diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,31213 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "global_step": 5198,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.282051282051282e-06,
+      "loss": 0.9911,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.564102564102564e-06,
+      "loss": 1.3057,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.846153846153847e-06,
+      "loss": 1.3057,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.128205128205128e-06,
+      "loss": 1.2979,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.41025641025641e-06,
+      "loss": 1.3423,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 1.3032,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.974358974358976e-06,
+      "loss": 1.2786,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0256410256410256e-05,
+      "loss": 1.2852,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.153846153846154e-05,
+      "loss": 1.2168,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.282051282051282e-05,
+      "loss": 1.2642,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4102564102564104e-05,
+      "loss": 1.1765,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 1.2004,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 1.1538,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.794871794871795e-05,
+      "loss": 1.1597,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.923076923076923e-05,
+      "loss": 1.0989,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0512820512820512e-05,
+      "loss": 1.1257,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1794871794871795e-05,
+      "loss": 1.1636,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.307692307692308e-05,
+      "loss": 1.0798,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.435897435897436e-05,
+      "loss": 1.0796,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.564102564102564e-05,
+      "loss": 1.1372,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6923076923076923e-05,
+      "loss": 1.0229,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8205128205128207e-05,
+      "loss": 1.1206,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.948717948717949e-05,
+      "loss": 1.0796,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0769230769230774e-05,
+      "loss": 1.001,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.205128205128206e-05,
+      "loss": 1.1108,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 1.0603,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.461538461538462e-05,
+      "loss": 1.0583,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.58974358974359e-05,
+      "loss": 1.0632,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.717948717948718e-05,
+      "loss": 1.0579,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.846153846153846e-05,
+      "loss": 1.0059,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.974358974358974e-05,
+      "loss": 1.0317,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1025641025641023e-05,
+      "loss": 1.0796,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.230769230769231e-05,
+      "loss": 1.0,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.358974358974359e-05,
+      "loss": 1.0261,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4871794871794874e-05,
+      "loss": 1.0081,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.615384615384616e-05,
+      "loss": 0.989,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7435897435897435e-05,
+      "loss": 1.0193,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.871794871794872e-05,
+      "loss": 0.3206,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-05,
+      "loss": 1.0166,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.128205128205128e-05,
+      "loss": 1.0398,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.256410256410257e-05,
+      "loss": 1.0105,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.384615384615385e-05,
+      "loss": 1.0217,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.512820512820514e-05,
+      "loss": 1.053,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.6410256410256414e-05,
+      "loss": 0.9395,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.769230769230769e-05,
+      "loss": 1.0188,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.897435897435898e-05,
+      "loss": 1.0288,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.025641025641026e-05,
+      "loss": 0.99,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.153846153846155e-05,
+      "loss": 0.9556,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.282051282051282e-05,
+      "loss": 0.979,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.410256410256412e-05,
+      "loss": 0.9739,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.538461538461539e-05,
+      "loss": 0.9648,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.666666666666667e-05,
+      "loss": 1.0012,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.794871794871795e-05,
+      "loss": 1.0083,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.923076923076924e-05,
+      "loss": 0.9692,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.051282051282052e-05,
+      "loss": 0.9604,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.17948717948718e-05,
+      "loss": 0.9893,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.307692307692307e-05,
+      "loss": 0.3201,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.435897435897436e-05,
+      "loss": 0.9736,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.564102564102564e-05,
+      "loss": 0.9995,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.692307692307693e-05,
+      "loss": 1.0271,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.820512820512821e-05,
+      "loss": 0.9863,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.948717948717948e-05,
+      "loss": 0.967,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.076923076923078e-05,
+      "loss": 0.9482,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.205128205128205e-05,
+      "loss": 0.9287,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.333333333333334e-05,
+      "loss": 0.9961,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.461538461538461e-05,
+      "loss": 0.97,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.58974358974359e-05,
+      "loss": 0.9419,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.717948717948718e-05,
+      "loss": 0.8909,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.846153846153847e-05,
+      "loss": 0.9736,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.974358974358975e-05,
+      "loss": 0.9631,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.102564102564103e-05,
+      "loss": 0.9502,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.230769230769232e-05,
+      "loss": 0.9346,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.35897435897436e-05,
+      "loss": 0.8926,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.487179487179487e-05,
+      "loss": 0.9878,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.615384615384617e-05,
+      "loss": 0.936,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.743589743589744e-05,
+      "loss": 0.9268,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.871794871794872e-05,
+      "loss": 0.9359,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001,
+      "loss": 0.9084,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010128205128205129,
+      "loss": 0.9607,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010256410256410256,
+      "loss": 1.0171,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010384615384615386,
+      "loss": 0.9541,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010512820512820514,
+      "loss": 0.9966,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010641025641025641,
+      "loss": 1.0237,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001076923076923077,
+      "loss": 1.0005,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010897435897435896,
+      "loss": 0.9644,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011025641025641027,
+      "loss": 0.9412,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011153846153846154,
+      "loss": 0.9492,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011282051282051283,
+      "loss": 1.0034,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001141025641025641,
+      "loss": 0.9585,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011538461538461538,
+      "loss": 0.2935,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011666666666666668,
+      "loss": 0.9119,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011794871794871796,
+      "loss": 0.9658,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011923076923076923,
+      "loss": 0.9883,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012051282051282052,
+      "loss": 0.9814,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012179487179487179,
+      "loss": 0.9363,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001230769230769231,
+      "loss": 0.3049,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012435897435897437,
+      "loss": 0.9514,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012564102564102564,
+      "loss": 0.9421,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012692307692307693,
+      "loss": 0.9761,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012820512820512823,
+      "loss": 0.8861,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001294871794871795,
+      "loss": 0.2889,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013076923076923077,
+      "loss": 0.9326,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013205128205128204,
+      "loss": 1.0322,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013333333333333334,
+      "loss": 0.9812,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013461538461538464,
+      "loss": 0.9231,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001358974358974359,
+      "loss": 0.2635,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013717948717948718,
+      "loss": 0.9312,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013846153846153847,
+      "loss": 1.01,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013974358974358974,
+      "loss": 0.9314,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014102564102564104,
+      "loss": 0.9646,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001423076923076923,
+      "loss": 0.9473,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001435897435897436,
+      "loss": 0.9421,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014487179487179488,
