5854
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1471841409427797e-06,
+      "loss": 0.4591,
+      "step": 5855
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1457976538307946e-06,
+      "loss": 0.4605,
+      "step": 5856
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.144411896647049e-06,
+      "loss": 0.4513,
+      "step": 5857
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1430268696539853e-06,
+      "loss": 0.4801,
+      "step": 5858
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1416425731139075e-06,
+      "loss": 0.4817,
+      "step": 5859
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1402590072889807e-06,
+      "loss": 0.4761,
+      "step": 5860
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1388761724412296e-06,
+      "loss": 0.445,
+      "step": 5861
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1374940688325432e-06,
+      "loss": 0.4463,
+      "step": 5862
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1361126967246715e-06,
+      "loss": 0.4694,
+      "step": 5863
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1347320563792263e-06,
+      "loss": 0.5065,
+      "step": 5864
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1333521480576815e-06,
+      "loss": 0.4623,
+      "step": 5865
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1319729720213674e-06,
+      "loss": 0.4587,
+      "step": 5866
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1305945285314813e-06,
+      "loss": 0.4747,
+      "step": 5867
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1292168178490798e-06,
+      "loss": 0.4502,
+      "step": 5868
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1278398402350804e-06,
+      "loss": 0.4531,
+      "step": 5869
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1264635959502622e-06,
+      "loss": 0.4725,
+      "step": 5870
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.125088085255267e-06,
+      "loss": 0.4263,
+      "step": 5871
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1237133084105928e-06,
+      "loss": 0.4498,
+      "step": 5872
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.122339265676603e-06,
+      "loss": 0.4732,
+      "step": 5873
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1209659573135212e-06,
+      "loss": 0.457,
+      "step": 5874
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.119593383581431e-06,
+      "loss": 0.4713,
+      "step": 5875
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1182215447402777e-06,
+      "loss": 0.4639,
+      "step": 5876
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1168504410498682e-06,
+      "loss": 0.5048,
+      "step": 5877
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1154800727698662e-06,
+      "loss": 0.4551,
+      "step": 5878
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1141104401598008e-06,
+      "loss": 0.4485,
+      "step": 5879
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.112741543479059e-06,
+      "loss": 0.4698,
+      "step": 5880
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.11137338298689e-06,
+      "loss": 0.4913,
+      "step": 5881
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.110005958942403e-06,
+      "loss": 0.458,
+      "step": 5882
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1086392716045686e-06,
+      "loss": 0.4719,
+      "step": 5883
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1072733212322146e-06,
+      "loss": 0.536,
+      "step": 5884
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1059081080840323e-06,
+      "loss": 0.4424,
+      "step": 5885
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1045436324185733e-06,
+      "loss": 0.4765,
+      "step": 5886
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.103179894494249e-06,
+      "loss": 0.4657,
+      "step": 5887
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1018168945693303e-06,
+      "loss": 0.4453,
+      "step": 5888
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1004546329019506e-06,
+      "loss": 0.4467,
+      "step": 5889
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.099093109750099e-06,
+      "loss": 0.4486,
+      "step": 5890
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.097732325371631e-06,
+      "loss": 0.431,
+      "step": 5891
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0963722800242555e-06,
+      "loss": 0.485,
+      "step": 5892
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0950129739655446e-06,
+      "loss": 0.4341,
+      "step": 5893
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0936544074529349e-06,
+      "loss": 0.478,
+      "step": 5894
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0922965807437142e-06,
+      "loss": 0.4502,
+      "step": 5895
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0909394940950364e-06,
+      "loss": 0.4312,
+      "step": 5896
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0895831477639145e-06,
+      "loss": 0.4439,
+      "step": 5897
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0882275420072175e-06,
+      "loss": 0.506,
+      "step": 5898
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0868726770816762e-06,
+      "loss": 0.439,
+      "step": 5899
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0855185532438871e-06,
+      "loss": 0.4638,
+      "step": 5900
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0841651707502959e-06,
+      "loss": 0.4667,
+      "step": 5901
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.082812529857215e-06,
+      "loss": 0.4506,
+      "step": 5902
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0814606308208152e-06,
+      "loss": 0.4355,
+      "step": 5903
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0801094738971235e-06,
+      "loss": 0.4421,
+      "step": 5904
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0787590593420283e-06,
+      "loss": 0.4436,
+      "step": 5905
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0774093874112829e-06,
+      "loss": 0.4759,
+      "step": 5906
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.07606045836049e-06,
+      "loss": 0.461,
+      "step": 5907
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0747122724451187e-06,
+      "loss": 0.454,
+      "step": 5908
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0733648299204963e-06,
+      "loss": 0.4735,
+      "step": 5909
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0720181310418048e-06,
+      "loss": 0.4709,
+      "step": 5910
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0706721760640897e-06,
+      "loss": 0.4534,
+      "step": 5911
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0693269652422588e-06,
+      "loss": 0.4884,
+      "step": 5912
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0679824988310704e-06,
+      "loss": 0.4888,
+      "step": 5913
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0666387770851478e-06,
+      "loss": 0.4737,
+      "step": 5914
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0652958002589731e-06,
+      "loss": 0.4717,
+      "step": 5915
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0639535686068837e-06,
+      "loss": 0.4546,
+      "step": 5916
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0626120823830787e-06,
+      "loss": 0.4874,
+      "step": 5917
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0612713418416164e-06,
+      "loss": 0.4789,
+      "step": 5918
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0599313472364125e-06,
+      "loss": 0.4992,
+      "step": 5919
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.058592098821244e-06,
+      "loss": 0.4686,
+      "step": 5920
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0572535968497405e-06,
+      "loss": 0.477,
+      "step": 5921
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0559158415753968e-06,
+      "loss": 0.4843,
+      "step": 5922
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0545788332515627e-06,
+      "loss": 0.4486,
+      "step": 5923
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0532425721314483e-06,
+      "loss": 0.4827,
+      "step": 5924
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0519070584681213e-06,
+      "loss": 0.4948,
+      "step": 5925
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0505722925145096e-06,
+      "loss": 0.4779,
+      "step": 5926
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0492382745233948e-06,
+      "loss": 0.5021,
+      "step": 5927
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.047905004747421e-06,
+      "loss": 0.485,
+      "step": 5928
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0465724834390901e-06,
+      "loss": 0.4848,
+      "step": 5929
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0452407108507616e-06,
+      "loss": 0.4782,
+      "step": 5930
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0439096872346533e-06,
+      "loss": 0.4509,
+      "step": 5931
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0425794128428423e-06,
+      "loss": 0.4476,
+      "step": 5932
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0412498879272598e-06,
+      "loss": 0.4639,
+      "step": 5933
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0399211127397002e-06,
+      "loss": 0.4602,
+      "step": 5934
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0385930875318123e-06,
+      "loss": 0.4893,
+      "step": 5935
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0372658125551054e-06,
+      "loss": 0.4983,
+      "step": 5936
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0359392880609448e-06,
+      "loss": 0.4689,
+      "step": 5937
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.034613514300556e-06,
+      "loss": 0.4585,
+      "step": 5938
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.033288491525018e-06,
+      "loss": 0.4504,
+      "step": 5939
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0319642199852714e-06,
+      "loss": 0.4782,
+      "step": 5940
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0306406999321135e-06,
+      "loss": 0.4738,
+      "step": 5941
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0293179316161995e-06,
+      "loss": 0.4451,
+      "step": 5942
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0279959152880431e-06,
+      "loss": 0.5014,
+      "step": 5943
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0266746511980114e-06,
+      "loss": 0.4382,
+      "step": 5944
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.025354139596334e-06,
+      "loss": 0.4271,
+      "step": 5945
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0240343807330965e-06,
+      "loss": 0.4552,
+      "step": 5946
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0227153748582386e-06,
+      "loss": 0.4785,
+      "step": 5947
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0213971222215629e-06,
+      "loss": 0.4754,
+      "step": 5948
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0200796230727284e-06,
+      "loss": 0.4748,
+      "step": 5949
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0187628776612458e-06,
+      "loss": 0.4888,
+      "step": 5950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0174468862364885e-06,
+      "loss": 0.4876,
+      "step": 5951
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.016131649047687e-06,
+      "loss": 0.4825,
+      "step": 5952
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0148171663439233e-06,
+      "loss": 0.4862,
+      "step": 5953
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0135034383741454e-06,
+      "loss": 0.4462,
+      "step": 5954
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0121904653871534e-06,
+      "loss": 0.4741,
+      "step": 5955
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.010878247631602e-06,
+      "loss": 0.4564,
+      "step": 5956
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.009566785356007e-06,
+      "loss": 0.4784,
+      "step": 5957
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0082560788087414e-06,
+      "loss": 0.4497,
+      "step": 5958
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0069461282380282e-06,
+      "loss": 0.4391,
+      "step": 5959
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.005636933891958e-06,
+      "loss": 0.4784,
+      "step": 5960
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0043284960184713e-06,
+      "loss": 0.4847,
+      "step": 5961
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0030208148653647e-06,
+      "loss": 0.4886,
+      "step": 5962
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0017138906802948e-06,
+      "loss": 0.4588,
+      "step": 5963
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0004077237107751e-06,
+      "loss": 0.4551,
+      "step": 5964
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.991023142041688e-07,
+      "loss": 0.4754,
+      "step": 5965
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.977976624077074e-07,
+      "loss": 0.4463,
+      "step": 5966
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.964937685684683e-07,
+      "loss": 0.4756,
+      "step": 5967
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.951906329333904e-07,
+      "loss": 0.4938,
+      "step": 5968
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.938882557492697e-07,
+      "loss": 0.4348,
+      "step": 5969
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.92586637262754e-07,
+      "loss": 0.4842,
+      "step": 5970
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.91285777720352e-07,
+      "loss": 0.4873,
+      "step": 5971
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.899856773684268e-07,
+      "loss": 0.4632,
+      "step": 5972
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.886863364531984e-07,
+      "loss": 0.4813,
+      "step": 5973
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.873877552207423e-07,
+      "loss": 0.4779,
+      "step": 5974
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.860899339169915e-07,
+      "loss": 0.4775,
+      "step": 5975
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.84792872787731e-07,
+      "loss": 0.4958,
+      "step": 5976
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.834965720786066e-07,
+      "loss": 0.4793,
+      "step": 5977
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.822010320351182e-07,
+      "loss": 0.4628,
+      "step": 5978
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.80906252902622e-07,
+      "loss": 0.4705,
+      "step": 5979
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.796122349263288e-07,
+      "loss": 0.4442,
+      "step": 5980
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.78318978351308e-07,
+      "loss": 0.4579,
+      "step": 5981
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.770264834224814e-07,
+      "loss": 0.4608,
+      "step": 5982
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.757347503846282e-07,
+      "loss": 0.4377,
+      "step": 5983
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.744437794823836e-07,
+      "loss": 0.4564,
+      "step": 5984
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.731535709602385e-07,
+      "loss": 0.4394,
+      "step": 5985
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.718641250625387e-07,
+      "loss": 0.4592,
+      "step": 5986
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.705754420334874e-07,
+      "loss": 0.4569,
+      "step": 5987
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.6928752211714e-07,
+      "loss": 0.4625,
+      "step": 5988
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.680003655574093e-07,
+      "loss": 0.4716,
+      "step": 5989
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.66713972598064e-07,
+      "loss": 0.4343,
+      "step": 5990
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.654283434827282e-07,
+      "loss": 0.4737,
+      "step": 5991
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.641434784548814e-07,
+      "loss": 0.4758,
+      "step": 5992
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.628593777578559e-07,
+      "loss": 0.4487,
+      "step": 5993
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.61576041634842e-07,
+      "loss": 0.481,
+      "step": 5994
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.602934703288836e-07,
+      "loss": 0.4795,
+      "step": 5995
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.590116640828822e-07,
+      "loss": 0.4355,
+      "step": 5996
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.57730623139591e-07,
+      "loss": 0.4718,
+      "step": 5997
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.564503477416231e-07,
+      "loss": 0.4409,
+      "step": 5998
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.55170838131439e-07,
+      "loss": 0.456,
+      "step": 5999
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.538920945513613e-07,
+      "loss": 0.4715,
+      "step": 6000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.526141172435644e-07,
+      "loss": 0.468,
+      "step": 6001
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.513369064500783e-07,
+      "loss": 0.4793,
+      "step": 6002
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.500604624127874e-07,
+      "loss": 0.4886,
+      "step": 6003
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.487847853734323e-07,
+      "loss": 0.4502,
+      "step": 6004
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.475098755736051e-07,
+      "loss": 0.4554,
+      "step": 6005
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.462357332547551e-07,
+      "loss": 0.44,
+      "step": 6006
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.449623586581863e-07,
+      "loss": 0.4701,
+      "step": 6007
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.436897520250566e-07,
+      "loss": 0.4548,
+      "step": 6008
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.42417913596379e-07,
+      "loss": 0.4969,
+      "step": 6009
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.411468436130216e-07,
+      "loss": 0.5093,
+      "step": 6010
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.398765423157036e-07,
+      "loss": 0.4626,
+      "step": 6011
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.386070099450017e-07,
+      "loss": 0.4556,
+      "step": 6012
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.373382467413477e-07,
+      "loss": 0.4426,
+      "step": 6013
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.360702529450256e-07,
+      "loss": 0.4332,
+      "step": 6014
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.34803028796174e-07,
+      "loss": 0.4636,
+      "step": 6015
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.33536574534788e-07,
+      "loss": 0.4436,
+      "step": 6016
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.322708904007127e-07,
+      "loss": 0.4812,
+      "step": 6017
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.310059766336515e-07,
+      "loss": 0.4695,
+      "step": 6018
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.297418334731573e-07,
+      "loss": 0.4799,
+      "step": 6019
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.284784611586433e-07,
+      "loss": 0.4905,
+      "step": 6020
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.272158599293729e-07,
+      "loss": 0.4738,
+      "step": 6021
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.259540300244618e-07,
+      "loss": 0.469,
+      "step": 6022
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.246929716828829e-07,
+      "loss": 0.493,
+      "step": 6023
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.234326851434632e-07,
+      "loss": 0.4715,
+      "step": 6024
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.221731706448789e-07,
+      "loss": 0.4594,
+      "step": 6025
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.209144284256643e-07,
+      "loss": 0.47,
+      "step": 6026
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.19656458724209e-07,
+      "loss": 0.4729,
+      "step": 6027
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.183992617787507e-07,
+      "loss": 0.4784,
+      "step": 6028
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.171428378273839e-07,
+      "loss": 0.4864,
+      "step": 6029
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.158871871080594e-07,
+      "loss": 0.4819,
+      "step": 6030
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.146323098585746e-07,
+      "loss": 0.4666,
+      "step": 6031
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.133782063165846e-07,
+      "loss": 0.4143,
+      "step": 6032
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.121248767196028e-07,
+      "loss": 0.4445,
+      "step": 6033
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.108723213049864e-07,
+      "loss": 0.4698,
+      "step": 6034
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.096205403099522e-07,
+      "loss": 0.4648,
+      "step": 6035
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.083695339715698e-07,
+      "loss": 0.4344,
+      "step": 6036
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.071193025267582e-07,
+      "loss": 0.4371,
+      "step": 6037
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.05869846212294e-07,
+      "loss": 0.4899,
+      "step": 6038
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.046211652648074e-07,
+      "loss": 0.487,
+      "step": 6039
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.033732599207773e-07,
+      "loss": 0.5116,
+      "step": 6040
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.021261304165391e-07,
+      "loss": 0.468,
+      "step": 6041
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.008797769882821e-07,
+      "loss": 0.4934,
+      "step": 6042
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.996341998720442e-07,
+      "loss": 0.4448,
+      "step": 6043
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.983893993037201e-07,
+      "loss": 0.4551,
+      "step": 6044
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.971453755190574e-07,
+      "loss": 0.4518,
+      "step": 6045
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.95902128753654e-07,
+      "loss": 0.4933,
+      "step": 6046
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.946596592429646e-07,
+      "loss": 0.4822,
+      "step": 6047
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.93417967222292e-07,
+      "loss": 0.5035,
+      "step": 6048
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.921770529267942e-07,
+      "loss": 0.4771,
+      "step": 6049
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.909369165914833e-07,
+      "loss": 0.4624,
+      "step": 6050
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.896975584512208e-07,
+      "loss": 0.4626,
+      "step": 6051
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.884589787407244e-07,
+      "loss": 0.4656,
+      "step": 6052
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.872211776945622e-07,
+      "loss": 0.4835,
+      "step": 6053