+      "loss": 1.0051,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014615384615384615,
+      "loss": 0.9463,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014743589743589745,
+      "loss": 0.9583,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014871794871794872,
+      "loss": 0.967,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 0.2896,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015128205128205128,
+      "loss": 0.9856,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015256410256410255,
+      "loss": 0.9418,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015384615384615385,
+      "loss": 1.0398,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015512820512820515,
+      "loss": 0.9056,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015641025641025642,
+      "loss": 0.9714,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001576923076923077,
+      "loss": 1.0107,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015897435897435896,
+      "loss": 0.8979,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00016025641025641028,
+      "loss": 0.9434,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00016153846153846155,
+      "loss": 0.947,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00016282051282051282,
+      "loss": 0.9348,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001641025641025641,
+      "loss": 0.9946,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001653846153846154,
+      "loss": 0.8452,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001666666666666667,
+      "loss": 0.9248,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00016794871794871796,
+      "loss": 0.9482,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00016923076923076923,
+      "loss": 0.9539,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017051282051282053,
+      "loss": 0.8955,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001717948717948718,
+      "loss": 0.9031,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001730769230769231,
+      "loss": 0.9282,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017435897435897436,
+      "loss": 0.9341,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017564102564102566,
+      "loss": 0.905,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017692307692307693,
+      "loss": 0.9768,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017820512820512823,
+      "loss": 0.9822,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001794871794871795,
+      "loss": 0.8977,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018076923076923077,
+      "loss": 0.9932,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018205128205128207,
+      "loss": 0.9468,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018333333333333334,
+      "loss": 0.9482,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018461538461538463,
+      "loss": 0.9897,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001858974358974359,
+      "loss": 0.9683,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001871794871794872,
+      "loss": 0.9407,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018846153846153847,
+      "loss": 0.9431,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018974358974358974,
+      "loss": 0.9031,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019102564102564104,
+      "loss": 0.9187,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019230769230769233,
+      "loss": 0.9221,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001935897435897436,
+      "loss": 0.9373,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019487179487179487,
+      "loss": 0.9734,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019615384615384615,
+      "loss": 0.279,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019743589743589744,
+      "loss": 0.9597,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019871794871794874,
+      "loss": 0.8979,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.8833,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999998058827846,
+      "loss": 1.0046,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999992235312134,
+      "loss": 0.9485,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999982529455128,
+      "loss": 0.9712,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999968941260594,
+      "loss": 0.9333,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999951470733808,
+      "loss": 0.9351,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999993011788155,
+      "loss": 0.9385,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999904882712116,
+      "loss": 0.9622,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000199998757652353,
+      "loss": 0.9269,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000199998427654624,
+      "loss": 0.3242,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999980588340624,
+      "loss": 0.9358,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999976511908113,
+      "loss": 0.947,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999720472502902,
+      "loss": 0.9563,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999671943688883,
+      "loss": 0.9199,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999619532657916,
+      "loss": 0.9216,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999563239430352,
+      "loss": 0.9082,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999503064028042,
+      "loss": 0.8121,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999943900647435,
+      "loss": 0.9695,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999371066794147,
+      "loss": 0.9285,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999299245013804,
+      "loss": 0.9438,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999922354116121,
+      "loss": 0.9358,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999914395526575,
+      "loss": 0.9241,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999906048735833,
+      "loss": 0.9148,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019998973137471353,
+      "loss": 0.8811,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019998881905638726,
+      "loss": 0.9421,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999878679189587,
+      "loss": 0.2761,
+      "step": 181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998687796279717,
+      "loss": 0.9131,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998584918828695,
+      "loss": 0.9875,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998478159582748,
+      "loss": 0.8845,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998367518583317,
+      "loss": 0.9155,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998252995873366,
+      "loss": 0.969,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999813459149735,
+      "loss": 0.9573,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999801230550124,
+      "loss": 0.3327,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997886137932512,
+      "loss": 0.9065,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997756088840148,
+      "loss": 0.8701,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997622158274636,
+      "loss": 0.2866,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997484346287973,
+      "loss": 0.9377,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997342652933666,
+      "loss": 0.9258,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997197078266724,
+      "loss": 1.0151,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997047622343658,
+      "loss": 0.9512,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996894285222497,
+      "loss": 0.8955,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996737066962775,
+      "loss": 0.9612,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996575967625523,
+      "loss": 0.9817,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999641098727329,
+      "loss": 0.9448,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996242125970126,
+      "loss": 0.8928,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996069383781587,
+      "loss": 0.9072,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995892760774737,
+      "loss": 0.9883,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999571225701815,
+      "loss": 0.9705,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995527872581902,
+      "loss": 0.9417,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995339607537579,
+      "loss": 0.8994,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995147461958267,
+      "loss": 0.3016,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999495143591857,
+      "loss": 0.9155,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999475152949459,
+      "loss": 0.9224,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994547742763934,
+      "loss": 0.9155,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994340075805725,
+      "loss": 0.9629,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994128528700583,
+      "loss": 0.9463,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993913101530635,
+      "loss": 0.9382,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993693794379525,
+      "loss": 0.8479,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999347060733239,
+      "loss": 0.8938,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993243540475877,
+      "loss": 0.9324,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993012593898146,
+      "loss": 0.929,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019992777767688854,
+      "loss": 0.9507,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019992539061939175,
+      "loss": 0.2731,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019992296476741776,
+      "loss": 0.2989,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019992050012190843,
+      "loss": 0.9783,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991799668382058,
+      "loss": 0.2691,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991545445412613,
+      "loss": 0.8954,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999128734338121,
+      "loss": 0.9231,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991025362388044,
+      "loss": 0.9924,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019990759502534834,
+      "loss": 0.9756,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019990489763924797,
+      "loss": 0.9124,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019990216146662646,
+      "loss": 0.2668,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019989938650854616,
+      "loss": 0.8877,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019989657276608437,
+      "loss": 0.9661,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019989372024033352,
+      "loss": 0.8909,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000199890828932401,
+      "loss": 0.8787,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019988789884340936,
+      "loss": 0.9041,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019988492997449615,
+      "loss": 0.9189,
+      "step": 233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019988192232681397,
+      "loss": 0.9385,
+      "step": 234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019987887590153055,
+      "loss": 0.9155,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019987579069982855,
+      "loss": 0.9202,
+      "step": 236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019987266672290575,
+      "loss": 0.9717,
+      "step": 237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019986950397197503,
+      "loss": 0.9509,
+      "step": 238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019986630244826426,
+      "loss": 0.9622,
+      "step": 239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019986306215301637,
+      "loss": 0.9265,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019985978308748937,
+      "loss": 0.9668,
+      "step": 241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019985646525295632,
+      "loss": 0.9453,
+      "step": 242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998531086507053,
+      "loss": 0.8621,
+      "step": 243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019984971328203946,
+      "loss": 0.887,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019984627914827697,
+      "loss": 0.918,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019984280625075113,
+      "loss": 0.9229,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998392945908102,
+      "loss": 0.9832,
+      "step": 247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019983574416981755,
+      "loss": 0.9089,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019983215498915158,
+      "loss": 0.9111,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998285270502057,
+      "loss": 0.8679,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019982486035438846,
+      "loss": 0.8566,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019982115490312332,
+      "loss": 0.3115,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998174106978489,
+      "loss": 0.9272,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019981362774001887,
+      "loss": 0.9312,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019980980603110185,
+      "loss": 0.916,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019980594557258155,
+      "loss": 0.9451,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019980204636595682,
+      "loss": 0.8926,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019979810841274132,
+      "loss": 0.8425,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019979413171446403,
+      "loss": 0.9263,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019979011627266883,
+      "loss": 0.9692,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019978606208891458,
+      "loss": 0.916,
+      "step": 261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997819691647753,
+      "loss": 0.9194,
+      "step": 262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019977783750183998,
+      "loss": 0.7986,
+      "step": 263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019977366710171272,
+      "loss": 0.9453,
+      "step": 264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019976945796601258,
+      "loss": 0.9231,
+      "step": 265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019976521009637364,
+      "loss": 0.9114,
+      "step": 266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997609234944452,
+      "loss": 0.917,
+      "step": 267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019975659816189137,
+      "loss": 0.9189,
+      "step": 268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997522341003914,
+      "loss": 1.002,
+      "step": 269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019974783131163957,
+      "loss": 0.8383,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019974338979734523,
+      "loss": 0.9153,
+      "step": 271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019973890955923268,
+      "loss": 0.9324,
+      "step": 272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019973439059904133,
+      "loss": 0.9473,
+      "step": 273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019972983291852563,
+      "loss": 0.9465,
+      "step": 274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019972523651945497,
+      "loss": 0.9299,
+      "step": 275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019972060140361383,
+      "loss": 0.9521,
+      "step": 276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019971592757280177,
+      "loss": 0.916,
+      "step": 277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997112150288333,
+      "loss": 0.9607,
+      "step": 278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000199706463773538,
+      "loss": 0.9417,
+      "step": 279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997016738087605,
+      "loss": 0.9414,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019969684513636033,
+      "loss": 0.9453,
+      "step": 281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019969197775821225,
+      "loss": 0.9053,
+      "step": 282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019968707167620592,
+      "loss": 0.9067,
+      "step": 283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019968212689224602,
+      "loss": 0.8933,
+      "step": 284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019967714340825233,
+      "loss": 0.9597,
+      "step": 285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019967212122615956,
+      "loss": 0.8433,
+      "step": 286
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019966706034791752,
+      "loss": 0.897,
+      "step": 287
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019966196077549105,
+      "loss": 0.833,
+      "step": 288
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001996568225108599,
+      "loss": 0.8732,
+      "step": 289
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019965164555601898,
+      "loss": 0.938,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019964642991297815,
+      "loss": 0.9111,
+      "step": 291
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001996411755837623,
+      "loss": 0.9495,
+      "step": 292
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019963588257041135,
+      "loss": 0.8972,
+      "step": 293
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019963055087498022,
+      "loss": 0.887,
+      "step": 294
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019962518049953886,
+      "loss": 0.9553,
+      "step": 295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019961977144617224,
+      "loss": 0.8906,
+      "step": 296
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001996143237169803,
+      "loss": 0.9443,
+      "step": 297
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019960883731407812,
+      "loss": 0.314,
+      "step": 298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019960331223959564,
+      "loss": 0.9031,
+      "step": 299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995977484956779,
+      "loss": 0.9324,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019959214608448496,
+      "loss": 0.8633,
+      "step": 301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019958650500819184,
+      "loss": 0.9441,
+      "step": 302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995808252689886,
+      "loss": 0.9751,
+      "step": 303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019957510686908033,
+      "loss": 0.8794,
+      "step": 304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019956934981068712,
+      "loss": 0.8608,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019956355409604402,
+      "loss": 0.9194,
+      "step": 306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019955771972740116,
+      "loss": 0.866,
+      "step": 307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995518467070236,
+      "loss": 0.9438,
+      "step": 308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019954593503719153,
+      "loss": 0.8555,
+      "step": 309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019953998472019995,
+      "loss": 0.9121,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995339957583591,
+      "loss": 0.9248,
+      "step": 311
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019952796815399403,
+      "loss": 0.8989,
+      "step": 312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019952190190944484,
+      "loss": 0.8491,
+      "step": 313
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019951579702706668,
+      "loss": 0.9358,
+      "step": 314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019950965350922975,
+      "loss": 0.9697,
+      "step": 315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019950347135831907,
+      "loss": 0.8517,
+      "step": 316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994972505767348,
+      "loss": 0.9263,
+      "step": 317
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019949099116689208,
+      "loss": 0.9241,
+      "step": 318
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000199484693131221,
+      "loss": 0.9119,
+      "step": 319
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019947835647216668,
+      "loss": 0.9045,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019947198119218924,
+      "loss": 0.896,
+      "step": 321
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994655672937638,
+      "loss": 0.8889,
+      "step": 322
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019945911477938042,
+      "loss": 0.2776,
+      "step": 323
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994526236515442,
+      "loss": 0.9895,
+      "step": 324
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019944609391277528,
+      "loss": 0.9255,
+      "step": 325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994395255656086,
+      "loss": 0.9658,
+      "step": 326
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019943291861259431,
+      "loss": 0.8491,
+      "step": 327
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019942627305629746,
+      "loss": 0.9607,
+      "step": 328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019941958889929806,
+      "loss": 0.9033,
+      "step": 329
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994128661441911,