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.859841555471538e-07,
+      "loss": 0.4674,
+      "step": 6054
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.847479125327735e-07,
+      "loss": 0.4656,
+      "step": 6055
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.835124488855479e-07,
+      "loss": 0.4852,
+      "step": 6056
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.822777648394537e-07,
+      "loss": 0.4908,
+      "step": 6057
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.810438606283234e-07,
+      "loss": 0.4777,
+      "step": 6058
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.798107364858399e-07,
+      "loss": 0.4652,
+      "step": 6059
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.785783926455365e-07,
+      "loss": 0.4722,
+      "step": 6060
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.773468293408022e-07,
+      "loss": 0.5072,
+      "step": 6061
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.761160468048763e-07,
+      "loss": 0.4789,
+      "step": 6062
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.748860452708507e-07,
+      "loss": 0.4601,
+      "step": 6063
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.736568249716693e-07,
+      "loss": 0.4947,
+      "step": 6064
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.724283861401295e-07,
+      "loss": 0.4654,
+      "step": 6065
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.712007290088759e-07,
+      "loss": 0.4766,
+      "step": 6066
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.699738538104103e-07,
+      "loss": 0.4621,
+      "step": 6067
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.687477607770844e-07,
+      "loss": 0.4754,
+      "step": 6068
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.675224501411022e-07,
+      "loss": 0.4551,
+      "step": 6069
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.662979221345197e-07,
+      "loss": 0.4752,
+      "step": 6070
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.650741769892424e-07,
+      "loss": 0.4488,
+      "step": 6071
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.638512149370304e-07,
+      "loss": 0.4365,
+      "step": 6072
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.626290362094936e-07,
+      "loss": 0.4643,
+      "step": 6073
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.614076410380956e-07,
+      "loss": 0.4517,
+      "step": 6074
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.601870296541498e-07,
+      "loss": 0.4739,
+      "step": 6075
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.589672022888223e-07,
+      "loss": 0.4584,
+      "step": 6076
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.57748159173129e-07,
+      "loss": 0.4572,
+      "step": 6077
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.565299005379391e-07,
+      "loss": 0.4629,
+      "step": 6078
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.553124266139734e-07,
+      "loss": 0.4465,
+      "step": 6079
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.540957376317993e-07,
+      "loss": 0.4694,
+      "step": 6080
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.528798338218441e-07,
+      "loss": 0.4857,
+      "step": 6081
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.516647154143815e-07,
+      "loss": 0.4931,
+      "step": 6082
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.504503826395344e-07,
+      "loss": 0.4523,
+      "step": 6083
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.492368357272801e-07,
+      "loss": 0.4739,
+      "step": 6084
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.48024074907447e-07,
+      "loss": 0.4958,
+      "step": 6085
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.468121004097113e-07,
+      "loss": 0.4733,
+      "step": 6086
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.45600912463606e-07,
+      "loss": 0.4851,
+      "step": 6087
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.443905112985118e-07,
+      "loss": 0.4664,
+      "step": 6088
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.431808971436578e-07,
+      "loss": 0.4686,
+      "step": 6089
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.419720702281292e-07,
+      "loss": 0.4362,
+      "step": 6090
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.407640307808601e-07,
+      "loss": 0.4527,
+      "step": 6091
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.395567790306314e-07,
+      "loss": 0.4467,
+      "step": 6092
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.383503152060823e-07,
+      "loss": 0.4862,
+      "step": 6093
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.371446395356991e-07,
+      "loss": 0.4611,
+      "step": 6094
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.359397522478163e-07,
+      "loss": 0.4808,
+      "step": 6095
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.347356535706236e-07,
+      "loss": 0.4677,
+      "step": 6096
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.335323437321574e-07,
+      "loss": 0.4856,
+      "step": 6097
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.323298229603072e-07,
+      "loss": 0.4777,
+      "step": 6098
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.311280914828129e-07,
+      "loss": 0.4399,
+      "step": 6099
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.299271495272642e-07,
+      "loss": 0.5085,
+      "step": 6100
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.28726997321102e-07,
+      "loss": 0.468,
+      "step": 6101
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.275276350916178e-07,
+      "loss": 0.4663,
+      "step": 6102
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.263290630659504e-07,
+      "loss": 0.4687,
+      "step": 6103
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.251312814710933e-07,
+      "loss": 0.4661,
+      "step": 6104
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.239342905338877e-07,
+      "loss": 0.4616,
+      "step": 6105
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.227380904810256e-07,
+      "loss": 0.4737,
+      "step": 6106
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.215426815390504e-07,
+      "loss": 0.4734,
+      "step": 6107
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.203480639343552e-07,
+      "loss": 0.4515,
+      "step": 6108
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.191542378931805e-07,
+      "loss": 0.4707,
+      "step": 6109
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.179612036416195e-07,
+      "loss": 0.4515,
+      "step": 6110
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.167689614056162e-07,
+      "loss": 0.4824,
+      "step": 6111
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.155775114109632e-07,
+      "loss": 0.483,
+      "step": 6112
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.143868538833027e-07,
+      "loss": 0.4192,
+      "step": 6113
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.131969890481289e-07,
+      "loss": 0.4672,
+      "step": 6114
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.12007917130782e-07,
+      "loss": 0.4581,
+      "step": 6115
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.108196383564554e-07,
+      "loss": 0.4418,
+      "step": 6116
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.096321529501921e-07,
+      "loss": 0.4813,
+      "step": 6117
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.084454611368831e-07,
+      "loss": 0.4685,
+      "step": 6118
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.072595631412705e-07,
+      "loss": 0.4758,
+      "step": 6119
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.060744591879472e-07,
+      "loss": 0.4626,
+      "step": 6120
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.048901495013511e-07,
+      "loss": 0.4884,
+      "step": 6121
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.037066343057742e-07,
+      "loss": 0.4775,
+      "step": 6122
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.025239138253571e-07,
+      "loss": 0.484,
+      "step": 6123
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.013419882840884e-07,
+      "loss": 0.4768,
+      "step": 6124
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.001608579058084e-07,
+      "loss": 0.4817,
+      "step": 6125
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.989805229142039e-07,
+      "loss": 0.4542,
+      "step": 6126
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.978009835328138e-07,
+      "loss": 0.4713,
+      "step": 6127
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.966222399850243e-07,
+      "loss": 0.4651,
+      "step": 6128
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.954442924940725e-07,
+      "loss": 0.4641,
+      "step": 6129
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.942671412830439e-07,
+      "loss": 0.4533,
+      "step": 6130
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.930907865748744e-07,
+      "loss": 0.4646,
+      "step": 6131
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.919152285923453e-07,
+      "loss": 0.4648,
+      "step": 6132
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.907404675580915e-07,
+      "loss": 0.4818,
+      "step": 6133
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.895665036945955e-07,
+      "loss": 0.4669,
+      "step": 6134
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.88393337224187e-07,
+      "loss": 0.4905,
+      "step": 6135
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.872209683690474e-07,
+      "loss": 0.4981,
+      "step": 6136
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.860493973512068e-07,
+      "loss": 0.4658,
+      "step": 6137
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.848786243925404e-07,
+      "loss": 0.4793,
+      "step": 6138
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.837086497147766e-07,
+      "loss": 0.4417,
+      "step": 6139
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.825394735394914e-07,
+      "loss": 0.4961,
+      "step": 6140
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.813710960881088e-07,
+      "loss": 0.4825,
+      "step": 6141
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.802035175819022e-07,
+      "loss": 0.4213,
+      "step": 6142
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.790367382419944e-07,
+      "loss": 0.466,
+      "step": 6143
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.778707582893541e-07,
+      "loss": 0.4576,
+      "step": 6144
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.767055779448007e-07,
+      "loss": 0.4675,
+      "step": 6145
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.755411974290028e-07,
+      "loss": 0.4938,
+      "step": 6146
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.743776169624767e-07,
+      "loss": 0.467,
+      "step": 6147
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.732148367655879e-07,
+      "loss": 0.4399,
+      "step": 6148
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.720528570585473e-07,
+      "loss": 0.4242,
+      "step": 6149
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.708916780614173e-07,
+      "loss": 0.4888,
+      "step": 6150
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.697312999941098e-07,
+      "loss": 0.4602,
+      "step": 6151
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.685717230763795e-07,
+      "loss": 0.4677,
+      "step": 6152
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.674129475278341e-07,
+      "loss": 0.456,
+      "step": 6153
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.66254973567932e-07,
+      "loss": 0.4392,
+      "step": 6154
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.650978014159711e-07,
+      "loss": 0.4629,
+      "step": 6155
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.639414312911047e-07,
+      "loss": 0.4717,
+      "step": 6156
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.627858634123336e-07,
+      "loss": 0.4666,
+      "step": 6157
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.61631097998502e-07,
+      "loss": 0.4861,
+      "step": 6158
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.604771352683049e-07,
+      "loss": 0.4545,
+      "step": 6159
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.593239754402898e-07,
+      "loss": 0.4559,
+      "step": 6160
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.581716187328442e-07,
+      "loss": 0.4676,
+      "step": 6161
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.570200653642085e-07,
+      "loss": 0.4828,
+      "step": 6162
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.558693155524705e-07,
+      "loss": 0.4659,
+      "step": 6163
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.54719369515563e-07,
+      "loss": 0.4405,
+      "step": 6164
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.53570227471268e-07,
+      "loss": 0.4494,
+      "step": 6165
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.524218896372198e-07,
+      "loss": 0.4739,
+      "step": 6166
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.512743562308928e-07,
+      "loss": 0.455,
+      "step": 6167
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.50127627469614e-07,
+      "loss": 0.4684,
+      "step": 6168
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.489817035705576e-07,
+      "loss": 0.4556,
+      "step": 6169
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.478365847507418e-07,
+      "loss": 0.4705,
+      "step": 6170
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.466922712270358e-07,
+      "loss": 0.4471,
+      "step": 6171
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.455487632161584e-07,
+      "loss": 0.4868,
+      "step": 6172
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.444060609346693e-07,
+      "loss": 0.4703,
+      "step": 6173
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.432641645989819e-07,
+      "loss": 0.4955,
+      "step": 6174
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.421230744253516e-07,
+      "loss": 0.4551,
+      "step": 6175
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.40982790629885e-07,
+      "loss": 0.4499,
+      "step": 6176
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.398433134285349e-07,
+      "loss": 0.4622,
+      "step": 6177
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.387046430371014e-07,
+      "loss": 0.4673,
+      "step": 6178
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.375667796712316e-07,
+      "loss": 0.5079,
+      "step": 6179
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.364297235464207e-07,
+      "loss": 0.4739,
+      "step": 6180
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.352934748780077e-07,
+      "loss": 0.4772,
+      "step": 6181
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.341580338811827e-07,
+      "loss": 0.4718,
+      "step": 6182
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.330234007709808e-07,
+      "loss": 0.4422,
+      "step": 6183
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.318895757622851e-07,
+      "loss": 0.4775,
+      "step": 6184
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.30756559069824e-07,
+      "loss": 0.4473,
+      "step": 6185
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.296243509081763e-07,
+      "loss": 0.4796,
+      "step": 6186
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.284929514917627e-07,
+      "loss": 0.4454,
+      "step": 6187
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.273623610348535e-07,
+      "loss": 0.4698,
+      "step": 6188
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.262325797515668e-07,
+      "loss": 0.4565,
+      "step": 6189
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.25103607855866e-07,
+      "loss": 0.4702,
+      "step": 6190
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.239754455615606e-07,
+      "loss": 0.4663,
+      "step": 6191
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.228480930823101e-07,
+      "loss": 0.4872,
+      "step": 6192
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.217215506316149e-07,
+      "loss": 0.4236,
+      "step": 6193
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.205958184228268e-07,
+      "loss": 0.4356,
+      "step": 6194
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.194708966691427e-07,
+      "loss": 0.4589,
+      "step": 6195
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.183467855836058e-07,
+      "loss": 0.4731,
+      "step": 6196
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.172234853791061e-07,
+      "loss": 0.462,
+      "step": 6197
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.161009962683807e-07,
+      "loss": 0.44,
+      "step": 6198
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.149793184640102e-07,
+      "loss": 0.4614,
+      "step": 6199
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.138584521784247e-07,
+      "loss": 0.4821,
+      "step": 6200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.127383976238994e-07,
+      "loss": 0.4754,
+      "step": 6201
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.116191550125562e-07,
+      "loss": 0.473,
+      "step": 6202
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.105007245563633e-07,
+      "loss": 0.4655,
+      "step": 6203
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.09383106467133e-07,
+      "loss": 0.4678,
+      "step": 6204
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.082663009565266e-07,
+      "loss": 0.4744,
+      "step": 6205
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.071503082360515e-07,
+      "loss": 0.4725,
+      "step": 6206
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.060351285170559e-07,
+      "loss": 0.484,
+      "step": 6207
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.049207620107423e-07,
+      "loss": 0.4596,
+      "step": 6208
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.03807208928155e-07,
+      "loss": 0.4722,
+      "step": 6209
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.026944694801818e-07,
+      "loss": 0.4714,
+      "step": 6210
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.015825438775603e-07,
+      "loss": 0.4908,
+      "step": 6211
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.004714323308731e-07,
+      "loss": 0.4792,
+      "step": 6212
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.993611350505453e-07,
+      "loss": 0.4756,
+      "step": 6213
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.982516522468536e-07,
+      "loss": 0.4721,
+      "step": 6214
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.971429841299176e-07,
+      "loss": 0.4583,
+      "step": 6215
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.960351309097007e-07,
+      "loss": 0.4646,
+      "step": 6216
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.949280927960139e-07,
+      "loss": 0.4488,
+      "step": 6217
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.938218699985155e-07,
+      "loss": 0.4816,
+      "step": 6218
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.927164627267036e-07,
+      "loss": 0.4692,
+      "step": 6219
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.916118711899295e-07,
+      "loss": 0.4559,
+      "step": 6220
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.905080955973864e-07,
+      "loss": 0.4682,
+      "step": 6221
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.894051361581106e-07,
+      "loss": 0.4856,
+      "step": 6222
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.883029930809876e-07,
+      "loss": 0.4678,
+      "step": 6223
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.87201666574746e-07,
+      "loss": 0.4555,
+      "step": 6224
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.861011568479592e-07,
+      "loss": 0.4879,
+      "step": 6225
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.850014641090513e-07,
+      "loss": 0.4831,
+      "step": 6226
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.839025885662842e-07,
+      "loss": 0.4815,
+      "step": 6227
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.828045304277697e-07,
+      "loss": 0.4709,
+      "step": 6228
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.817072899014648e-07,
+      "loss": 0.4879,
+      "step": 6229
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.806108671951677e-07,
+      "loss": 0.4582,
+      "step": 6230
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.795152625165258e-07,
+      "loss": 0.4975,
+      "step": 6231
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.784204760730306e-07,
+      "loss": 0.4382,
+      "step": 6232
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.773265080720176e-07,
+      "loss": 0.455,
+      "step": 6233
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.762333587206682e-07,
+      "loss": 0.4468,
+      "step": 6234
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.751410282260102e-07,
+      "loss": 0.4717,
+      "step": 6235
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.740495167949118e-07,
+      "loss": 0.4756,
+      "step": 6236
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.729588246340901e-07,
+      "loss": 0.4853,
+      "step": 6237
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.71868951950106e-07,
+      "loss": 0.4556,
+      "step": 6238
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.707798989493647e-07,
+      "loss": 0.4981,
+      "step": 6239
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.696916658381175e-07,
+      "loss": 0.4685,
+      "step": 6240
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.686042528224596e-07,
+      "loss": 0.4936,
+      "step": 6241
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.675176601083288e-07,
+      "loss": 0.4351,
+      "step": 6242
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.664318879015114e-07,
+      "loss": 0.4403,
+      "step": 6243
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.653469364076348e-07,
+      "loss": 0.4554,
+      "step": 6244
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.642628058321743e-07,
+      "loss": 0.4606,
+      "step": 6245
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.631794963804467e-07,
+      "loss": 0.4838,
+      "step": 6246
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.620970082576161e-07,
+      "loss": 0.4721,
+      "step": 6247
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.610153416686877e-07,
+      "loss": 0.4626,
+      "step": 6248
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.59934496818514e-07,
+      "loss": 0.5113,
+      "step": 6249
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.588544739117903e-07,
+      "loss": 0.4754,
+      "step": 6250
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.577752731530573e-07,
+      "loss": 0.4274,
+      "step": 6251
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.566968947467e-07,
+      "loss": 0.483,
+      "step": 6252