+      "loss": 0.8882,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019940610479358669,
+      "loss": 0.8464,
+      "step": 331
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019939930485010968,
+      "loss": 0.9331,
+      "step": 332
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019939246631640013,
+      "loss": 0.9185,
+      "step": 333
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019938558919511298,
+      "loss": 0.8958,
+      "step": 334
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019937867348891815,
+      "loss": 0.8589,
+      "step": 335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019937171920050056,
+      "loss": 0.9194,
+      "step": 336
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019936472633256012,
+      "loss": 0.9211,
+      "step": 337
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019935769488781168,
+      "loss": 0.8997,
+      "step": 338
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001993506248689851,
+      "loss": 0.9136,
+      "step": 339
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001993435162788252,
+      "loss": 0.9336,
+      "step": 340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001993363691200918,
+      "loss": 0.895,
+      "step": 341
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019932918339555963,
+      "loss": 0.9387,
+      "step": 342
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019932195910801848,
+      "loss": 0.8525,
+      "step": 343
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019931469626027304,
+      "loss": 0.9592,
+      "step": 344
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019930739485514302,
+      "loss": 0.8782,
+      "step": 345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019930005489546306,
+      "loss": 0.8682,
+      "step": 346
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019929267638408278,
+      "loss": 0.9072,
+      "step": 347
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019928525932386677,
+      "loss": 0.9475,
+      "step": 348
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019927780371769462,
+      "loss": 0.8984,
+      "step": 349
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019927030956846083,
+      "loss": 0.912,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001992627768790749,
+      "loss": 0.8569,
+      "step": 351
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019925520565246122,
+      "loss": 0.9143,
+      "step": 352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001992475958915593,
+      "loss": 0.3037,
+      "step": 353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019923994759932343,
+      "loss": 0.8801,
+      "step": 354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019923226077872298,
+      "loss": 0.97,
+      "step": 355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001992245354327422,
+      "loss": 0.8887,
+      "step": 356
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019921677156438045,
+      "loss": 0.9031,
+      "step": 357
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019920896917665178,
+      "loss": 0.8975,
+      "step": 358
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001992011282725854,
+      "loss": 0.9578,
+      "step": 359
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019919324885522545,
+      "loss": 0.9612,
+      "step": 360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000199185330927631,
+      "loss": 0.9155,
+      "step": 361
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019917737449287598,
+      "loss": 0.9084,
+      "step": 362
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019916937955404942,
+      "loss": 0.8813,
+      "step": 363
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019916134611425523,
+      "loss": 0.8828,
+      "step": 364
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019915327417661225,
+      "loss": 0.9014,
+      "step": 365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019914516374425429,
+      "loss": 0.8719,
+      "step": 366
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019913701482033008,
+      "loss": 0.9233,
+      "step": 367
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019912882740800333,
+      "loss": 0.8855,
+      "step": 368
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001991206015104527,
+      "loss": 0.8965,
+      "step": 369
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001991123371308717,
+      "loss": 0.9089,
+      "step": 370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019910403427246895,
+      "loss": 0.9546,
+      "step": 371
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001990956929384678,
+      "loss": 0.9199,
+      "step": 372
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001990873131321067,
+      "loss": 0.9067,
+      "step": 373
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019907889485663895,
+      "loss": 0.8865,
+      "step": 374
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019907043811533283,
+      "loss": 0.9202,
+      "step": 375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019906194291147156,
+      "loss": 0.8845,
+      "step": 376
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019905340924835323,
+      "loss": 0.8867,
+      "step": 377
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019904483712929094,
+      "loss": 0.9412,
+      "step": 378
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019903622655761266,
+      "loss": 0.8591,
+      "step": 379
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001990275775366613,
+      "loss": 0.9077,
+      "step": 380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019901889006979472,
+      "loss": 0.8979,
+      "step": 381
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001990101641603857,
+      "loss": 0.9138,
+      "step": 382
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019900139981182194,
+      "loss": 0.8848,
+      "step": 383
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019899259702750604,
+      "loss": 0.8813,
+      "step": 384
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019898375581085557,
+      "loss": 0.8862,
+      "step": 385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019897487616530297,
+      "loss": 0.9031,
+      "step": 386
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019896595809429566,
+      "loss": 0.9258,
+      "step": 387
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001989570016012959,
+      "loss": 0.9084,
+      "step": 388
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019894800668978095,
+      "loss": 0.8884,
+      "step": 389
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001989389733632429,
+      "loss": 0.9243,
+      "step": 390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019892990162518882,
+      "loss": 0.9348,
+      "step": 391
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019892079147914072,
+      "loss": 0.9243,
+      "step": 392
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019891164292863537,
+      "loss": 0.9343,
+      "step": 393
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019890245597722463,
+      "loss": 0.8966,
+      "step": 394
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019889323062847515,
+      "loss": 0.9424,
+      "step": 395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001988839668859686,
+      "loss": 0.9041,
+      "step": 396
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019887466475330138,
+      "loss": 0.9358,
+      "step": 397
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019886532423408495,
+      "loss": 0.8796,
+      "step": 398
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019885594533194562,
+      "loss": 0.9397,
+      "step": 399
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019884652805052464,
+      "loss": 0.9268,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019883707239347803,
+      "loss": 0.8513,
+      "step": 401
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001988275783644769,
+      "loss": 0.8855,
+      "step": 402
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001988180459672071,
+      "loss": 0.895,
+      "step": 403
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019880847520536946,
+      "loss": 0.9417,
+      "step": 404
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019879886608267967,
+      "loss": 0.9067,
+      "step": 405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001987892186028683,
+      "loss": 0.938,
+      "step": 406
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019877953276968086,
+      "loss": 0.9353,
+      "step": 407
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019876980858687775,
+      "loss": 0.8748,
+      "step": 408
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019876004605823419,
+      "loss": 0.8679,
+      "step": 409
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001987502451875403,
+      "loss": 0.2799,
+      "step": 410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001987404059786012,
+      "loss": 0.289,
+      "step": 411
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019873052843523677,
+      "loss": 0.8762,
+      "step": 412
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001987206125612818,
+      "loss": 0.9111,
+      "step": 413
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019871065836058596,
+      "loss": 0.9058,
+      "step": 414
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019870066583701387,
+      "loss": 0.9092,
+      "step": 415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019869063499444495,
+      "loss": 0.855,
+      "step": 416
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019868056583677345,
+      "loss": 0.9392,
+      "step": 417
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019867045836790867,
+      "loss": 0.9333,
+      "step": 418
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019866031259177462,
+      "loss": 0.8779,
+      "step": 419
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019865012851231023,
+      "loss": 0.9309,
+      "step": 420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019863990613346933,
+      "loss": 0.9219,
+      "step": 421
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001986296454592206,
+      "loss": 0.9312,
+      "step": 422
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019861934649354762,
+      "loss": 0.9087,
+      "step": 423
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019860900924044872,
+      "loss": 0.9385,
+      "step": 424
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019859863370393725,
+      "loss": 0.8926,