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.556193388969456e-07,
+      "loss": 0.4466,
+      "step": 6253
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.54542605807868e-07,
+      "loss": 0.4643,
+      "step": 6254
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.53466695683384e-07,
+      "loss": 0.4532,
+      "step": 6255
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.523916087272552e-07,
+      "loss": 0.4578,
+      "step": 6256
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.513173451430871e-07,
+      "loss": 0.4465,
+      "step": 6257
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.502439051343295e-07,
+      "loss": 0.4537,
+      "step": 6258
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.491712889042745e-07,
+      "loss": 0.4729,
+      "step": 6259
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.4809949665606e-07,
+      "loss": 0.4614,
+      "step": 6260
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.47028528592668e-07,
+      "loss": 0.485,
+      "step": 6261
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.459583849169232e-07,
+      "loss": 0.4566,
+      "step": 6262
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.448890658314955e-07,
+      "loss": 0.4706,
+      "step": 6263
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.438205715388989e-07,
+      "loss": 0.4509,
+      "step": 6264
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.427529022414875e-07,
+      "loss": 0.4587,
+      "step": 6265
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.416860581414636e-07,
+      "loss": 0.501,
+      "step": 6266
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.406200394408707e-07,
+      "loss": 0.4599,
+      "step": 6267
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.395548463415974e-07,
+      "loss": 0.4566,
+      "step": 6268
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.384904790453755e-07,
+      "loss": 0.4319,
+      "step": 6269
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.374269377537812e-07,
+      "loss": 0.4273,
+      "step": 6270
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.363642226682304e-07,
+      "loss": 0.4674,
+      "step": 6271
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.353023339899877e-07,
+      "loss": 0.4648,
+      "step": 6272
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.342412719201579e-07,
+      "loss": 0.4934,
+      "step": 6273
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.331810366596908e-07,
+      "loss": 0.4622,
+      "step": 6274
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.321216284093801e-07,
+      "loss": 0.4797,
+      "step": 6275
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.310630473698598e-07,
+      "loss": 0.4562,
+      "step": 6276
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.300052937416101e-07,
+      "loss": 0.4407,
+      "step": 6277
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.289483677249547e-07,
+      "loss": 0.4784,
+      "step": 6278
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.278922695200573e-07,
+      "loss": 0.4595,
+      "step": 6279
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.268369993269291e-07,
+      "loss": 0.4679,
+      "step": 6280
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.257825573454229e-07,
+      "loss": 0.4838,
+      "step": 6281
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.247289437752324e-07,
+      "loss": 0.4563,
+      "step": 6282
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.236761588158968e-07,
+      "loss": 0.4676,
+      "step": 6283
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.226242026667995e-07,
+      "loss": 0.475,
+      "step": 6284
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.215730755271627e-07,
+      "loss": 0.484,
+      "step": 6285
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.205227775960538e-07,
+      "loss": 0.4403,
+      "step": 6286
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.194733090723876e-07,
+      "loss": 0.4847,
+      "step": 6287
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.184246701549141e-07,
+      "loss": 0.4804,
+      "step": 6288
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.173768610422303e-07,
+      "loss": 0.4544,
+      "step": 6289
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.163298819327779e-07,
+      "loss": 0.4798,
+      "step": 6290
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.152837330248357e-07,
+      "loss": 0.499,
+      "step": 6291
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.142384145165298e-07,
+      "loss": 0.469,
+      "step": 6292
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.131939266058295e-07,
+      "loss": 0.511,
+      "step": 6293
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.121502694905434e-07,
+      "loss": 0.4541,
+      "step": 6294
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.111074433683256e-07,
+      "loss": 0.4698,
+      "step": 6295
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.10065448436672e-07,
+      "loss": 0.4576,
+      "step": 6296
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.09024284892919e-07,
+      "loss": 0.4369,
+      "step": 6297
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.079839529342474e-07,
+      "loss": 0.4758,
+      "step": 6298
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.069444527576834e-07,
+      "loss": 0.4356,
+      "step": 6299
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.059057845600907e-07,
+      "loss": 0.4635,
+      "step": 6300
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.048679485381781e-07,
+      "loss": 0.4586,
+      "step": 6301
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.038309448884955e-07,
+      "loss": 0.4608,
+      "step": 6302
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.027947738074363e-07,
+      "loss": 0.4507,
+      "step": 6303
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.017594354912354e-07,
+      "loss": 0.4849,
+      "step": 6304
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.007249301359707e-07,
+      "loss": 0.4439,
+      "step": 6305
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.996912579375625e-07,
+      "loss": 0.4551,
+      "step": 6306
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.986584190917733e-07,
+      "loss": 0.4998,
+      "step": 6307
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.976264137942056e-07,
+      "loss": 0.4668,
+      "step": 6308
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.965952422403066e-07,
+      "loss": 0.461,
+      "step": 6309
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.955649046253648e-07,
+      "loss": 0.4658,
+      "step": 6310
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.945354011445104e-07,
+      "loss": 0.4672,
+      "step": 6311
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.935067319927162e-07,
+      "loss": 0.4669,
+      "step": 6312
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.924788973647983e-07,
+      "loss": 0.4826,
+      "step": 6313
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.914518974554106e-07,
+      "loss": 0.4759,
+      "step": 6314
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.904257324590523e-07,
+      "loss": 0.4515,
+      "step": 6315
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.89400402570064e-07,
+      "loss": 0.4792,
+      "step": 6316
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.883759079826279e-07,
+      "loss": 0.4366,
+      "step": 6317
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.873522488907679e-07,
+      "loss": 0.4468,
+      "step": 6318
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.863294254883511e-07,
+      "loss": 0.4534,
+      "step": 6319
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.85307437969082e-07,
+      "loss": 0.4899,
+      "step": 6320
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.842862865265114e-07,
+      "loss": 0.4604,
+      "step": 6321
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.8326597135403e-07,
+      "loss": 0.4815,
+      "step": 6322
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.822464926448706e-07,
+      "loss": 0.4849,
+      "step": 6323
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.812278505921065e-07,
+      "loss": 0.4562,
+      "step": 6324
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.802100453886545e-07,
+      "loss": 0.4843,
+      "step": 6325
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.791930772272697e-07,
+      "loss": 0.4681,
+      "step": 6326
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.781769463005516e-07,
+      "loss": 0.4546,
+      "step": 6327
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.771616528009405e-07,
+      "loss": 0.4688,
+      "step": 6328
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.761471969207172e-07,
+      "loss": 0.4582,
+      "step": 6329
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.751335788520062e-07,
+      "loss": 0.4604,
+      "step": 6330
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.741207987867692e-07,
+      "loss": 0.4578,
+      "step": 6331
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.731088569168125e-07,
+      "loss": 0.4412,
+      "step": 6332
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.72097753433784e-07,
+      "loss": 0.4601,
+      "step": 6333
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.710874885291684e-07,
+      "loss": 0.4791,
+      "step": 6334
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.700780623942975e-07,
+      "loss": 0.4789,
+      "step": 6335
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.690694752203419e-07,
+      "loss": 0.4855,
+      "step": 6336
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.680617271983107e-07,
+      "loss": 0.4456,
+      "step": 6337
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.670548185190577e-07,
+      "loss": 0.4826,
+      "step": 6338
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.660487493732769e-07,
+      "loss": 0.4716,
+      "step": 6339
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.650435199514997e-07,
+      "loss": 0.4568,
+      "step": 6340
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.640391304441051e-07,
+      "loss": 0.4533,
+      "step": 6341
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.630355810413085e-07,
+      "loss": 0.4661,
+      "step": 6342
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.620328719331658e-07,
+      "loss": 0.4413,
+      "step": 6343
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.610310033095756e-07,
+      "loss": 0.4654,
+      "step": 6344
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.600299753602789e-07,
+      "loss": 0.481,
+      "step": 6345
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.590297882748513e-07,
+      "loss": 0.4687,
+      "step": 6346
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.580304422427163e-07,
+      "loss": 0.4394,
+      "step": 6347
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.570319374531358e-07,
+      "loss": 0.4718,
+      "step": 6348
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.560342740952097e-07,
+      "loss": 0.4569,
+      "step": 6349
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.550374523578805e-07,
+      "loss": 0.4505,
+      "step": 6350
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.540414724299331e-07,
+      "loss": 0.4821,
+      "step": 6351
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.530463344999875e-07,
+      "loss": 0.4888,
+      "step": 6352
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.520520387565131e-07,
+      "loss": 0.4633,
+      "step": 6353
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.510585853878109e-07,
+      "loss": 0.4681,
+      "step": 6354
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.500659745820269e-07,
+      "loss": 0.4821,
+      "step": 6355
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.490742065271482e-07,
+      "loss": 0.4661,
+      "step": 6356
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.480832814109982e-07,
+      "loss": 0.447,
+      "step": 6357
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.470931994212453e-07,
+      "loss": 0.4461,
+      "step": 6358
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.461039607453949e-07,
+      "loss": 0.4624,
+      "step": 6359
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.451155655707952e-07,
+      "loss": 0.4584,
+      "step": 6360
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.441280140846333e-07,
+      "loss": 0.4691,
+      "step": 6361
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.431413064739366e-07,
+      "loss": 0.4988,
+      "step": 6362
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.421554429255721e-07,
+      "loss": 0.4333,
+      "step": 6363
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.411704236262477e-07,
+      "loss": 0.4363,
+      "step": 6364
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.401862487625115e-07,
+      "loss": 0.4564,
+      "step": 6365
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.39202918520752e-07,
+      "loss": 0.4824,
+      "step": 6366
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.382204330871965e-07,
+      "loss": 0.4424,
+      "step": 6367
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.372387926479145e-07,
+      "loss": 0.4822,
+      "step": 6368
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.362579973888116e-07,
+      "loss": 0.4574,
+      "step": 6369
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.352780474956376e-07,
+      "loss": 0.4377,
+      "step": 6370
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.342989431539791e-07,
+      "loss": 0.4789,
+      "step": 6371
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.33320684549265e-07,
+      "loss": 0.4359,
+      "step": 6372
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.323432718667621e-07,
+      "loss": 0.4381,
+      "step": 6373
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.313667052915788e-07,
+      "loss": 0.4498,
+      "step": 6374
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.303909850086602e-07,
+      "loss": 0.4941,
+      "step": 6375
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.294161112027946e-07,
+      "loss": 0.4541,
+      "step": 6376
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.284420840586085e-07,
+      "loss": 0.4401,
+      "step": 6377
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.274689037605674e-07,
+      "loss": 0.4866,
+      "step": 6378
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.264965704929781e-07,
+      "loss": 0.4324,
+      "step": 6379
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.255250844399845e-07,
+      "loss": 0.4323,
+      "step": 6380
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.245544457855723e-07,
+      "loss": 0.4797,
+      "step": 6381
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.235846547135659e-07,
+      "loss": 0.4824,
+      "step": 6382
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.226157114076291e-07,
+      "loss": 0.4627,
+      "step": 6383
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.216476160512657e-07,
+      "loss": 0.4492,
+      "step": 6384
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.206803688278189e-07,
+      "loss": 0.4665,
+      "step": 6385
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.197139699204689e-07,
+      "loss": 0.4528,
+      "step": 6386
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.187484195122388e-07,
+      "loss": 0.4745,
+      "step": 6387
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.17783717785989e-07,
+      "loss": 0.4432,
+      "step": 6388
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.168198649244199e-07,
+      "loss": 0.4749,
+      "step": 6389
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1585686111007e-07,
+      "loss": 0.4413,
+      "step": 6390
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.148947065253196e-07,
+      "loss": 0.4345,
+      "step": 6391
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.139334013523839e-07,
+      "loss": 0.4696,
+      "step": 6392
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.129729457733207e-07,
+      "loss": 0.4571,
+      "step": 6393
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.120133399700261e-07,
+      "loss": 0.4356,
+      "step": 6394
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.110545841242353e-07,
+      "loss": 0.4654,
+      "step": 6395
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.10096678417522e-07,
+      "loss": 0.4701,
+      "step": 6396
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.091396230313001e-07,
+      "loss": 0.484,
+      "step": 6397
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.081834181468204e-07,
+      "loss": 0.4702,
+      "step": 6398
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.072280639451732e-07,
+      "loss": 0.4825,
+      "step": 6399
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0627356060729e-07,
+      "loss": 0.5063,
+      "step": 6400
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.053199083139382e-07,
+      "loss": 0.4749,
+      "step": 6401
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.043671072457263e-07,
+      "loss": 0.4622,
+      "step": 6402
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.034151575831003e-07,
+      "loss": 0.4504,
+      "step": 6403
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.024640595063446e-07,
+      "loss": 0.4617,
+      "step": 6404
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.015138131955838e-07,
+      "loss": 0.4716,
+      "step": 6405
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.005644188307779e-07,
+      "loss": 0.4406,
+      "step": 6406
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.996158765917308e-07,
+      "loss": 0.4191,
+      "step": 6407
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.986681866580829e-07,
+      "loss": 0.5054,
+      "step": 6408
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.977213492093091e-07,
+      "loss": 0.4614,
+      "step": 6409
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.967753644247286e-07,
+      "loss": 0.4826,
+      "step": 6410
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.958302324834968e-07,
+      "loss": 0.4589,
+      "step": 6411
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.948859535646061e-07,
+      "loss": 0.4932,
+      "step": 6412
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.939425278468885e-07,
+      "loss": 0.4875,
+      "step": 6413
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.929999555090176e-07,
+      "loss": 0.4723,
+      "step": 6414
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.920582367295002e-07,
+      "loss": 0.469,
+      "step": 6415
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.911173716866841e-07,
+      "loss": 0.4419,
+      "step": 6416
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.901773605587562e-07,
+      "loss": 0.4544,
+      "step": 6417
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.892382035237387e-07,
+      "loss": 0.4547,
+      "step": 6418
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.882999007594935e-07,
+      "loss": 0.4862,
+      "step": 6419
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.873624524437243e-07,
+      "loss": 0.4088,
+      "step": 6420
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.864258587539672e-07,
+      "loss": 0.4383,
+      "step": 6421
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.854901198675993e-07,
+      "loss": 0.4585,
+      "step": 6422
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.845552359618372e-07,
+      "loss": 0.4401,
+      "step": 6423
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.836212072137314e-07,
+      "loss": 0.4823,
+      "step": 6424
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.826880338001732e-07,
+      "loss": 0.4706,
+      "step": 6425
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.817557158978947e-07,
+      "loss": 0.4715,
+      "step": 6426
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.808242536834601e-07,
+      "loss": 0.4635,
+      "step": 6427
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.798936473332749e-07,
+      "loss": 0.4776,
+      "step": 6428
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.789638970235833e-07,
+      "loss": 0.4621,
+      "step": 6429
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.780350029304642e-07,
+      "loss": 0.4477,
+      "step": 6430
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.771069652298376e-07,
+      "loss": 0.4808,
+      "step": 6431
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.761797840974591e-07,
+      "loss": 0.441,
+      "step": 6432
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7525345970892367e-07,
+      "loss": 0.4537,
+      "step": 6433
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.743279922396632e-07,
+      "loss": 0.4547,
+      "step": 6434
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.734033818649469e-07,
+      "loss": 0.4726,
+      "step": 6435
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.724796287598821e-07,
+      "loss": 0.4391,
+      "step": 6436
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7155673309941417e-07,
+      "loss": 0.4714,
+      "step": 6437
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.706346950583257e-07,
+      "loss": 0.4812,
+      "step": 6438
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6971351481123694e-07,
+      "loss": 0.4675,
+      "step": 6439
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6879319253260615e-07,
+      "loss": 0.4437,
+      "step": 6440
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.678737283967272e-07,
+      "loss": 0.4701,
+      "step": 6441
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6695512257773357e-07,
+      "loss": 0.4829,
+      "step": 6442
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.660373752495956e-07,
+      "loss": 0.4701,
+      "step": 6443
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6512048658612085e-07,
+      "loss": 0.4897,
+      "step": 6444
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.642044567609544e-07,
+      "loss": 0.4675,
+      "step": 6445
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.632892859475796e-07,
+      "loss": 0.4851,
+      "step": 6446
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6237497431931344e-07,
+      "loss": 0.4558,
+      "step": 6447
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.614615220493146e-07,
+      "loss": 0.4793,
+      "step": 6448
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.605489293105775e-07,
+      "loss": 0.4628,
+      "step": 6449
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5963719627593273e-07,
+      "loss": 0.4036,
+      "step": 6450
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.587263231180489e-07,
+      "loss": 0.4556,
+      "step": 6451