+      "step": 425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019858821988804132,
+      "loss": 0.8113,
+      "step": 426
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019857776779680394,
+      "loss": 0.9358,
+      "step": 427
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019856727743428296,
+      "loss": 0.887,
+      "step": 428
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019855674880455112,
+      "loss": 0.9102,
+      "step": 429
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019854618191169599,
+      "loss": 0.8877,
+      "step": 430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019853557675982,
+      "loss": 0.9578,
+      "step": 431
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001985249333530404,
+      "loss": 0.9041,
+      "step": 432
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019851425169548938,
+      "loss": 0.8508,
+      "step": 433
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001985035317913139,
+      "loss": 0.8428,
+      "step": 434
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001984927736446758,
+      "loss": 0.3389,
+      "step": 435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001984819772597518,
+      "loss": 0.3084,
+      "step": 436
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019847114264073338,
+      "loss": 0.8809,
+      "step": 437
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001984602697918269,
+      "loss": 0.9092,
+      "step": 438
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001984493587172536,
+      "loss": 0.9248,
+      "step": 439
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019843840942124956,
+      "loss": 0.8965,
+      "step": 440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019842742190806566,
+      "loss": 0.8606,
+      "step": 441
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019841639618196758,
+      "loss": 0.9517,
+      "step": 442
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019840533224723593,
+      "loss": 0.8877,
+      "step": 443
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019839423010816615,
+      "loss": 0.9128,
+      "step": 444
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001983830897690684,
+      "loss": 0.908,
+      "step": 445
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019837191123426776,
+      "loss": 0.9297,
+      "step": 446
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019836069450810418,
+      "loss": 0.9121,
+      "step": 447
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001983494395949323,
+      "loss": 0.9331,
+      "step": 448
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001983381464991217,
+      "loss": 0.9141,
+      "step": 449
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019832681522505676,
+      "loss": 0.9043,
+      "step": 450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019831544577713663,
+      "loss": 0.8525,
+      "step": 451
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001983040381597754,
+      "loss": 0.9585,
+      "step": 452
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019829259237740177,
+      "loss": 0.9253,
+      "step": 453
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019828110843445954,
+      "loss": 0.8291,
+      "step": 454
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019826958633540706,
+      "loss": 0.864,
+      "step": 455
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019825802608471765,
+      "loss": 0.8962,
+      "step": 456
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001982464276868794,
+      "loss": 0.8533,
+      "step": 457
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001982347911463952,
+      "loss": 0.9326,
+      "step": 458
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019822311646778277,
+      "loss": 0.9263,
+      "step": 459
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019821140365557457,
+      "loss": 0.8862,
+      "step": 460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019819965271431796,
+      "loss": 0.8906,
+      "step": 461
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019818786364857505,
+      "loss": 0.8521,
+      "step": 462
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019817603646292276,
+      "loss": 0.8979,
+      "step": 463
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019816417116195285,
+      "loss": 0.9126,
+      "step": 464
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001981522677502718,
+      "loss": 0.859,
+      "step": 465
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019814032623250092,
+      "loss": 0.9004,
+      "step": 466
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001981283466132763,
+      "loss": 0.9695,
+      "step": 467
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019811632889724887,
+      "loss": 0.8478,
+      "step": 468
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019810427308908436,
+      "loss": 0.8408,
+      "step": 469
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019809217919346315,
+      "loss": 0.8328,
+      "step": 470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001980800472150806,
+      "loss": 0.8301,
+      "step": 471
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019806787715864673,
+      "loss": 0.9663,
+      "step": 472
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019805566902888634,
+      "loss": 0.9055,
+      "step": 473
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019804342283053915,
+      "loss": 0.9392,
+      "step": 474
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019803113856835938,
+      "loss": 0.9172,
+      "step": 475
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019801881624711637,
+      "loss": 0.9683,
+      "step": 476
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000198006455871594,
+      "loss": 0.9082,
+      "step": 477
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000197994057446591,
+      "loss": 0.9014,
+      "step": 478
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019798162097692089,
+      "loss": 0.844,
+      "step": 479
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019796914646741186,
+      "loss": 0.8979,
+      "step": 480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000197956633922907,
+      "loss": 0.8506,
+      "step": 481
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019794408334826414,
+      "loss": 0.8776,
+      "step": 482
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001979314947483558,
+      "loss": 0.8352,
+      "step": 483
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001979188681280693,
+      "loss": 0.9214,
+      "step": 484
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019790620349230674,
+      "loss": 0.9463,
+      "step": 485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000197893500845985,
+      "loss": 0.8682,
+      "step": 486
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019788076019403565,
+      "loss": 0.8999,
+      "step": 487
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019786798154140507,
+      "loss": 0.8364,
+      "step": 488
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019785516489305435,
+      "loss": 0.9453,
+      "step": 489
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019784231025395937,
+      "loss": 0.395,
+      "step": 490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019782941762911074,
+      "loss": 0.9045,
+      "step": 491
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019781648702351382,
+      "loss": 0.9019,
+      "step": 492
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001978035184421887,
+      "loss": 0.9182,
+      "step": 493
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001977905118901703,
+      "loss": 0.9197,
+      "step": 494
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019777746737250807,
+      "loss": 0.8865,
+      "step": 495
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001977643848942665,
+      "loss": 0.8723,
+      "step": 496
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019775126446052458,
+      "loss": 0.3273,
+      "step": 497
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001977381060763761,
+      "loss": 0.8708,
+      "step": 498
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019772490974692962,
+      "loss": 0.8721,
+      "step": 499
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019771167547730844,
+      "loss": 0.8936,
+      "step": 500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001976984032726505,
+      "loss": 0.9412,
+      "step": 501
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001976850931381086,
+      "loss": 0.9299,
+      "step": 502
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019767174507885008,
+      "loss": 0.9163,
+      "step": 503
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019765835910005722,
+      "loss": 0.8875,
+      "step": 504
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019764493520692687,
+      "loss": 0.8464,
+      "step": 505
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019763147340467067,
+      "loss": 0.853,
+      "step": 506
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019761797369851497,
+      "loss": 0.9778,
+      "step": 507
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019760443609370074,
+      "loss": 0.8367,
+      "step": 508
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019759086059548383,
+      "loss": 0.8718,
+      "step": 509
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019757724720913467,
+      "loss": 0.8755,
+      "step": 510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019756359593993845,
+      "loss": 0.9292,
+      "step": 511
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001975499067931951,
+      "loss": 0.8811,
+      "step": 512
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001975361797742192,
+      "loss": 0.3101,
+      "step": 513
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019752241488834002,
+      "loss": 0.947,
+      "step": 514
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001975086121409016,
+      "loss": 0.9167,
+      "step": 515
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001974947715372626,
+      "loss": 0.9021,
+      "step": 516
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019748089308279648,
+      "loss": 0.8361,
+      "step": 517
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001974669767828913,
+      "loss": 0.8503,
+      "step": 518
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001974530226429498,
+      "loss": 0.8882,
+      "step": 519
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019743903066838953,