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5781631000943347e-07,
+      "loss": 0.4227,
+      "step": 6452
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5690715712242706e-07,
+      "loss": 0.4488,
+      "step": 6453
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.559988646292102e-07,
+      "loss": 0.459,
+      "step": 6454
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5509143270180024e-07,
+      "loss": 0.4875,
+      "step": 6455
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5418486151205144e-07,
+      "loss": 0.4323,
+      "step": 6456
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5327915123165534e-07,
+      "loss": 0.462,
+      "step": 6457
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.523743020321381e-07,
+      "loss": 0.4782,
+      "step": 6458
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5147031408486585e-07,
+      "loss": 0.5007,
+      "step": 6459
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.505671875610401e-07,
+      "loss": 0.4415,
+      "step": 6460
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4966492263169905e-07,
+      "loss": 0.4653,
+      "step": 6461
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.487635194677192e-07,
+      "loss": 0.4384,
+      "step": 6462
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4786297823981307e-07,
+      "loss": 0.4771,
+      "step": 6463
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.46963299118528e-07,
+      "loss": 0.4349,
+      "step": 6464
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.460644822742505e-07,
+      "loss": 0.4673,
+      "step": 6465
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.451665278772044e-07,
+      "loss": 0.5107,
+      "step": 6466
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4426943609744586e-07,
+      "loss": 0.481,
+      "step": 6467
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4337320710487297e-07,
+      "loss": 0.4661,
+      "step": 6468
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.424778410692182e-07,
+      "loss": 0.4473,
+      "step": 6469
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4158333816004926e-07,
+      "loss": 0.4659,
+      "step": 6470
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4068969854677234e-07,
+      "loss": 0.4514,
+      "step": 6471
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.397969223986298e-07,
+      "loss": 0.4645,
+      "step": 6472
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3890500988469754e-07,
+      "loss": 0.468,
+      "step": 6473
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.380139611738937e-07,
+      "loss": 0.4854,
+      "step": 6474
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3712377643496897e-07,
+      "loss": 0.4596,
+      "step": 6475
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3623445583650946e-07,
+      "loss": 0.4673,
+      "step": 6476
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3534599954694e-07,
+      "loss": 0.4614,
+      "step": 6477
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3445840773452154e-07,
+      "loss": 0.4426,
+      "step": 6478
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3357168056734843e-07,
+      "loss": 0.4688,
+      "step": 6479
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3268581821335756e-07,
+      "loss": 0.4412,
+      "step": 6480
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3180082084031407e-07,
+      "loss": 0.4858,
+      "step": 6481
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3091668861582503e-07,
+      "loss": 0.4597,
+      "step": 6482
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.300334217073321e-07,
+      "loss": 0.4439,
+      "step": 6483
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.29151020282112e-07,
+      "loss": 0.5104,
+      "step": 6484
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2826948450727823e-07,
+      "loss": 0.4452,
+      "step": 6485
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2738881454978066e-07,
+      "loss": 0.4617,
+      "step": 6486
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.265090105764058e-07,
+      "loss": 0.412,
+      "step": 6487
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2563007275377486e-07,
+      "loss": 0.4896,
+      "step": 6488
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.247520012483458e-07,
+      "loss": 0.4538,
+      "step": 6489
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2387479622641114e-07,
+      "loss": 0.4405,
+      "step": 6490
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.229984578541013e-07,
+      "loss": 0.4762,
+      "step": 6491
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2212298629738136e-07,
+      "loss": 0.4768,
+      "step": 6492
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.212483817220525e-07,
+      "loss": 0.4621,
+      "step": 6493
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.203746442937523e-07,
+      "loss": 0.4906,
+      "step": 6494
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1950177417795336e-07,
+      "loss": 0.4566,
+      "step": 6495
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1862977153996344e-07,
+      "loss": 0.4671,
+      "step": 6496
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1775863654492767e-07,
+      "loss": 0.4849,
+      "step": 6497
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.168883693578252e-07,
+      "loss": 0.4822,
+      "step": 6498
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1601897014347193e-07,
+      "loss": 0.46,
+      "step": 6499
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.15150439066519e-07,
+      "loss": 0.4694,
+      "step": 6500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1428277629145476e-07,
+      "loss": 0.4384,
+      "step": 6501
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.134159819825989e-07,
+      "loss": 0.4866,
+      "step": 6502
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1255005630411073e-07,
+      "loss": 0.4846,
+      "step": 6503
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.116849994199834e-07,
+      "loss": 0.4792,
+      "step": 6504
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1082081149404594e-07,
+      "loss": 0.441,
+      "step": 6505
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0995749268996356e-07,
+      "loss": 0.4651,
+      "step": 6506
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0909504317123316e-07,
+      "loss": 0.4458,
+      "step": 6507
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.082334631011919e-07,
+      "loss": 0.4742,
+      "step": 6508
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0737275264300934e-07,
+      "loss": 0.441,
+      "step": 6509
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.065129119596922e-07,
+      "loss": 0.4594,
+      "step": 6510
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0565394121407973e-07,
+      "loss": 0.5014,
+      "step": 6511
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.047958405688507e-07,
+      "loss": 0.4489,
+      "step": 6512
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0393861018651394e-07,
+      "loss": 0.4592,
+      "step": 6513
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0308225022941685e-07,
+      "loss": 0.4818,
+      "step": 6514
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.022267608597413e-07,
+      "loss": 0.4634,
+      "step": 6515
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0137214223950507e-07,
+      "loss": 0.4786,
+      "step": 6516
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.005183945305585e-07,
+      "loss": 0.4468,
+      "step": 6517
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9966551789459075e-07,
+      "loss": 0.4518,
+      "step": 6518
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.988135124931214e-07,
+      "loss": 0.4647,
+      "step": 6519
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9796237848750917e-07,
+      "loss": 0.4433,
+      "step": 6520
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.971121160389452e-07,
+      "loss": 0.4616,
+      "step": 6521
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9626272530845723e-07,
+      "loss": 0.4484,
+      "step": 6522
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9541420645690665e-07,
+      "loss": 0.483,
+      "step": 6523
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.945665596449916e-07,
+      "loss": 0.4972,
+      "step": 6524
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9371978503324135e-07,
+      "loss": 0.4713,
+      "step": 6525
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9287388278202385e-07,
+      "loss": 0.4383,
+      "step": 6526
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.920288530515398e-07,
+      "loss": 0.4583,
+      "step": 6527
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9118469600182565e-07,
+      "loss": 0.4695,
+      "step": 6528
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.903414117927512e-07,
+      "loss": 0.5014,
+      "step": 6529
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.894990005840238e-07,
+      "loss": 0.4551,
+      "step": 6530
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8865746253518135e-07,
+      "loss": 0.4638,
+      "step": 6531
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8781679780560077e-07,
+      "loss": 0.4817,
+      "step": 6532
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8697700655448755e-07,
+      "loss": 0.4263,
+      "step": 6533
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8613808894088943e-07,
+      "loss": 0.4266,
+      "step": 6534
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8530004512368433e-07,
+      "loss": 0.4851,
+      "step": 6535
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.844628752615831e-07,
+      "loss": 0.4679,
+      "step": 6536
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.83626579513135e-07,
+      "loss": 0.461,
+      "step": 6537
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.827911580367216e-07,
+      "loss": 0.4397,
+      "step": 6538
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.819566109905587e-07,
+      "loss": 0.4578,
+      "step": 6539
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8112293853269646e-07,
+      "loss": 0.4887,
+      "step": 6540
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.802901408210224e-07,
+      "loss": 0.4589,
+      "step": 6541
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7945821801325334e-07,
+      "loss": 0.4504,
+      "step": 6542
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.786271702669447e-07,
+      "loss": 0.4437,
+      "step": 6543
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.777969977394841e-07,
+      "loss": 0.4751,
+      "step": 6544
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7696770058809285e-07,
+      "loss": 0.4908,
+      "step": 6545
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7613927896982706e-07,
+      "loss": 0.4642,
+      "step": 6546
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7531173304158067e-07,
+      "loss": 0.4269,
+      "step": 6547
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7448506296007494e-07,
+      "loss": 0.4664,
+      "step": 6548
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.736592688818702e-07,
+      "loss": 0.4943,
+      "step": 6549
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.728343509633603e-07,
+      "loss": 0.494,
+      "step": 6550
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.720103093607702e-07,
+      "loss": 0.4745,
+      "step": 6551
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7118714423016235e-07,
+      "loss": 0.4763,
+      "step": 6552
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.703648557274325e-07,
+      "loss": 0.4551,
+      "step": 6553
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.69543444008309e-07,
+      "loss": 0.446,
+      "step": 6554
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6872290922835505e-07,
+      "loss": 0.5008,
+      "step": 6555
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.679032515429681e-07,
+      "loss": 0.4522,
+      "step": 6556
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.670844711073784e-07,
+      "loss": 0.4809,
+      "step": 6557
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.662665680766508e-07,
+      "loss": 0.4555,
+      "step": 6558
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.654495426056837e-07,
+      "loss": 0.4685,
+      "step": 6559
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.646333948492098e-07,
+      "loss": 0.4688,
+      "step": 6560
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.638181249617967e-07,
+      "loss": 0.4254,
+      "step": 6561
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.630037330978414e-07,
+      "loss": 0.4794,
+      "step": 6562
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.621902194115795e-07,
+      "loss": 0.4576,
+      "step": 6563
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.613775840570777e-07,
+      "loss": 0.4688,
+      "step": 6564
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.605658271882373e-07,
+      "loss": 0.4394,
+      "step": 6565
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5975494895879257e-07,
+      "loss": 0.4575,
+      "step": 6566
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.589449495223124e-07,
+      "loss": 0.4521,
+      "step": 6567
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.581358290321979e-07,
+      "loss": 0.4364,
+      "step": 6568
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.573275876416843e-07,
+      "loss": 0.4534,
+      "step": 6569
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.565202255038402e-07,
+      "loss": 0.4477,
+      "step": 6570
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.55713742771569e-07,
+      "loss": 0.4799,
+      "step": 6571
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5490813959760506e-07,
+      "loss": 0.4666,
+      "step": 6572
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.541034161345197e-07,
+      "loss": 0.454,
+      "step": 6573
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5329957253471327e-07,
+      "loss": 0.4719,
+      "step": 6574
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5249660895042226e-07,
+      "loss": 0.4771,
+      "step": 6575
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5169452553371673e-07,
+      "loss": 0.4491,
+      "step": 6576
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5089332243649854e-07,
+      "loss": 0.4663,
+      "step": 6577
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.500929998105035e-07,
+      "loss": 0.4621,
+      "step": 6578
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4929355780730256e-07,
+      "loss": 0.4707,
+      "step": 6579
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.484949965782952e-07,
+      "loss": 0.4817,
+      "step": 6580
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.476973162747183e-07,
+      "loss": 0.4662,
+      "step": 6581
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.469005170476408e-07,
+      "loss": 0.4499,
+      "step": 6582
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.461045990479645e-07,
+      "loss": 0.4589,
+      "step": 6583
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.453095624264252e-07,
+      "loss": 0.4707,
+      "step": 6584
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.445154073335888e-07,
+      "loss": 0.4589,
+      "step": 6585
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4372213391985763e-07,
+      "loss": 0.4499,
+      "step": 6586
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.429297423354655e-07,
+      "loss": 0.446,
+      "step": 6587
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4213823273048006e-07,
+      "loss": 0.4608,
+      "step": 6588
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.413476052548009e-07,
+      "loss": 0.4782,
+      "step": 6589
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4055786005816215e-07,
+      "loss": 0.4219,
+      "step": 6590
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3976899729012746e-07,
+      "loss": 0.4771,
+      "step": 6591
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3898101710009734e-07,
+      "loss": 0.4935,
+      "step": 6592
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3819391963730406e-07,
+      "loss": 0.477,
+      "step": 6593
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.37407705050809e-07,
+      "loss": 0.4407,
+      "step": 6594
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3662237348951245e-07,
+      "loss": 0.4698,
+      "step": 6595
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.358379251021443e-07,
+      "loss": 0.4587,
+      "step": 6596
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3505436003726576e-07,
+      "loss": 0.4868,
+      "step": 6597
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3427167844327313e-07,
+      "loss": 0.485,
+      "step": 6598
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.334898804683956e-07,
+      "loss": 0.486,
+      "step": 6599
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3270896626069136e-07,
+      "loss": 0.4666,
+      "step": 6600
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.319289359680561e-07,
+      "loss": 0.5037,
+      "step": 6601
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.31149789738216e-07,
+      "loss": 0.4333,
+      "step": 6602
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.303715277187286e-07,
+      "loss": 0.4482,
+      "step": 6603
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2959415005698493e-07,
+      "loss": 0.4814,
+      "step": 6604
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.288176569002105e-07,
+      "loss": 0.4361,
+      "step": 6605
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.280420483954583e-07,
+      "loss": 0.4395,
+      "step": 6606
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2726732468961963e-07,
+      "loss": 0.4469,
+      "step": 6607
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2649348592941553e-07,
+      "loss": 0.4686,
+      "step": 6608
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2572053226139824e-07,
+      "loss": 0.4708,
+      "step": 6609
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.24948463831955e-07,
+      "loss": 0.4389,
+      "step": 6610
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2417728078730226e-07,
+      "loss": 0.4699,
+      "step": 6611
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.234069832734904e-07,
+      "loss": 0.4837,
+      "step": 6612
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2263757143640494e-07,
+      "loss": 0.4592,
+      "step": 6613
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2186904542175876e-07,
+      "loss": 0.4993,
+      "step": 6614
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.211014053750994e-07,
+      "loss": 0.4658,
+      "step": 6615
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2033465144180767e-07,
+      "loss": 0.4493,
+      "step": 6616
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1956878376709355e-07,
+      "loss": 0.4707,
+      "step": 6617
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1880380249600216e-07,
+      "loss": 0.4574,
+      "step": 6618
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.180397077734082e-07,
+      "loss": 0.4685,
+      "step": 6619
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.172764997440214e-07,
+      "loss": 0.4394,
+      "step": 6620
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1651417855238066e-07,
+      "loss": 0.4969,
+      "step": 6621
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1575274434285983e-07,
+      "loss": 0.4799,
+      "step": 6622
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.149921972596609e-07,
+      "loss": 0.423,
+      "step": 6623
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.142325374468214e-07,
+      "loss": 0.4814,
+      "step": 6624
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.134737650482095e-07,
+      "loss": 0.4511,
+      "step": 6625
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1271588020752476e-07,
+      "loss": 0.4458,
+      "step": 6626
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1195888306830015e-07,
+      "loss": 0.4191,
+      "step": 6627
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1120277377389984e-07,
+      "loss": 0.4432,
+      "step": 6628
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1044755246751766e-07,
+      "loss": 0.4524,
+      "step": 6629
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.096932192921831e-07,
+      "loss": 0.4791,
+      "step": 6630
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.089397743907546e-07,
+      "loss": 0.4592,
+      "step": 6631
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.081872179059236e-07,
+      "loss": 0.4965,
+      "step": 6632
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.074355499802134e-07,
+      "loss": 0.4585,
+      "step": 6633
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.066847707559789e-07,
+      "loss": 0.4742,
+      "step": 6634
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0593488037540533e-07,
+      "loss": 0.467,
+      "step": 6635
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0518587898051065e-07,
+      "loss": 0.4696,
+      "step": 6636
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0443776671314593e-07,
+      "loss": 0.4703,
+      "step": 6637
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.036905437149912e-07,
+      "loss": 0.4669,
+      "step": 6638
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0294421012756095e-07,
+      "loss": 0.4403,
+      "step": 6639
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.021987660921971e-07,
+      "loss": 0.5021,
+      "step": 6640
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.014542117500774e-07,
+      "loss": 0.458,
+      "step": 6641
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.007105472422084e-07,
+      "loss": 0.4293,
+      "step": 6642
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9996777270942966e-07,
+      "loss": 0.4838,
+      "step": 6643
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.992258882924115e-07,
+      "loss": 0.4899,
+      "step": 6644
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.984848941316565e-07,
+      "loss": 0.4728,
+      "step": 6645
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.977447903674957e-07,
+      "loss": 0.4667,
+      "step": 6646
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9700557714009484e-07,
+      "loss": 0.4886,
+      "step": 6647
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9626725458945016e-07,
+      "loss": 0.4534,
+      "step": 6648
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9552982285538823e-07,
+      "loss": 0.4612,
+      "step": 6649
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.947932820775684e-07,
+      "loss": 0.4713,
+      "step": 6650
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.940576323954808e-07,
+      "loss": 0.4375,
+      "step": 6651
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9332287394844505e-07,
+      "loss": 0.4511,
+      "step": 6652
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9258900687561374e-07,
+      "loss": 0.4182,
+      "step": 6653
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9185603131597073e-07,
+      "loss": 0.4576,
+      "step": 6654
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.911239474083311e-07,
+      "loss": 0.468,
+      "step": 6655
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9039275529133935e-07,
+      "loss": 0.4447,
+      "step": 6656
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8966245510347436e-07,
+      "loss": 0.4476,
+      "step": 6657
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8893304698304147e-07,
+      "loss": 0.4798,
+      "step": 6658
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8820453106818083e-07,
+      "loss": 0.4663,
+      "step": 6659
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.874769074968625e-07,
+      "loss": 0.4844,
+      "step": 6660
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8675017640688805e-07,
+      "loss": 0.4347,
+      "step": 6661
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.860243379358896e-07,
+      "loss": 0.4709,
+      "step": 6662
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.852993922213282e-07,
+      "loss": 0.4428,
+      "step": 6663
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.845753394004991e-07,
+      "loss": 0.48,
+      "step": 6664
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.838521796105276e-07,
+      "loss": 0.4631,
+      "step": 6665
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8312991298836746e-07,
+      "loss": 0.4638,
+      "step": 6666
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8240853967080705e-07,
+      "loss": 0.4896,
+      "step": 6667
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8168805979446325e-07,
+      "loss": 0.4481,
+      "step": 6668
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.809684734957835e-07,
+      "loss": 0.4571,
+      "step": 6669
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.802497809110472e-07,
+      "loss": 0.4393,
+      "step": 6670
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.795319821763648e-07,
+      "loss": 0.4315,
+      "step": 6671
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7881507742767433e-07,
+      "loss": 0.5021,
+      "step": 6672
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.780990668007477e-07,
+      "loss": 0.4359,
+      "step": 6673
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7738395043118873e-07,
+      "loss": 0.4615,
+      "step": 6674
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.766697284544273e-07,
+      "loss": 0.4663,
+      "step": 6675
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7595640100572696e-07,
+      "loss": 0.4684,
+      "step": 6676
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.752439682201824e-07,
+      "loss": 0.4574,
+      "step": 6677
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7453243023271625e-07,
+      "loss": 0.5106,
+      "step": 6678
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.738217871780824e-07,
+      "loss": 0.4817,
+      "step": 6679
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7311203919086927e-07,
+      "loss": 0.4712,
+      "step": 6680
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.724031864054899e-07,
+      "loss": 0.4709,
+      "step": 6681
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.716952289561908e-07,
+      "loss": 0.4717,
+      "step": 6682
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.709881669770492e-07,
+      "loss": 0.4349,
+      "step": 6683
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.702820006019713e-07,
+      "loss": 0.4732,
+      "step": 6684
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.69576729964694e-07,
+      "loss": 0.4735,
+      "step": 6685
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6887235519878706e-07,
+      "loss": 0.4708,
+      "step": 6686
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.681688764376461e-07,
+      "loss": 0.4767,
+      "step": 6687
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6746629381450175e-07,
+      "loss": 0.4662,
+      "step": 6688
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.667646074624108e-07,
+      "loss": 0.4652,
+      "step": 6689
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.660638175142627e-07,
+      "loss": 0.4915,
+      "step": 6690
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.653639241027772e-07,
+      "loss": 0.4656,
+      "step": 6691
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.646649273605023e-07,
+      "loss": 0.4842,
+      "step": 6692
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6396682741981873e-07,
+      "loss": 0.4622,
+      "step": 6693
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.632696244129368e-07,
+      "loss": 0.4872,
+      "step": 6694
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6257331847189425e-07,
+      "loss": 0.4699,
+      "step": 6695
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.618779097285623e-07,
+      "loss": 0.4795,
+      "step": 6696
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6118339831464056e-07,
+      "loss": 0.4442,
+      "step": 6697
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.604897843616594e-07,
+      "loss": 0.4888,
+      "step": 6698
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.597970680009787e-07,
+      "loss": 0.4544,
+      "step": 6699
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5910524936378913e-07,
+      "loss": 0.4847,
+      "step": 6700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5841432858111026e-07,
+      "loss": 0.4617,
+      "step": 6701
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5772430578379194e-07,
+      "loss": 0.4552,
+      "step": 6702
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5703518110251404e-07,
+      "loss": 0.4882,
+      "step": 6703
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.563469546677871e-07,
+      "loss": 0.4547,
+      "step": 6704
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5565962660995093e-07,
+      "loss": 0.4735,
+      "step": 6705
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.549731970591751e-07,
+      "loss": 0.4791,
+      "step": 6706
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.542876661454585e-07,
+      "loss": 0.4725,
+      "step": 6707
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.536030339986306e-07,
+      "loss": 0.4451,
+      "step": 6708
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5291930074835103e-07,
+      "loss": 0.4503,
+      "step": 6709
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.522364665241084e-07,
+      "loss": 0.464,
+      "step": 6710
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5155453145522147e-07,
+      "loss": 0.4706,
+      "step": 6711
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5087349567083754e-07,
+      "loss": 0.4495,
+      "step": 6712
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.501933592999356e-07,
+      "loss": 0.4696,
+      "step": 6713
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.495141224713232e-07,
+      "loss": 0.4656,
+      "step": 6714
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4883578531363693e-07,
+      "loss": 0.4784,
+      "step": 6715
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4815834795534454e-07,
+      "loss": 0.4404,
+      "step": 6716
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4748181052474283e-07,
+      "loss": 0.449,
+      "step": 6717
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.468061731499566e-07,
+      "loss": 0.4746,
+      "step": 6718
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4613143595894226e-07,
+      "loss": 0.4758,
+      "step": 6719
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.45457599079485e-07,
+      "loss": 0.4777,
+      "step": 6720
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.447846626391981e-07,
+      "loss": 0.4558,
+      "step": 6721
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.441126267655275e-07,
+      "loss": 0.4376,
+      "step": 6722
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.434414915857464e-07,
+      "loss": 0.4924,
+      "step": 6723
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.427712572269569e-07,
+      "loss": 0.4682,
+      "step": 6724
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.421019238160921e-07,
+      "loss": 0.4599,
+      "step": 6725
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.414334914799138e-07,
+      "loss": 0.4619,
+      "step": 6726
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.407659603450113e-07,
+      "loss": 0.456,
+      "step": 6727
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.400993305378069e-07,
+      "loss": 0.4831,
+      "step": 6728
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.394336021845511e-07,
+      "loss": 0.4622,
+      "step": 6729
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.387687754113205e-07,
+      "loss": 0.4729,
+      "step": 6730
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.381048503440242e-07,
+      "loss": 0.4849,
+      "step": 6731
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3744182710840113e-07,
+      "loss": 0.4311,
+      "step": 6732
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.367797058300153e-07,
+      "loss": 0.4752,
+      "step": 6733
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3611848663426474e-07,
+      "loss": 0.4456,
+      "step": 6734
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3545816964637425e-07,
+      "loss": 0.445,
+      "step": 6735
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3479875499139716e-07,
+      "loss": 0.4494,
+      "step": 6736
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3414024279421743e-07,
+      "loss": 0.4688,
+      "step": 6737
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3348263317954644e-07,
+      "loss": 0.4493,
+      "step": 6738
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3282592627192568e-07,
+      "loss": 0.4573,
+      "step": 6739
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3217012219572789e-07,
+      "loss": 0.4583,
+      "step": 6740
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3151522107514924e-07,
+      "loss": 0.4556,
+      "step": 6741
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3086122303422053e-07,
+      "loss": 0.4703,
+      "step": 6742
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3020812819679882e-07,
+      "loss": 0.5029,
+      "step": 6743
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2955593668656962e-07,
+      "loss": 0.4359,
+      "step": 6744
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2890464862704853e-07,
+      "loss": 0.4529,
+      "step": 6745
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2825426414158024e-07,
+      "loss": 0.4522,
+      "step": 6746
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2760478335333734e-07,
+      "loss": 0.4369,
+      "step": 6747
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.26956206385322e-07,
+      "loss": 0.4681,
+      "step": 6748
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2630853336036595e-07,
+      "loss": 0.4586,
+      "step": 6749
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2566176440112663e-07,
+      "loss": 0.4707,
+      "step": 6750
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.250158996300944e-07,
+      "loss": 0.442,
+      "step": 6751
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2437093916958474e-07,
+      "loss": 0.4311,
+      "step": 6752
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2372688314174496e-07,
+      "loss": 0.4852,
+      "step": 6753
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2308373166854856e-07,
+      "loss": 0.4309,
+      "step": 6754
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2244148487180038e-07,
+      "loss": 0.4547,
+      "step": 6755
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.218001428731298e-07,
+      "loss": 0.4944,
+      "step": 6756
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.211597057939996e-07,
+      "loss": 0.4868,
+      "step": 6757
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.205201737556978e-07,
+      "loss": 0.5029,
+      "step": 6758
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.198815468793425e-07,
+      "loss": 0.4589,
+      "step": 6759
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1924382528588085e-07,
+      "loss": 0.4795,
+      "step": 6760
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1860700909608678e-07,
+      "loss": 0.4652,
+      "step": 6761
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1797109843056442e-07,
+      "loss": 0.451,
+      "step": 6762
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.173360934097446e-07,
+      "loss": 0.4823,
+      "step": 6763
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1670199415388893e-07,
+      "loss": 0.4641,
+      "step": 6764
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.160688007830858e-07,
+      "loss": 0.46,
+      "step": 6765
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1543651341725315e-07,
+      "loss": 0.4138,
+      "step": 6766
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1480513217613574e-07,
+      "loss": 0.426,
+      "step": 6767
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.141746571793085e-07,
+      "loss": 0.4981,
+      "step": 6768
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1354508854617317e-07,
+      "loss": 0.4385,
+      "step": 6769
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1291642639596155e-07,
+      "loss": 0.4681,
+      "step": 6770
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1228867084773286e-07,
+      "loss": 0.4695,
+      "step": 6771
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.116618220203742e-07,
+      "loss": 0.4617,
+      "step": 6772
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1103588003260124e-07,
+      "loss": 0.4866,
+      "step": 6773
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1041084500295793e-07,
+      "loss": 0.4521,
+      "step": 6774
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.097867170498169e-07,
+      "loss": 0.5021,
+      "step": 6775
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.09163496291378e-07,
+      "loss": 0.4546,
+      "step": 6776
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0854118284567128e-07,
+      "loss": 0.5051,
+      "step": 6777
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0791977683055242e-07,
+      "loss": 0.4529,
+      "step": 6778
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0729927836370677e-07,
+      "loss": 0.4688,
+      "step": 6779
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0667968756264756e-07,
+      "loss": 0.4657,
+      "step": 6780
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0606100454471534e-07,
+      "loss": 0.4933,
+      "step": 6781
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.054432294270803e-07,
+      "loss": 0.4884,
+      "step": 6782
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0482636232673936e-07,
+      "loss": 0.4775,
+      "step": 6783
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0421040336051856e-07,
+      "loss": 0.4363,
+      "step": 6784
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.035953526450707e-07,
+      "loss": 0.4648,
+      "step": 6785
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0298121029687645e-07,
+      "loss": 0.4541,
+      "step": 6786
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.023679764322467e-07,
+      "loss": 0.4664,
+      "step": 6787
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0175565116731744e-07,
+      "loss": 0.4597,
+      "step": 6788
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0114423461805532e-07,
+      "loss": 0.4636,
+      "step": 6789
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0053372690025219e-07,
+      "loss": 0.443,
+      "step": 6790
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9992412812952944e-07,
+      "loss": 0.4872,
+      "step": 6791
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9931543842133693e-07,
+      "loss": 0.4832,
+      "step": 6792
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9870765789094915e-07,
+      "loss": 0.4189,
+      "step": 6793
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.981007866534723e-07,
+      "loss": 0.4574,
+      "step": 6794
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9749482482383952e-07,
+      "loss": 0.451,
+      "step": 6795
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9688977251680895e-07,
+      "loss": 0.4492,
+      "step": 6796
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9628562984696952e-07,
+      "loss": 0.4429,
+      "step": 6797
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9568239692873692e-07,
+      "loss": 0.4741,
+      "step": 6798
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9508007387635363e-07,
+      "loss": 0.4527,
+      "step": 6799
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.944786608038901e-07,
+      "loss": 0.4715,
+      "step": 6800
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.93878157825248e-07,
+      "loss": 0.4485,
+      "step": 6801
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.932785650541502e-07,
+      "loss": 0.4471,
+      "step": 6802
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.926798826041526e-07,
+      "loss": 0.495,
+      "step": 6803
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9208211058863614e-07,
+      "loss": 0.4327,
+      "step": 6804
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.914852491208097e-07,
+      "loss": 0.4584,
+      "step": 6805
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9088929831370895e-07,
+      "loss": 0.4768,
+      "step": 6806
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.902942582802003e-07,
+      "loss": 0.4377,
+      "step": 6807
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.897001291329742e-07,
+      "loss": 0.4731,
+      "step": 6808
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8910691098454948e-07,
+      "loss": 0.4426,
+      "step": 6809
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8851460394727406e-07,
+      "loss": 0.4412,
+      "step": 6810
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8792320813332044e-07,
+      "loss": 0.488,
+      "step": 6811
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8733272365469014e-07,
+      "loss": 0.4575,
+      "step": 6812
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8674315062321368e-07,
+      "loss": 0.4679,
+      "step": 6813
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8615448915054622e-07,
+      "loss": 0.4556,
+      "step": 6814
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8556673934817249e-07,
+      "loss": 0.4587,
+      "step": 6815
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8497990132740173e-07,
+      "loss": 0.4332,
+      "step": 6816
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8439397519937342e-07,
+      "loss": 0.4256,
+      "step": 6817
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8380896107505264e-07,
+      "loss": 0.4617,
+      "step": 6818
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8322485906523302e-07,
+      "loss": 0.4497,
+      "step": 6819
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.826416692805344e-07,
+      "loss": 0.4733,
+      "step": 6820
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8205939183140508e-07,
+      "loss": 0.4399,
+      "step": 6821
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8147802682811798e-07,
+      "loss": 0.4563,
+      "step": 6822
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.808975743807756e-07,
+      "loss": 0.4713,
+      "step": 6823
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8031803459930718e-07,
+      "loss": 0.4832,
+      "step": 6824
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.797394075934694e-07,
+      "loss": 0.481,
+      "step": 6825
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7916169347284407e-07,
+      "loss": 0.4094,
+      "step": 6826
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7858489234684362e-07,
+      "loss": 0.4561,
+      "step": 6827
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7800900432470402e-07,
+      "loss": 0.467,
+      "step": 6828
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7743402951548972e-07,
+      "loss": 0.4675,
+      "step": 6829
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7685996802809357e-07,
+      "loss": 0.4633,
+      "step": 6830
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.762868199712331e-07,
+      "loss": 0.4463,
+      "step": 6831
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.757145854534542e-07,
+      "loss": 0.4792,
+      "step": 6832
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.751432645831308e-07,
+      "loss": 0.4665,
+      "step": 6833
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7457285746846075e-07,
+      "loss": 0.4762,
+      "step": 6834
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.74003364217471e-07,
+      "loss": 0.4714,
+      "step": 6835
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.734347849380158e-07,
+      "loss": 0.4433,
+      "step": 6836
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7286711973777458e-07,
+      "loss": 0.4844,
+      "step": 6837
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7230036872425583e-07,
+      "loss": 0.4754,
+      "step": 6838
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7173453200479306e-07,
+      "loss": 0.4506,
+      "step": 6839
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7116960968654728e-07,
+      "loss": 0.4986,
+      "step": 6840
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7060560187650566e-07,
+      "loss": 0.4574,
+      "step": 6841
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7004250868148388e-07,
+      "loss": 0.4395,
+      "step": 6842
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6948033020812326e-07,
+      "loss": 0.4893,
+      "step": 6843
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6891906656289202e-07,
+      "loss": 0.4826,
+      "step": 6844
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6835871785208403e-07,
+      "loss": 0.4647,
+      "step": 6845
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6779928418182156e-07,
+      "loss": 0.4558,
+      "step": 6846
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.672407656580538e-07,
+      "loss": 0.4734,
+      "step": 6847
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6668316238655446e-07,
+      "loss": 0.4595,
+      "step": 6848
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.661264744729263e-07,
+      "loss": 0.4529,
+      "step": 6849
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6557070202259773e-07,