+      "loss": 0.8999,
+      "step": 520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019742500086464263,
+      "loss": 0.8962,
+      "step": 521
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019741093323715595,
+      "loss": 0.9241,
+      "step": 522
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019739682779139106,
+      "loss": 0.3688,
+      "step": 523
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001973826845328241,
+      "loss": 0.9177,
+      "step": 524
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001973685034669461,
+      "loss": 0.8789,
+      "step": 525
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019735428459926249,
+      "loss": 0.9102,
+      "step": 526
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001973400279352936,
+      "loss": 0.9148,
+      "step": 527
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019732573348057436,
+      "loss": 0.8635,
+      "step": 528
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019731140124065437,
+      "loss": 0.8542,
+      "step": 529
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019729703122109787,
+      "loss": 0.938,
+      "step": 530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019728262342748383,
+      "loss": 0.8928,
+      "step": 531
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019726817786540584,
+      "loss": 0.8386,
+      "step": 532
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019725369454047214,
+      "loss": 0.9136,
+      "step": 533
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019723917345830568,
+      "loss": 0.8992,
+      "step": 534
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019722461462454404,
+      "loss": 0.9404,
+      "step": 535
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019721001804483948,
+      "loss": 0.8589,
+      "step": 536
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019719538372485885,
+      "loss": 0.8921,
+      "step": 537
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019718071167028375,
+      "loss": 0.9631,
+      "step": 538
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019716600188681037,
+      "loss": 0.8569,
+      "step": 539
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019715125438014947,
+      "loss": 0.8831,
+      "step": 540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019713646915602663,
+      "loss": 0.8987,
+      "step": 541
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019712164622018197,
+      "loss": 0.9204,
+      "step": 542
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019710678557837024,
+      "loss": 0.8647,
+      "step": 543
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019709188723636086,
+      "loss": 0.8503,
+      "step": 544
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019707695119993793,
+      "loss": 0.8987,
+      "step": 545
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019706197747490001,
+      "loss": 0.8782,
+      "step": 546
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019704696606706053,
+      "loss": 0.8743,
+      "step": 547
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019703191698224742,
+      "loss": 0.8623,
+      "step": 548
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001970168302263032,
+      "loss": 0.9177,
+      "step": 549
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019700170580508512,
+      "loss": 0.9138,
+      "step": 550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019698654372446493,
+      "loss": 0.9297,
+      "step": 551
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001969713439903292,
+      "loss": 0.9153,
+      "step": 552
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019695610660857887,
+      "loss": 0.8796,
+      "step": 553
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019694083158512964,
+      "loss": 0.8894,
+      "step": 554
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019692551892591185,
+      "loss": 0.9272,
+      "step": 555
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019691016863687035,
+      "loss": 0.9231,
+      "step": 556
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019689478072396467,
+      "loss": 0.9248,
+      "step": 557
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019687935519316897,
+      "loss": 0.8452,
+      "step": 558
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019686389205047188,
+      "loss": 0.8677,
+      "step": 559
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001968483913018768,
+      "loss": 0.8682,
+      "step": 560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001968328529534016,
+      "loss": 0.9131,
+      "step": 561
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019681727701107885,
+      "loss": 0.8431,
+      "step": 562
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019680166348095564,
+      "loss": 0.8701,
+      "step": 563
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019678601236909371,
+      "loss": 0.8575,
+      "step": 564
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019677032368156931,
+      "loss": 0.8516,
+      "step": 565
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019675459742447338,
+      "loss": 0.8704,
+      "step": 566
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019673883360391136,
+      "loss": 0.8801,
+      "step": 567
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019672303222600333,
+      "loss": 0.9387,
+      "step": 568
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019670719329688388,
+      "loss": 0.9016,
+      "step": 569
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001966913168227023,
+      "loss": 0.9346,
+      "step": 570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019667540280962235,
+      "loss": 0.8096,
+      "step": 571
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001966594512638224,
+      "loss": 0.8234,
+      "step": 572
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001966434621914954,
+      "loss": 0.9094,
+      "step": 573
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001966274355988488,
+      "loss": 0.8857,
+      "step": 574
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019661137149210473,
+      "loss": 0.8481,
+      "step": 575
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019659526987749986,
+      "loss": 0.8904,
+      "step": 576
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019657913076128532,
+      "loss": 0.3234,
+      "step": 577
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001965629541497269,
+      "loss": 0.3547,
+      "step": 578
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019654674004910492,
+      "loss": 0.8792,
+      "step": 579
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019653048846571426,
+      "loss": 0.8748,
+      "step": 580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019651419940586436,
+      "loss": 0.9282,
+      "step": 581
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019649787287587912,
+      "loss": 0.8821,
+      "step": 582
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019648150888209714,
+      "loss": 0.9197,
+      "step": 583
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019646510743087143,
+      "loss": 0.9214,
+      "step": 584
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019644866852856966,
+      "loss": 0.9382,
+      "step": 585
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019643219218157392,
+      "loss": 0.8728,
+      "step": 586
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019641567839628093,
+      "loss": 0.8884,
+      "step": 587
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001963991271791019,
+      "loss": 0.8987,
+      "step": 588
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019638253853646257,
+      "loss": 0.8979,
+      "step": 589
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019636591247480323,
+      "loss": 0.8469,
+      "step": 590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019634924900057866,
+      "loss": 0.9065,
+      "step": 591
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001963325481202583,
+      "loss": 0.3423,
+      "step": 592
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001963158098403259,
+      "loss": 0.8816,
+      "step": 593
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019629903416727986,
+      "loss": 0.8619,
+      "step": 594
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001962822211076331,
+      "loss": 0.8503,
+      "step": 595
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019626537066791297,
+      "loss": 0.8447,
+      "step": 596
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019624848285466147,
+      "loss": 0.8896,
+      "step": 597
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019623155767443498,
+      "loss": 0.8535,
+      "step": 598
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019621459513380444,
+      "loss": 0.8662,
+      "step": 599
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019619759523935532,
+      "loss": 0.8684,
+      "step": 600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019618055799768756,
+      "loss": 0.9297,
+      "step": 601
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019616348341541557,
+      "loss": 0.8193,
+      "step": 602
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019614637149916836,
+      "loss": 0.8699,
+      "step": 603
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019612922225558925,
+      "loss": 0.8719,
+      "step": 604
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019611203569133627,
+      "loss": 0.8955,
+      "step": 605
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019609481181308177,
+      "loss": 0.8882,
+      "step": 606
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019607755062751273,
+      "loss": 0.9277,
+      "step": 607
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019606025214133046,
+      "loss": 0.8838,
+      "step": 608
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019604291636125085,
+      "loss": 0.3339,
+      "step": 609
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019602554329400428,
+      "loss": 0.3107,
+      "step": 610
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019600813294633554,
+      "loss": 0.8386,
+      "step": 611
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019599068532500394,
+      "loss": 0.8208,
+      "step": 612
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019597320043678322,
+      "loss": 0.8516,
+      "step": 613
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019595567828846163,
+      "loss": 0.9041,