+      "loss": 0.4258,
+      "step": 6850
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6501584514082293e-07,
+      "loss": 0.499,
+      "step": 6851
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6446190393268336e-07,
+      "loss": 0.4608,
+      "step": 6852
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6390887850308847e-07,
+      "loss": 0.4671,
+      "step": 6853
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.633567689567711e-07,
+      "loss": 0.4597,
+      "step": 6854
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6280557539829432e-07,
+      "loss": 0.4658,
+      "step": 6855
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6225529793204576e-07,
+      "loss": 0.4803,
+      "step": 6856
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6170593666223867e-07,
+      "loss": 0.4874,
+      "step": 6857
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6115749169291483e-07,
+      "loss": 0.4603,
+      "step": 6858
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6060996312794118e-07,
+      "loss": 0.4647,
+      "step": 6859
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6006335107101034e-07,
+      "loss": 0.4485,
+      "step": 6860
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5951765562564448e-07,
+      "loss": 0.4435,
+      "step": 6861
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5897287689518926e-07,
+      "loss": 0.4882,
+      "step": 6862
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.584290149828166e-07,
+      "loss": 0.465,
+      "step": 6863
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5788606999152745e-07,
+      "loss": 0.4747,
+      "step": 6864
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5734404202414677e-07,
+      "loss": 0.479,
+      "step": 6865
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.568029311833258e-07,
+      "loss": 0.4674,
+      "step": 6866
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.562627375715442e-07,
+      "loss": 0.5077,
+      "step": 6867
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5572346129110572e-07,
+      "loss": 0.441,
+      "step": 6868
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5518510244414143e-07,
+      "loss": 0.4544,
+      "step": 6869
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5464766113260866e-07,
+      "loss": 0.4295,
+      "step": 6870
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5411113745828987e-07,
+      "loss": 0.4633,
+      "step": 6871
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5357553152279492e-07,
+      "loss": 0.4629,
+      "step": 6872
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.53040843427561e-07,
+      "loss": 0.4515,
+      "step": 6873
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5250707327384816e-07,
+      "loss": 0.4817,
+      "step": 6874
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5197422116274562e-07,
+      "loss": 0.474,
+      "step": 6875
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5144228719516708e-07,
+      "loss": 0.4721,
+      "step": 6876
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5091127147185302e-07,
+      "loss": 0.4559,
+      "step": 6877
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.503811740933697e-07,
+      "loss": 0.4639,
+      "step": 6878
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4985199516010963e-07,
+      "loss": 0.4712,
+      "step": 6879
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4932373477229146e-07,
+      "loss": 0.457,
+      "step": 6880
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.487963930299602e-07,
+      "loss": 0.4788,
+      "step": 6881
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.482699700329865e-07,
+      "loss": 0.4557,
+      "step": 6882
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4774446588106617e-07,
+      "loss": 0.4634,
+      "step": 6883
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.472198806737224e-07,
+      "loss": 0.49,
+      "step": 6884
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4669621451030346e-07,
+      "loss": 0.443,
+      "step": 6885
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4617346748998396e-07,
+      "loss": 0.4133,
+      "step": 6886
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4565163971176467e-07,
+      "loss": 0.4493,
+      "step": 6887
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4513073127447264e-07,
+      "loss": 0.4518,
+      "step": 6888
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4461074227675842e-07,
+      "loss": 0.4695,
+      "step": 6889
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4409167281710102e-07,
+      "loss": 0.4705,
+      "step": 6890
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.43573522993804e-07,
+      "loss": 0.4642,
+      "step": 6891
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4305629290499833e-07,
+      "loss": 0.4428,
+      "step": 6892
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.42539982648639e-07,
+      "loss": 0.4674,
+      "step": 6893
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4202459232250722e-07,
+      "loss": 0.4704,
+      "step": 6894
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.415101220242099e-07,
+      "loss": 0.486,
+      "step": 6895
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4099657185118077e-07,
+      "loss": 0.4743,
+      "step": 6896
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4048394190067816e-07,
+      "loss": 0.4707,
+      "step": 6897
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.399722322697866e-07,
+      "loss": 0.4892,
+      "step": 6898
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3946144305541697e-07,
+      "loss": 0.4481,
+      "step": 6899
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.389515743543035e-07,
+      "loss": 0.4303,
+      "step": 6900
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3844262626300898e-07,
+      "loss": 0.4624,
+      "step": 6901
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.379345988779196e-07,
+      "loss": 0.4669,
+      "step": 6902
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.37427492295249e-07,
+      "loss": 0.4759,
+      "step": 6903
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.369213066110353e-07,
+      "loss": 0.4493,
+      "step": 6904
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3641604192114244e-07,
+      "loss": 0.4359,
+      "step": 6905
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3591169832125995e-07,
+      "loss": 0.5021,
+      "step": 6906
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.354082759069031e-07,
+      "loss": 0.4909,
+      "step": 6907
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3490577477341226e-07,
+      "loss": 0.4682,
+      "step": 6908
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3440419501595358e-07,
+      "loss": 0.4672,
+      "step": 6909
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.339035367295194e-07,
+      "loss": 0.4427,
+      "step": 6910
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.334038000089266e-07,
+      "loss": 0.4588,
+      "step": 6911
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3290498494881789e-07,
+      "loss": 0.4361,
+      "step": 6912
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3240709164366038e-07,
+      "loss": 0.4591,
+      "step": 6913
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3191012018774928e-07,
+      "loss": 0.486,
+      "step": 6914
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3141407067520207e-07,
+      "loss": 0.4525,
+      "step": 6915
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.309189431999641e-07,
+      "loss": 0.4788,
+      "step": 6916
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3042473785580546e-07,
+      "loss": 0.4489,
+      "step": 6917
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2993145473632008e-07,
+      "loss": 0.4489,
+      "step": 6918
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.294390939349288e-07,
+      "loss": 0.4497,
+      "step": 6919
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.289476555448771e-07,
+      "loss": 0.4439,
+      "step": 6920
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2845713965923656e-07,
+      "loss": 0.4534,
+      "step": 6921
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2796754637090402e-07,
+      "loss": 0.4379,
+      "step": 6922
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2747887577259976e-07,
+      "loss": 0.4697,
+      "step": 6923
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.269911279568714e-07,
+      "loss": 0.4612,
+      "step": 6924
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.265043030160923e-07,
+      "loss": 0.5029,
+      "step": 6925
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2601840104245756e-07,
+      "loss": 0.4884,
+      "step": 6926
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2553342212799024e-07,
+      "loss": 0.4761,
+      "step": 6927
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.250493663645397e-07,
+      "loss": 0.4646,
+      "step": 6928
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2456623384377754e-07,
+      "loss": 0.4865,
+      "step": 6929
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2408402465720172e-07,
+      "loss": 0.4446,
+      "step": 6930
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2360273889613637e-07,
+      "loss": 0.461,
+      "step": 6931
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2312237665172922e-07,
+      "loss": 0.4852,
+      "step": 6932
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2264293801495297e-07,
+      "loss": 0.4775,
+      "step": 6933
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.221644230766078e-07,
+      "loss": 0.4753,
+      "step": 6934
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2168683192731568e-07,
+      "loss": 0.5141,
+      "step": 6935
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.212101646575259e-07,
+      "loss": 0.4425,
+      "step": 6936
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.207344213575129e-07,
+      "loss": 0.4394,
+      "step": 6937
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2025960211737408e-07,
+      "loss": 0.4822,
+      "step": 6938
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1978570702703306e-07,
+      "loss": 0.4819,
+      "step": 6939
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1931273617623974e-07,
+      "loss": 0.4694,
+      "step": 6940
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.188406896545663e-07,
+      "loss": 0.4821,
+      "step": 6941
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1836956755141239e-07,
+      "loss": 0.4356,
+      "step": 6942
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.178993699560016e-07,
+      "loss": 0.4585,
+      "step": 6943
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1743009695738106e-07,
+      "loss": 0.4605,
+      "step": 6944
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1696174864442466e-07,
+      "loss": 0.4715,
+      "step": 6945
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1649432510583092e-07,
+      "loss": 0.494,
+      "step": 6946
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1602782643012234e-07,
+      "loss": 0.4753,
+      "step": 6947
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1556225270564769e-07,
+      "loss": 0.4769,
+      "step": 6948
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1509760402057812e-07,
+      "loss": 0.4334,
+      "step": 6949
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1463388046291268e-07,
+      "loss": 0.4791,
+      "step": 6950
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1417108212047223e-07,
+      "loss": 0.4326,
+      "step": 6951
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1370920908090555e-07,
+      "loss": 0.4763,
+      "step": 6952
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1324826143168321e-07,
+      "loss": 0.4645,
+      "step": 6953
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1278823926010263e-07,
+      "loss": 0.4552,
+      "step": 6954
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.123291426532841e-07,
+      "loss": 0.4932,
+      "step": 6955
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1187097169817418e-07,
+      "loss": 0.4834,
+      "step": 6956
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1141372648154348e-07,
+      "loss": 0.4759,
+      "step": 6957
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.109574070899877e-07,
+      "loss": 0.4394,
+      "step": 6958
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1050201360992718e-07,
+      "loss": 0.4662,
+      "step": 6959
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1004754612760627e-07,
+      "loss": 0.5053,
+      "step": 6960
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0959400472909332e-07,
+      "loss": 0.4782,
+      "step": 6961
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0914138950028408e-07,
+      "loss": 0.4939,
+      "step": 6962
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0868970052689553e-07,
+      "loss": 0.4232,
+      "step": 6963
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0823893789447148e-07,
+      "loss": 0.4524,
+      "step": 6964
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0778910168838031e-07,
+      "loss": 0.4753,
+      "step": 6965
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0734019199381329e-07,
+      "loss": 0.4292,
+      "step": 6966
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0689220889578744e-07,
+      "loss": 0.4651,
+      "step": 6967
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0644515247914378e-07,
+      "loss": 0.4636,
+      "step": 6968
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0599902282854902e-07,
+      "loss": 0.4521,
+      "step": 6969
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0555382002849223e-07,
+      "loss": 0.4501,
+      "step": 6970
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0510954416328989e-07,
+      "loss": 0.4439,
+      "step": 6971
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.046661953170791e-07,
+      "loss": 0.4689,
+      "step": 6972
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0422377357382441e-07,
+      "loss": 0.4297,
+      "step": 6973
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0378227901731431e-07,
+      "loss": 0.4717,
+      "step": 6974
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0334171173116081e-07,
+      "loss": 0.4616,
+      "step": 6975
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0290207179880107e-07,
+      "loss": 0.438,
+      "step": 6976
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0246335930349627e-07,
+      "loss": 0.4326,
+      "step": 6977
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0202557432833105e-07,
+      "loss": 0.4692,
+      "step": 6978
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.015887169562163e-07,
+      "loss": 0.4859,
+      "step": 6979
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0115278726988697e-07,
+      "loss": 0.455,
+      "step": 6980
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0071778535189925e-07,
+      "loss": 0.4806,
+      "step": 6981
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0028371128463833e-07,
+      "loss": 0.4695,
+      "step": 6982
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.985056515031066e-08,
+      "loss": 0.4778,
+      "step": 6983
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.941834703094677e-08,
+      "loss": 0.4608,
+      "step": 6984
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.898705700840338e-08,
+      "loss": 0.4436,
+      "step": 6985
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.855669516436017e-08,
+      "loss": 0.4757,
+      "step": 6986
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.812726158032027e-08,
+      "loss": 0.4588,
+      "step": 6987
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.76987563376136e-08,
+      "loss": 0.4779,
+      "step": 6988
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.72711795173914e-08,
+      "loss": 0.4698,
+      "step": 6989
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.684453120063054e-08,
+      "loss": 0.4557,
+      "step": 6990
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.641881146813248e-08,
+      "loss": 0.4699,
+      "step": 6991
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.599402040052163e-08,
+      "loss": 0.4843,
+      "step": 6992
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.557015807824588e-08,
+      "loss": 0.4613,
+      "step": 6993
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.514722458158099e-08,
+      "loss": 0.5001,
+      "step": 6994
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.472521999062234e-08,
+      "loss": 0.473,
+      "step": 6995
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.430414438529156e-08,
+      "loss": 0.446,
+      "step": 6996
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.388399784533541e-08,
+      "loss": 0.4602,
+      "step": 6997
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.346478045032137e-08,
+      "loss": 0.4752,
+      "step": 6998
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.304649227964313e-08,
+      "loss": 0.4828,
+      "step": 6999
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.262913341252011e-08,
+      "loss": 0.4775,
+      "step": 7000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.221270392799187e-08,
+      "loss": 0.4487,
+      "step": 7001
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.17972039049253e-08,
+      "loss": 0.4301,
+      "step": 7002
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.138263342200915e-08,
+      "loss": 0.4908,
+      "step": 7003
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.096899255775671e-08,
+      "loss": 0.4701,
+      "step": 7004
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.055628139050587e-08,
+      "loss": 0.4757,
+      "step": 7005
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.014449999841745e-08,
+      "loss": 0.4514,
+      "step": 7006
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.973364845947741e-08,
+      "loss": 0.4512,
+      "step": 7007
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.932372685149404e-08,
+      "loss": 0.4164,
+      "step": 7008
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.891473525210082e-08,
+      "loss": 0.4254,
+      "step": 7009
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.850667373875465e-08,
+      "loss": 0.4831,
+      "step": 7010
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.809954238873597e-08,
+      "loss": 0.4686,
+      "step": 7011
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.769334127914975e-08,
+      "loss": 0.4433,
+      "step": 7012
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.728807048692444e-08,
+      "loss": 0.453,
+      "step": 7013
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.688373008881146e-08,
+      "loss": 0.4599,
+      "step": 7014
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.64803201613884e-08,
+      "loss": 0.4672,
+      "step": 7015
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.607784078105308e-08,
+      "loss": 0.49,
+      "step": 7016
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.567629202403061e-08,
+      "loss": 0.4591,
+      "step": 7017
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.527567396636738e-08,
+      "loss": 0.4603,
+      "step": 7018
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.487598668393548e-08,
+      "loss": 0.4593,
+      "step": 7019
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.447723025242882e-08,
+      "loss": 0.4675,
+      "step": 7020
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.407940474736587e-08,
+      "loss": 0.4403,
+      "step": 7021
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.368251024408969e-08,
+      "loss": 0.469,
+      "step": 7022
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.328654681776516e-08,
+      "loss": 0.4815,
+      "step": 7023
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.289151454338229e-08,
+      "loss": 0.4168,
+      "step": 7024
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.249741349575401e-08,
+      "loss": 0.4438,
+      "step": 7025
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.210424374951787e-08,
+      "loss": 0.4617,
+      "step": 7026
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.171200537913371e-08,
+      "loss": 0.4425,
+      "step": 7027
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.132069845888547e-08,
+      "loss": 0.4027,
+      "step": 7028
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.093032306288107e-08,
+      "loss": 0.4517,
+      "step": 7029
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.054087926505194e-08,
+      "loss": 0.4629,
+      "step": 7030
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.015236713915298e-08,
+      "loss": 0.4632,
+      "step": 7031
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.976478675876254e-08,
+      "loss": 0.4593,
+      "step": 7032
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.937813819728191e-08,
+      "loss": 0.4555,
+      "step": 7033
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.899242152793696e-08,
+      "loss": 0.4552,
+      "step": 7034
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.860763682377704e-08,
+      "loss": 0.4502,
+      "step": 7035
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.822378415767384e-08,
+      "loss": 0.4774,
+      "step": 7036
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.784086360232424e-08,
+      "loss": 0.5157,
+      "step": 7037
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.745887523024686e-08,
+      "loss": 0.4607,
+      "step": 7038
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.707781911378498e-08,
+      "loss": 0.4378,
+      "step": 7039
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.669769532510474e-08,
+      "loss": 0.4615,
+      "step": 7040
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.631850393619577e-08,
+      "loss": 0.48,
+      "step": 7041
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.594024501887177e-08,
+      "loss": 0.4386,
+      "step": 7042
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.556291864476873e-08,
+      "loss": 0.4935,
+      "step": 7043
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.518652488534784e-08,
+      "loss": 0.4552,
+      "step": 7044
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.481106381189041e-08,
+      "loss": 0.4482,
+      "step": 7045
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.443653549550512e-08,
+      "loss": 0.456,
+      "step": 7046
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.406294000711967e-08,
+      "loss": 0.4421,
+      "step": 7047
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.369027741749024e-08,
+      "loss": 0.4661,
+      "step": 7048
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.331854779719205e-08,
+      "loss": 0.4514,
+      "step": 7049
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.29477512166249e-08,
+      "loss": 0.4177,
+      "step": 7050
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.25778877460126e-08,
+      "loss": 0.5033,
+      "step": 7051
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.22089574554019e-08,
+      "loss": 0.4622,
+      "step": 7052
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.184096041466194e-08,
+      "loss": 0.4632,
+      "step": 7053
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.147389669348692e-08,
+      "loss": 0.4679,
+      "step": 7054
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.110776636139238e-08,
+      "loss": 0.5032,
+      "step": 7055
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.07425694877184e-08,
+      "loss": 0.4786,
+      "step": 7056
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.037830614162744e-08,
+      "loss": 0.4597,
+      "step": 7057
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.001497639210596e-08,
+      "loss": 0.4858,
+      "step": 7058
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.965258030796284e-08,
+      "loss": 0.4446,
+      "step": 7059
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.929111795782983e-08,
+      "loss": 0.4712,
+      "step": 7060
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.893058941016439e-08,
+      "loss": 0.4701,
+      "step": 7061
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.857099473324302e-08,
+      "loss": 0.461,
+      "step": 7062
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.821233399516902e-08,
+      "loss": 0.4804,
+      "step": 7063
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.785460726386695e-08,
+      "loss": 0.4345,
+      "step": 7064
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.749781460708482e-08,
+      "loss": 0.4434,
+      "step": 7065
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.71419560923936e-08,
+      "loss": 0.4938,
+      "step": 7066
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.678703178718882e-08,
+      "loss": 0.4587,
+      "step": 7067
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.643304175868614e-08,
+      "loss": 0.4658,
+      "step": 7068
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.607998607392641e-08,
+      "loss": 0.4885,
+      "step": 7069
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.572786479977388e-08,
+      "loss": 0.4305,
+      "step": 7070
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.53766780029147e-08,
+      "loss": 0.4494,
+      "step": 7071
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.502642574985784e-08,
+      "loss": 0.4642,
+      "step": 7072
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.467710810693639e-08,
+      "loss": 0.4695,
+      "step": 7073
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.43287251403052e-08,
+      "loss": 0.4667,
+      "step": 7074
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.398127691594424e-08,
+      "loss": 0.4702,
+      "step": 7075
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.36347634996537e-08,
+      "loss": 0.4714,
+      "step": 7076
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.328918495705826e-08,
+      "loss": 0.4249,
+      "step": 7077
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.294454135360562e-08,
+      "loss": 0.4539,
+      "step": 7078
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.260083275456574e-08,
+      "loss": 0.463,
+      "step": 7079
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.225805922503215e-08,
+      "loss": 0.4766,
+      "step": 7080
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.191622082992177e-08,
+      "loss": 0.4593,
+      "step": 7081
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.157531763397228e-08,
+      "loss": 0.4547,
+      "step": 7082
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.123534970174594e-08,
+      "loss": 0.4692,
+      "step": 7083
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.089631709762844e-08,
+      "loss": 0.4736,
+      "step": 7084
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.055821988582677e-08,
+      "loss": 0.4583,
+      "step": 7085
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.022105813037193e-08,
+      "loss": 0.4472,
+      "step": 7086
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.988483189511674e-08,
+      "loss": 0.4862,
+      "step": 7087
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.9549541243738015e-08,
+      "loss": 0.457,
+      "step": 7088
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.9215186239734436e-08,
+      "loss": 0.4674,
+      "step": 7089
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.8881766946427554e-08,
+      "loss": 0.4351,
+      "step": 7090
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.854928342696242e-08,
+      "loss": 0.4644,
+      "step": 7091
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.821773574430645e-08,
+      "loss": 0.4347,
+      "step": 7092
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.788712396125051e-08,
+      "loss": 0.4393,
+      "step": 7093
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.7557448140405645e-08,
+      "loss": 0.4592,
+      "step": 7094
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.7228708344209125e-08,
+      "loss": 0.4792,
+      "step": 7095
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.690090463491838e-08,
+      "loss": 0.4585,
+      "step": 7096
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.657403707461539e-08,
+      "loss": 0.4438,
+      "step": 7097
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.6248105725204005e-08,
+      "loss": 0.4807,
+      "step": 7098
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.592311064840927e-08,
+      "loss": 0.4873,
+      "step": 7099
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.559905190578196e-08,
+      "loss": 0.4298,
+      "step": 7100
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.527592955869354e-08,
+      "loss": 0.4782,
+      "step": 7101
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.495374366833839e-08,
+      "loss": 0.4615,
+      "step": 7102
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.463249429573325e-08,
+      "loss": 0.4548,
+      "step": 7103
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.431218150171891e-08,
+      "loss": 0.4867,
+      "step": 7104
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.399280534695739e-08,
+      "loss": 0.488,
+      "step": 7105
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.367436589193364e-08,
+      "loss": 0.4749,
+      "step": 7106
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.335686319695499e-08,
+      "loss": 0.469,
+      "step": 7107
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.3040297322152215e-08,
+      "loss": 0.485,
+      "step": 7108
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.2724668327478466e-08,
+      "loss": 0.4608,
+      "step": 7109
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.240997627270872e-08,
+      "loss": 0.4574,
+      "step": 7110
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.209622121744029e-08,
+      "loss": 0.4233,
+      "step": 7111
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.178340322109454e-08,
+      "loss": 0.4641,
+      "step": 7112
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.147152234291519e-08,
+      "loss": 0.4452,
+      "step": 7113
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.116057864196555e-08,
+      "loss": 0.4736,
+      "step": 7114
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.085057217713574e-08,
+      "loss": 0.4603,
+      "step": 7115
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.054150300713601e-08,
+      "loss": 0.4672,
+      "step": 7116
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.0233371190498424e-08,
+      "loss": 0.4108,
+      "step": 7117
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.992617678557965e-08,
+      "loss": 0.469,
+      "step": 7118
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.9619919850557585e-08,
+      "loss": 0.4949,
+      "step": 7119
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.9314600443431945e-08,
+      "loss": 0.4701,
+      "step": 7120
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.901021862202593e-08,
+      "loss": 0.4821,
+      "step": 7121
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8706774443985635e-08,
+      "loss": 0.4729,
+      "step": 7122
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8404267966777886e-08,
+      "loss": 0.4597,
+      "step": 7123
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8102699247694064e-08,
+      "loss": 0.4744,
+      "step": 7124
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.780206834384515e-08,
+      "loss": 0.4476,
+      "step": 7125
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.7502375312167284e-08,
+      "loss": 0.4716,
+      "step": 7126
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.720362020941782e-08,
+      "loss": 0.4332,
+      "step": 7127
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6905803092175406e-08,
+      "loss": 0.4834,
+      "step": 7128
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6608924016843806e-08,
+      "loss": 0.4882,
+      "step": 7129
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6312983039646373e-08,
+      "loss": 0.4487,
+      "step": 7130
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.601798021663051e-08,
+      "loss": 0.4708,
+      "step": 7131
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.572391560366429e-08,
+      "loss": 0.4902,
+      "step": 7132
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5430789256440397e-08,
+      "loss": 0.4706,
+      "step": 7133
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.513860123047164e-08,
+      "loss": 0.4717,
+      "step": 7134
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.484735158109488e-08,
+      "loss": 0.4369,
+      "step": 7135
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.455704036346764e-08,
+      "loss": 0.4981,
+      "step": 7136
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4267667632570954e-08,
+      "loss": 0.4651,
+      "step": 7137
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.39792334432082e-08,
+      "loss": 0.4471,
+      "step": 7138
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3691737850002913e-08,
+      "loss": 0.4104,
+      "step": 7139
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.340518090740431e-08,
+      "loss": 0.4613,
+      "step": 7140
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.311956266968065e-08,
+      "loss": 0.4236,
+      "step": 7141
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.283488319092477e-08,
+      "loss": 0.4594,
+      "step": 7142
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.255114252504966e-08,
+      "loss": 0.4712,
+      "step": 7143
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.226834072579233e-08,
+      "loss": 0.4424,
+      "step": 7144
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1986477846710505e-08,
+      "loss": 0.4431,
+      "step": 7145
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1705553941185916e-08,
+      "loss": 0.4564,
+      "step": 7146
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1425569062419904e-08,
+      "loss": 0.4669,
+      "step": 7147
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1146523263438934e-08,
+      "loss": 0.4746,
+      "step": 7148
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.086841659708907e-08,
+      "loss": 0.4644,
+      "step": 7149
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.059124911603984e-08,
+      "loss": 0.4746,
+      "step": 7150
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0315020872782604e-08,
+      "loss": 0.4623,
+      "step": 7151
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.003973191963051e-08,
+      "loss": 0.4548,
+      "step": 7152
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.976538230872018e-08,
+      "loss": 0.4637,
+      "step": 7153
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.949197209200839e-08,
+      "loss": 0.4457,
+      "step": 7154
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.921950132127483e-08,
+      "loss": 0.4732,
+      "step": 7155
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.894797004812212e-08,
+      "loss": 0.4847,
+      "step": 7156
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.867737832397411e-08,
+      "loss": 0.4377,
+      "step": 7157
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.840772620007593e-08,
+      "loss": 0.4649,
+      "step": 7158
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8139013727497284e-08,
+      "loss": 0.4419,
+      "step": 7159
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.787124095712635e-08,
+      "loss": 0.4395,
+      "step": 7160
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7604407939677015e-08,
+      "loss": 0.4727,
+      "step": 7161
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7338514725682175e-08,
+      "loss": 0.4483,
+      "step": 7162
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.707356136549878e-08,
+      "loss": 0.4536,
+      "step": 7163
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.680954790930447e-08,
+      "loss": 0.4816,
+      "step": 7164
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6546474407100354e-08,
+      "loss": 0.4511,
+      "step": 7165
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.628434090870769e-08,
+      "loss": 0.4789,
+      "step": 7166
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.602314746377067e-08,
+      "loss": 0.4851,
+      "step": 7167
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.576289412175582e-08,
+      "loss": 0.4709,
+      "step": 7168
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.55035809319515e-08,
+      "loss": 0.4545,
+      "step": 7169
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.524520794346731e-08,
+      "loss": 0.456,
+      "step": 7170
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4987775205235216e-08,
+      "loss": 0.4736,
+      "step": 7171
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4731282766009566e-08,
+      "loss": 0.4438,
+      "step": 7172
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4475730674365384e-08,
+      "loss": 0.4389,
+      "step": 7173
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.422111897870117e-08,
+      "loss": 0.465,
+      "step": 7174
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.396744772723615e-08,
+      "loss": 0.4891,
+      "step": 7175
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3714716968011874e-08,
+      "loss": 0.4615,
+      "step": 7176
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3462926748891736e-08,
+      "loss": 0.4745,
+      "step": 7177
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.321207711756147e-08,
+      "loss": 0.4792,
+      "step": 7178
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.296216812152808e-08,
+      "loss": 0.4655,
+      "step": 7179
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.271319980811982e-08,
+      "loss": 0.4147,
+      "step": 7180
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2465172224488416e-08,
+      "loss": 0.4457,
+      "step": 7181
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2218085417606846e-08,
+      "loss": 0.4471,
+      "step": 7182
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1971939434268796e-08,
+      "loss": 0.469,
+      "step": 7183
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.172673432109086e-08,
+      "loss": 0.4685,
+      "step": 7184
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1482470124512e-08,
+      "loss": 0.4627,
+      "step": 7185
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1239146890790775e-08,
+      "loss": 0.483,
+      "step": 7186
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.099676466600976e-08,
+      "loss": 0.4753,
+      "step": 7187
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0755323496073356e-08,
+      "loss": 0.4375,
+      "step": 7188
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.051482342670553e-08,
+      "loss": 0.4712,
+      "step": 7189
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.027526450345375e-08,
+      "loss": 0.4653,
+      "step": 7190
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0036646771687275e-08,
+      "loss": 0.4635,
+      "step": 7191
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.979897027659662e-08,
+      "loss": 0.4623,
+      "step": 7192
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9562235063193557e-08,
+      "loss": 0.4632,
+      "step": 7193
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9326441176312782e-08,
+      "loss": 0.4861,
+      "step": 7194
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9091588660610238e-08,
+      "loss": 0.4701,
+      "step": 7195
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.885767756056368e-08,
+      "loss": 0.4546,
+      "step": 7196
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8624707920471563e-08,
+      "loss": 0.4419,
+      "step": 7197
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8392679784455257e-08,
+      "loss": 0.4778,
+      "step": 7198
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8161593196457394e-08,
+      "loss": 0.4331,
+      "step": 7199
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7931448200242405e-08,
+      "loss": 0.4857,
+      "step": 7200
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7702244839395986e-08,
+      "loss": 0.4839,
+      "step": 7201
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.747398315732619e-08,
+      "loss": 0.4633,
+      "step": 7202
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7246663197262324e-08,
+      "loss": 0.4443,
+      "step": 7203
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7020285002255508e-08,
+      "loss": 0.4795,
+      "step": 7204
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6794848615178116e-08,
+      "loss": 0.466,
+      "step": 7205
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6570354078724327e-08,
+      "loss": 0.4653,
+      "step": 7206
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.634680143541013e-08,
+      "loss": 0.4719,
+      "step": 7207
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6124190727573327e-08,
+      "loss": 0.46,
+      "step": 7208
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5902521997372977e-08,
+      "loss": 0.4663,
+      "step": 7209
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.568179528678938e-08,
+      "loss": 0.4471,
+      "step": 7210
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5462010637625213e-08,
+      "loss": 0.4573,
+      "step": 7211
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.52431680915044e-08,
+      "loss": 0.4271,
+      "step": 7212
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5025267689872677e-08,
+      "loss": 0.4743,
+      "step": 7213
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.480830947399704e-08,
+      "loss": 0.4664,
+      "step": 7214
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4592293484965167e-08,
+      "loss": 0.4417,
+      "step": 7215
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4377219763688787e-08,
+      "loss": 0.4615,
+      "step": 7216
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4163088350898646e-08,
+      "loss": 0.4402,
+      "step": 7217
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.394989928714786e-08,
+      "loss": 0.445,
+      "step": 7218
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.373765261281191e-08,
+      "loss": 0.4676,
+      "step": 7219
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3526348368087538e-08,
+      "loss": 0.4691,
+      "step": 7220
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3315986592991614e-08,
+      "loss": 0.4562,
+      "step": 7221
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3106567327363382e-08,
+      "loss": 0.4789,
+      "step": 7222
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2898090610864455e-08,
+      "loss": 0.4412,
+      "step": 7223
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2690556482977133e-08,
+      "loss": 0.433,
+      "step": 7224
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.248396498300498e-08,
+      "loss": 0.4857,
+      "step": 7225
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2278316150073364e-08,
+      "loss": 0.4582,
+      "step": 7226
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2073610023129467e-08,
+      "loss": 0.4509,
+      "step": 7227
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1869846640941163e-08,
+      "loss": 0.4686,
+      "step": 7228
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.166702604209814e-08,
+      "loss": 0.4888,
+      "step": 7229
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.146514826501134e-08,
+      "loss": 0.4531,
+      "step": 7230
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.126421334791462e-08,
+      "loss": 0.4585,
+      "step": 7231
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.106422132886088e-08,
+      "loss": 0.4605,
+      "step": 7232
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0865172245725373e-08,
+      "loss": 0.4944,
+      "step": 7233
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0667066136206283e-08,
+      "loss": 0.4673,
+      "step": 7234
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.046990303782137e-08,
+      "loss": 0.4563,
+      "step": 7235
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.027368298790966e-08,
+      "loss": 0.4405,
+      "step": 7236
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.007840602363309e-08,
+      "loss": 0.4587,
+      "step": 7237
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.98840721819743e-08,
+      "loss": 0.4514,
+      "step": 7238
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.969068149973663e-08,
+      "loss": 0.47,
+      "step": 7239
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.949823401354578e-08,
+      "loss": 0.4664,
+      "step": 7240
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9306729759848152e-08,
+      "loss": 0.4398,
+      "step": 7241
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9116168774911958e-08,
+      "loss": 0.4786,
+      "step": 7242
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8926551094826662e-08,
+      "loss": 0.4654,
+      "step": 7243
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8737876755502983e-08,
+      "loss": 0.4423,
+      "step": 7244
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.855014579267289e-08,
+      "loss": 0.47,
+      "step": 7245
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8363358241890172e-08,
+      "loss": 0.4909,
+      "step": 7246
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.817751413852875e-08,
+      "loss": 0.4659,
+      "step": 7247
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7992613517785474e-08,