+      "step": 614
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001959381188868419,
+      "loss": 0.8672,
+      "step": 615
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019592052223874117,
+      "loss": 0.8799,
+      "step": 616
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019590288835099107,
+      "loss": 0.8711,
+      "step": 617
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019588521723043766,
+      "loss": 0.8872,
+      "step": 618
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001958675088839415,
+      "loss": 0.8708,
+      "step": 619
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019584976331837758,
+      "loss": 0.8865,
+      "step": 620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019583198054063533,
+      "loss": 0.8674,
+      "step": 621
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019581416055761865,
+      "loss": 0.896,
+      "step": 622
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019579630337624587,
+      "loss": 0.8794,
+      "step": 623
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019577840900344976,
+      "loss": 0.9524,
+      "step": 624
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019576047744617752,
+      "loss": 0.9121,
+      "step": 625
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001957425087113908,
+      "loss": 0.8872,
+      "step": 626
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001957245028060657,
+      "loss": 0.8026,
+      "step": 627
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001957064597371927,
+      "loss": 0.9185,
+      "step": 628
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019568837951177676,
+      "loss": 0.8564,
+      "step": 629
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019567026213683725,
+      "loss": 0.8691,
+      "step": 630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019565210761940798,
+      "loss": 0.885,
+      "step": 631
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019563391596653712,
+      "loss": 0.9006,
+      "step": 632
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001956156871852873,
+      "loss": 0.9097,
+      "step": 633
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019559742128273558,
+      "loss": 0.9075,
+      "step": 634
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019557911826597338,
+      "loss": 0.8894,
+      "step": 635
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001955607781421066,
+      "loss": 0.8484,
+      "step": 636
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019554240091825548,
+      "loss": 0.8652,
+      "step": 637
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001955239866015547,
+      "loss": 0.8694,
+      "step": 638
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019550553519915337,
+      "loss": 0.8037,
+      "step": 639
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019548704671821493,
+      "loss": 0.8599,
+      "step": 640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001954685211659172,
+      "loss": 0.8967,
+      "step": 641
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019544995854945248,
+      "loss": 0.8542,
+      "step": 642
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001954313588760274,
+      "loss": 0.896,
+      "step": 643
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019541272215286304,
+      "loss": 0.8867,
+      "step": 644
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019539404838719477,
+      "loss": 0.8203,
+      "step": 645
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001953753375862724,
+      "loss": 0.8501,
+      "step": 646
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001953565897573601,
+      "loss": 0.8301,
+      "step": 647
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019533780490773643,
+      "loss": 0.8965,
+      "step": 648
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019531898304469433,
+      "loss": 0.8472,
+      "step": 649
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001953001241755411,
+      "loss": 0.7828,
+      "step": 650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001952812283075984,
+      "loss": 0.8533,
+      "step": 651
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019526229544820217,
+      "loss": 0.8994,
+      "step": 652
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019524332560470292,
+      "loss": 0.8818,
+      "step": 653
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019522431878446535,
+      "loss": 0.9468,
+      "step": 654
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019520527499486856,
+      "loss": 0.8687,
+      "step": 655
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019518619424330598,
+      "loss": 0.9468,
+      "step": 656
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019516707653718544,
+      "loss": 0.3445,
+      "step": 657
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019514792188392912,
+      "loss": 0.8647,
+      "step": 658
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019512873029097345,
+      "loss": 0.825,
+      "step": 659
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019510950176576932,
+      "loss": 0.8733,
+      "step": 660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001950902363157819,
+      "loss": 0.8599,
+      "step": 661
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019507093394849066,
+      "loss": 0.8724,
+      "step": 662
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001950515946713895,
+      "loss": 0.821,
+      "step": 663
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019503221849198656,
+      "loss": 0.9004,
+      "step": 664
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019501280541780433,
+      "loss": 0.8464,
+      "step": 665
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019499335545637966,
+      "loss": 0.8735,
+      "step": 666
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001949738686152637,
+      "loss": 0.8718,
+      "step": 667
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019495434490202188,
+      "loss": 0.9221,
+      "step": 668
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000194934784324234,
+      "loss": 0.8342,
+      "step": 669
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019491518688949416,
+      "loss": 0.9189,
+      "step": 670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019489555260541073,
+      "loss": 0.8093,
+      "step": 671
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001948758814796064,
+      "loss": 0.8364,
+      "step": 672
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019485617351971827,
+      "loss": 0.9026,
+      "step": 673
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019483642873339751,
+      "loss": 0.3491,
+      "step": 674
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019481664712830984,
+      "loss": 0.8035,
+      "step": 675
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019479682871213514,
+      "loss": 0.8469,
+      "step": 676
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019477697349256756,
+      "loss": 0.3552,
+      "step": 677
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001947570814773156,
+      "loss": 0.8887,
+      "step": 678
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019473715267410207,
+      "loss": 0.854,
+      "step": 679
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019471718709066394,
+      "loss": 0.3251,
+      "step": 680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019469718473475256,
+      "loss": 0.9128,
+      "step": 681
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019467714561413358,
+      "loss": 0.9216,
+      "step": 682
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019465706973658683,
+      "loss": 0.918,
+      "step": 683
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019463695710990645,
+      "loss": 0.9199,
+      "step": 684
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019461680774190092,
+      "loss": 0.8918,
+      "step": 685
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019459662164039284,
+      "loss": 0.8757,
+      "step": 686
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019457639881321917,
+      "loss": 0.8926,
+      "step": 687
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019455613926823114,
+      "loss": 0.8824,
+      "step": 688
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001945358430132942,
+      "loss": 0.9067,
+      "step": 689
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000194515510056288,
+      "loss": 0.875,
+      "step": 690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019449514040510655,
+      "loss": 0.9199,
+      "step": 691
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019447473406765804,
+      "loss": 0.3142,
+      "step": 692
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019445429105186487,
+      "loss": 0.8701,
+      "step": 693
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001944338113656638,
+      "loss": 0.8606,
+      "step": 694
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019441329501700567,
+      "loss": 0.8633,
+      "step": 695
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019439274201385568,
+      "loss": 0.8582,
+      "step": 696
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001943721523641932,
+      "loss": 0.8606,
+      "step": 697
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019435152607601188,
+      "loss": 0.8806,
+      "step": 698
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001943308631573195,
+      "loss": 0.874,
+      "step": 699
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019431016361613814,
+      "loss": 0.8782,
+      "step": 700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019428942746050406,
+      "loss": 0.8723,
+      "step": 701
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019426865469846774,
+      "loss": 0.8926,
+      "step": 702
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019424784533809392,
+      "loss": 0.301,
+      "step": 703
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019422699938746152,
+      "loss": 0.8911,
+      "step": 704
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019420611685466358,
+      "loss": 0.8943,
+      "step": 705
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019418519774780748,
+      "loss": 0.9265,
+      "step": 706
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019416424207501474,
+      "loss": 0.8665,
+      "step": 707
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019414324984442103,
+      "loss": 0.8574,
+      "step": 708
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001941222210641763,