+      "loss": 0.4707,
+      "step": 7248
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7808656414677884e-08,
+      "loss": 0.466,
+      "step": 7249
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7625642864044224e-08,
+      "loss": 0.469,
+      "step": 7250
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7443572900544548e-08,
+      "loss": 0.4788,
+      "step": 7251
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7262446558660163e-08,
+      "loss": 0.4656,
+      "step": 7252
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7082263872693627e-08,
+      "loss": 0.4795,
+      "step": 7253
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.690302487676987e-08,
+      "loss": 0.4473,
+      "step": 7254
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6724729604832845e-08,
+      "loss": 0.4233,
+      "step": 7255
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.654737809064888e-08,
+      "loss": 0.4783,
+      "step": 7256
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6370970367807214e-08,
+      "loss": 0.4651,
+      "step": 7257
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6195506469715017e-08,
+      "loss": 0.457,
+      "step": 7258
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.602098642960348e-08,
+      "loss": 0.4777,
+      "step": 7259
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.58474102805245e-08,
+      "loss": 0.4415,
+      "step": 7260
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5674778055350114e-08,
+      "loss": 0.4578,
+      "step": 7261
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.550308978677417e-08,
+      "loss": 0.453,
+      "step": 7262
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.533234550731233e-08,
+      "loss": 0.4623,
+      "step": 7263
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5162545249300387e-08,
+      "loss": 0.4348,
+      "step": 7264
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.499368904489651e-08,
+      "loss": 0.4649,
+      "step": 7265
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.482577692607956e-08,
+      "loss": 0.4952,
+      "step": 7266
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4658808924649659e-08,
+      "loss": 0.4767,
+      "step": 7267
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4492785072227621e-08,
+      "loss": 0.4901,
+      "step": 7268
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4327705400256075e-08,
+      "loss": 0.4676,
+      "step": 7269
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4163569939998346e-08,
+      "loss": 0.448,
+      "step": 7270
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.400037872254012e-08,
+      "loss": 0.4429,
+      "step": 7271
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3838131778786123e-08,
+      "loss": 0.4141,
+      "step": 7272
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3676829139464553e-08,
+      "loss": 0.455,
+      "step": 7273
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3516470835123197e-08,
+      "loss": 0.4715,
+      "step": 7274
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3357056896131649e-08,
+      "loss": 0.4891,
+      "step": 7275
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3198587352680204e-08,
+      "loss": 0.4579,
+      "step": 7276
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3041062234781521e-08,
+      "loss": 0.4611,
+      "step": 7277
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2884481572267294e-08,
+      "loss": 0.4665,
+      "step": 7278
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2728845394792133e-08,
+      "loss": 0.4716,
+      "step": 7279
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2574153731831907e-08,
+      "loss": 0.4588,
+      "step": 7280
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2420406612682067e-08,
+      "loss": 0.4681,
+      "step": 7281
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2267604066459881e-08,
+      "loss": 0.4776,
+      "step": 7282
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2115746122103867e-08,
+      "loss": 0.4673,
+      "step": 7283
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.196483280837435e-08,
+      "loss": 0.4831,
+      "step": 7284
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1814864153851802e-08,
+      "loss": 0.4093,
+      "step": 7285
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1665840186937948e-08,
+      "loss": 0.4599,
+      "step": 7286
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1517760935855215e-08,
+      "loss": 0.4515,
+      "step": 7287
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1370626428647835e-08,
+      "loss": 0.4702,
+      "step": 7288
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.122443669318185e-08,
+      "loss": 0.4584,
+      "step": 7289
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1079191757141783e-08,
+      "loss": 0.4412,
+      "step": 7290
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0934891648036183e-08,
+      "loss": 0.4433,
+      "step": 7291
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0791536393192637e-08,
+      "loss": 0.4653,
+      "step": 7292
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0649126019761092e-08,
+      "loss": 0.4704,
+      "step": 7293
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0507660554711085e-08,
+      "loss": 0.4355,
+      "step": 7294
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.036714002483452e-08,
+      "loss": 0.4952,
+      "step": 7295
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0227564456743999e-08,
+      "loss": 0.4971,
+      "step": 7296
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.008893387687282e-08,
+      "loss": 0.4949,
+      "step": 7297
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.951248311475537e-09,
+      "loss": 0.4983,
+      "step": 7298
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.814507786627958e-09,
+      "loss": 0.4888,
+      "step": 7299
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.678712328226592e-09,
+      "loss": 0.4454,
+      "step": 7300
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.543861961989198e-09,
+      "loss": 0.4683,
+      "step": 7301
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.409956713454237e-09,
+      "loss": 0.4872,
+      "step": 7302
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.276996607981426e-09,
+      "loss": 0.4691,
+      "step": 7303
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.144981670751174e-09,
+      "loss": 0.4493,
+      "step": 7304
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.013911926765151e-09,
+      "loss": 0.4306,
+      "step": 7305
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.883787400846277e-09,
+      "loss": 0.4592,
+      "step": 7306
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.754608117638174e-09,
+      "loss": 0.4919,
+      "step": 7307
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.626374101605161e-09,
+      "loss": 0.4545,
+      "step": 7308
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.499085377033922e-09,
+      "loss": 0.4756,
+      "step": 7309
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.372741968029063e-09,
+      "loss": 0.4736,
+      "step": 7310
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.247343898519777e-09,
+      "loss": 0.4512,
+      "step": 7311
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.12289119225429e-09,
+      "loss": 0.4716,
+      "step": 7312
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.999383872801525e-09,
+      "loss": 0.4637,
+      "step": 7313
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.876821963552216e-09,
+      "loss": 0.4819,
+      "step": 7314
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.755205487718353e-09,
+      "loss": 0.4679,
+      "step": 7315
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.634534468330956e-09,
+      "loss": 0.4465,
+      "step": 7316
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.514808928244522e-09,
+      "loss": 0.474,
+      "step": 7317
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.396028890132578e-09,
+      "loss": 0.4902,
+      "step": 7318
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.278194376491021e-09,
+      "loss": 0.4769,
+      "step": 7319
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.161305409634778e-09,
+      "loss": 0.4589,
+      "step": 7320
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.045362011701695e-09,
+      "loss": 0.4354,
+      "step": 7321
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.930364204649764e-09,
+      "loss": 0.4567,
+      "step": 7322
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.8163120102576755e-09,
+      "loss": 0.4718,
+      "step": 7323
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.7032054501253715e-09,
+      "loss": 0.5038,
+      "step": 7324
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.591044545672942e-09,
+      "loss": 0.4521,
+      "step": 7325
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.479829318142284e-09,
+      "loss": 0.4574,
+      "step": 7326
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.369559788596546e-09,
+      "loss": 0.4275,
+      "step": 7327
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.260235977917917e-09,
+      "loss": 0.4664,
+      "step": 7328
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.151857906812053e-09,
+      "loss": 0.4598,
+      "step": 7329
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.044425595803094e-09,
+      "loss": 0.4634,
+      "step": 7330
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.937939065237542e-09,
+      "loss": 0.4543,
+      "step": 7331
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.832398335282041e-09,
+      "loss": 0.4735,
+      "step": 7332
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.727803425924494e-09,
+      "loss": 0.481,
+      "step": 7333
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.624154356974609e-09,
+      "loss": 0.5327,
+      "step": 7334
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.521451148060575e-09,
+      "loss": 0.4656,
+      "step": 7335
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.419693818633498e-09,
+      "loss": 0.4314,
+      "step": 7336
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.318882387964075e-09,
+      "loss": 0.4555,
+      "step": 7337
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.2190168751459216e-09,
+      "loss": 0.4523,
+      "step": 7338
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.120097299090576e-09,
+      "loss": 0.4268,
+      "step": 7339
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.0221236785324975e-09,
+      "loss": 0.4386,
+      "step": 7340
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.925096032026844e-09,
+      "loss": 0.488,
+      "step": 7341
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.82901437794836e-09,
+      "loss": 0.4719,
+      "step": 7342
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.733878734494158e-09,
+      "loss": 0.4321,
+      "step": 7343
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6396891196809345e-09,
+      "loss": 0.4479,
+      "step": 7344
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5464455513472005e-09,
+      "loss": 0.4681,
+      "step": 7345
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.4541480471521625e-09,
+      "loss": 0.4647,
+      "step": 7346
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.362796624574617e-09,
+      "loss": 0.4945,
+      "step": 7347
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.272391300916279e-09,
+      "loss": 0.4903,
+      "step": 7348
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.182932093297898e-09,
+      "loss": 0.4402,
+      "step": 7349
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.094419018662032e-09,
+      "loss": 0.4506,
+      "step": 7350
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.00685209377083e-09,
+      "loss": 0.4294,
+      "step": 7351
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.920231335209357e-09,
+      "loss": 0.4537,
+      "step": 7352
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.834556759382268e-09,
+      "loss": 0.4654,
+      "step": 7353
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.749828382513809e-09,
+      "loss": 0.4927,
+      "step": 7354
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.666046220651698e-09,
+      "loss": 0.4649,
+      "step": 7355
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5832102896621314e-09,
+      "loss": 0.4742,
+      "step": 7356
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5013206052331162e-09,
+      "loss": 0.4779,
+      "step": 7357
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.420377182873358e-09,
+      "loss": 0.47,
+      "step": 7358
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.340380037912816e-09,
+      "loss": 0.4528,
+      "step": 7359
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2613291855010385e-09,
+      "loss": 0.4497,
+      "step": 7360
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1832246406093836e-09,
+      "loss": 0.456,
+      "step": 7361
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1060664180299072e-09,
+      "loss": 0.4715,
+      "step": 7362
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.02985453237481e-09,
+      "loss": 0.4821,
+      "step": 7363
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9545889980775453e-09,
+      "loss": 0.4757,
+      "step": 7364
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8802698293928233e-09,
+      "loss": 0.463,
+      "step": 7365
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8068970403949402e-09,
+      "loss": 0.4342,
+      "step": 7366
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7344706449800028e-09,
+      "loss": 0.4346,
+      "step": 7367
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6629906568642615e-09,
+      "loss": 0.467,
+      "step": 7368
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5924570895846656e-09,
+      "loss": 0.5088,
+      "step": 7369
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5228699564999737e-09,
+      "loss": 0.4879,
+      "step": 7370
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.454229270788533e-09,
+      "loss": 0.4683,
+      "step": 7371
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3865350454493895e-09,
+      "loss": 0.4659,
+      "step": 7372
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3197872933039546e-09,
+      "loss": 0.4907,
+      "step": 7373
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2539860269921164e-09,
+      "loss": 0.4424,
+      "step": 7374
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1891312589766846e-09,
+      "loss": 0.4565,
+      "step": 7375
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.125223001538945e-09,
+      "loss": 0.4853,
+      "step": 7376
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0622612667831053e-09,
+      "loss": 0.4643,
+      "step": 7377
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0002460666329606e-09,
+      "loss": 0.4555,
+      "step": 7378
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9391774128335595e-09,
+      "loss": 0.4884,
+      "step": 7379
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.879055316949541e-09,
+      "loss": 0.4524,
+      "step": 7380
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8198797903679066e-09,
+      "loss": 0.4853,
+      "step": 7381
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7616508442952485e-09,
+      "loss": 0.4885,
+      "step": 7382
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7043684897594115e-09,
+      "loss": 0.4601,
+      "step": 7383
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.64803273760894e-09,
+      "loss": 0.4942,
+      "step": 7384
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5926435985125222e-09,
+      "loss": 0.4711,
+      "step": 7385
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5382010829601001e-09,
+      "loss": 0.4999,
+      "step": 7386
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4847052012628703e-09,
+      "loss": 0.4592,
+      "step": 7387
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.432155963551618e-09,
+      "loss": 0.4998,
+      "step": 7388
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.380553379778382e-09,
+      "loss": 0.4962,
+      "step": 7389
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.329897459716456e-09,
+      "loss": 0.4822,
+      "step": 7390
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2801882129587218e-09,
+      "loss": 0.4806,
+      "step": 7391
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2314256489193154e-09,
+      "loss": 0.4466,
+      "step": 7392
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1836097768336273e-09,
+      "loss": 0.4231,
+      "step": 7393
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1367406057571917e-09,
+      "loss": 0.454,
+      "step": 7394
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0908181445662414e-09,
+      "loss": 0.4781,
+      "step": 7395
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0458424019577085e-09,
+      "loss": 0.4832,
+      "step": 7396
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.001813386448669e-09,
+      "loss": 0.4457,
+      "step": 7397
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.587311063791182e-10,
+      "loss": 0.4502,
+      "step": 7398
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.165955699069751e-10,
+      "loss": 0.4375,
+      "step": 7399
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.754067850119674e-10,
+      "loss": 0.463,
+      "step": 7400
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.351647594956325e-10,
+      "loss": 0.4657,
+      "step": 7401
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.958695009785411e-10,
+      "loss": 0.4683,
+      "step": 7402
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.575210169030733e-10,
+      "loss": 0.4578,
+      "step": 7403
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.20119314531198e-10,
+      "loss": 0.462,
+      "step": 7404
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.836644009461379e-10,
+      "loss": 0.4504,
+      "step": 7405
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.481562830529254e-10,
+      "loss": 0.4485,
+      "step": 7406
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.135949675756259e-10,
+      "loss": 0.4728,
+      "step": 7407
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.799804610595594e-10,
+      "loss": 0.411,
+      "step": 7408
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.473127698707447e-10,
+      "loss": 0.4761,
+      "step": 7409
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.155919001964548e-10,
+      "loss": 0.4477,
+      "step": 7410
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.848178580441065e-10,
+      "loss": 0.4609,
+      "step": 7411
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.5499064924126036e-10,
+      "loss": 0.4927,
+      "step": 7412
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.261102794367311e-10,
+      "loss": 0.453,
+      "step": 7413
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9817675410058763e-10,
+      "loss": 0.4551,
+      "step": 7414
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.711900785230427e-10,
+      "loss": 0.467,
+      "step": 7415
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.4515025781445276e-10,
+      "loss": 0.4209,
+      "step": 7416
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.200572969069837e-10,
+      "loss": 0.4313,
+      "step": 7417
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.959112005523901e-10,
+      "loss": 0.4474,
+      "step": 7418
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.727119733236805e-10,
+      "loss": 0.4529,
+      "step": 7419
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5045961961456257e-10,
+      "loss": 0.4353,
+      "step": 7420
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2915414363888778e-10,
+      "loss": 0.4393,
+      "step": 7421
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0879554943231682e-10,
+      "loss": 0.4947,
+      "step": 7422
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8938384084954408e-10,
+      "loss": 0.4805,
+      "step": 7423
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7091902156762818e-10,
+      "loss": 0.4594,
+      "step": 7424
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5340109508321654e-10,
+      "loss": 0.4845,
+      "step": 7425
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.368300647142107e-10,
+      "loss": 0.4634,
+      "step": 7426
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2120593359865595e-10,
+      "loss": 0.4469,
+      "step": 7427
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0652870469585186e-10,
+      "loss": 0.4672,
+      "step": 7428
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.279838078468661e-11,
+      "loss": 0.4842,
+      "step": 7429
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.001496446641277e-11,
+      "loss": 0.4595,
+      "step": 7430
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.817845816131652e-11,
+      "loss": 0.4417,
+      "step": 7431
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.728886411093815e-11,
+      "loss": 0.4391,
+      "step": 7432
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.734618437862715e-11,
+      "loss": 0.4808,
+      "step": 7433
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.835042084621154e-11,
+      "loss": 0.4846,
+      "step": 7434
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.030157521843879e-11,
+      "loss": 0.4824,
+      "step": 7435
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3199649018534887e-11,
+      "loss": 0.4939,
+      "step": 7436
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7044643592090126e-11,
+      "loss": 0.449,
+      "step": 7437
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1836560104838691e-11,
+      "loss": 0.4856,
+      "step": 7438
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.575399543213734e-12,
+      "loss": 0.4623,
+      "step": 7439
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.261162713792288e-12,
+      "loss": 0.443,
+      "step": 7440
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8938502444054708e-12,
+      "loss": 0.4618,
+      "step": 7441
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.73462583583384e-13,
+      "loss": 0.4594,
+      "step": 7442
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.5047,
+      "step": 7443
+    },
+    {
+      "epoch": 1.0,
+      "step": 7443,
+      "total_flos": 0.0,
+      "train_loss": 0.015140367998930135,
+      "train_runtime": 4068.4255,
+      "train_samples_per_second": 468.455,
+      "train_steps_per_second": 1.829
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 7443,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 100,
+  "total_flos": 0.0,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}