+      "loss": 0.8486,
+      "step": 709
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019410115574244462,
+      "loss": 0.7953,
+      "step": 710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019408005388740432,
+      "loss": 0.8788,
+      "step": 711
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019405891550724778,
+      "loss": 0.8668,
+      "step": 712
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019403774061018173,
+      "loss": 0.8916,
+      "step": 713
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019401652920442692,
+      "loss": 0.8938,
+      "step": 714
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019399528129821842,
+      "loss": 0.8743,
+      "step": 715
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001939739968998054,
+      "loss": 0.9658,
+      "step": 716
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001939526760174511,
+      "loss": 0.8777,
+      "step": 717
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001939313186594331,
+      "loss": 0.9163,
+      "step": 718
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019390992483404306,
+      "loss": 0.9253,
+      "step": 719
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001938884945495868,
+      "loss": 0.8262,
+      "step": 720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019386702781438425,
+      "loss": 0.8896,
+      "step": 721
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001938455246367696,
+      "loss": 0.8843,
+      "step": 722
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019382398502509107,
+      "loss": 0.9314,
+      "step": 723
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001938024089877111,
+      "loss": 0.8396,
+      "step": 724
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019378079653300625,
+      "loss": 0.8723,
+      "step": 725
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001937591476693672,
+      "loss": 0.8821,
+      "step": 726
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019373746240519883,
+      "loss": 0.8547,
+      "step": 727
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019371574074892006,
+      "loss": 0.9102,
+      "step": 728
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019369398270896403,
+      "loss": 0.8933,
+      "step": 729
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001936721882937779,
+      "loss": 0.8582,
+      "step": 730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001936503575118231,
+      "loss": 0.2961,
+      "step": 731
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019362849037157497,
+      "loss": 0.8706,
+      "step": 732
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001936065868815232,
+      "loss": 0.8848,
+      "step": 733
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019358464705017143,
+      "loss": 0.9031,
+      "step": 734
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019356267088603744,
+      "loss": 0.838,
+      "step": 735
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019354065839765315,
+      "loss": 0.8955,
+      "step": 736
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001935186095935646,
+      "loss": 0.8159,
+      "step": 737
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019349652448233187,
+      "loss": 0.8953,
+      "step": 738
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019347440307252911,
+      "loss": 0.8442,
+      "step": 739
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001934522453727447,
+      "loss": 0.8892,
+      "step": 740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019343005139158096,
+      "loss": 0.8689,
+      "step": 741
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019340782113765439,
+      "loss": 0.856,
+      "step": 742
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019338555461959552,
+      "loss": 0.8787,
+      "step": 743
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000193363251846049,
+      "loss": 0.8784,
+      "step": 744
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019334091282567352,
+      "loss": 0.8801,
+      "step": 745
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019331853756714184,
+      "loss": 0.8684,
+      "step": 746
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019329612607914086,
+      "loss": 0.9033,
+      "step": 747
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019327367837037142,
+      "loss": 0.8723,
+      "step": 748
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019325119444954855,
+      "loss": 0.3126,
+      "step": 749
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019322867432540125,
+      "loss": 0.2981,
+      "step": 750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019320611800667265,
+      "loss": 0.8894,
+      "step": 751
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019318352550211986,
+      "loss": 0.8948,
+      "step": 752
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019316089682051404,
+      "loss": 0.8817,
+      "step": 753
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019313823197064043,
+      "loss": 0.8129,
+      "step": 754
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019311553096129836,
+      "loss": 0.8826,
+      "step": 755
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001930927938013011,
+      "loss": 0.8748,
+      "step": 756
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000193070020499476,
+      "loss": 0.8887,
+      "step": 757
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019304721106466448,
+      "loss": 0.9619,
+      "step": 758
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019302436550572188,
+      "loss": 0.9275,
+      "step": 759
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019300148383151767,
+      "loss": 0.8779,
+      "step": 760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019297856605093533,
+      "loss": 0.9075,
+      "step": 761
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019295561217287226,
+      "loss": 0.8528,
+      "step": 762
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019293262220624002,
+      "loss": 0.8435,
+      "step": 763
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019290959615996407,
+      "loss": 0.9114,
+      "step": 764
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001928865340429839,
+      "loss": 0.8718,
+      "step": 765
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019286343586425307,
+      "loss": 0.8973,
+      "step": 766
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019284030163273905,
+      "loss": 0.8192,
+      "step": 767
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019281713135742333,
+      "loss": 0.8766,
+      "step": 768
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019279392504730145,
+      "loss": 0.8734,
+      "step": 769
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019277068271138288,
+      "loss": 0.8475,
+      "step": 770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019274740435869106,
+      "loss": 0.938,
+      "step": 771
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001927240899982635,
+      "loss": 0.8523,
+      "step": 772
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019270073963915162,
+      "loss": 0.8643,
+      "step": 773
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019267735329042086,
+      "loss": 0.887,
+      "step": 774
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019265393096115055,
+      "loss": 0.9155,
+      "step": 775
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019263047266043407,
+      "loss": 0.8655,
+      "step": 776
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019260697839737875,
+      "loss": 0.8381,
+      "step": 777
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019258344818110588,
+      "loss": 0.8882,
+      "step": 778
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019255988202075065,
+      "loss": 0.8262,
+      "step": 779
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001925362799254623,
+      "loss": 0.9011,
+      "step": 780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019251264190440397,
+      "loss": 0.8242,
+      "step": 781
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019248896796675275,
+      "loss": 0.8799,
+      "step": 782
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001924652581216997,
+      "loss": 0.8369,
+      "step": 783
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019244151237844974,
+      "loss": 0.885,
+      "step": 784
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019241773074622182,
+      "loss": 0.8689,
+      "step": 785
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019239391323424881,
+      "loss": 0.8521,
+      "step": 786
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019237005985177748,
+      "loss": 0.9075,
+      "step": 787
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019234617060806848,
+      "loss": 0.7673,
+      "step": 788
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001923222455123965,
+      "loss": 0.8029,
+      "step": 789
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019229828457405007,
+      "loss": 0.3447,
+      "step": 790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019227428780233165,
+      "loss": 0.896,
+      "step": 791
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019225025520655758,
+      "loss": 0.3647,
+      "step": 792
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001922261867960582,
+      "loss": 0.9297,
+      "step": 793
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019220208258017763,
+      "loss": 0.3006,
+      "step": 794
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019217794256827403,
+      "loss": 0.8767,
+      "step": 795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001921537667697193,
+      "loss": 0.8745,
+      "step": 796
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001921295551938994,
+      "loss": 0.8955,
+      "step": 797
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019210530785021405,
+      "loss": 0.8984,
+      "step": 798
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001920810247480769,
+      "loss": 0.9084,
+      "step": 799
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001920567058969155,
+      "loss": 0.8145,
+      "step": 800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001920323513061713,
+      "loss": 0.895,
+      "step": 801
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019200796098529956,
+      "loss": 0.8307,
+      "step": 802
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019198353494376938,
+      "loss": 0.8142,
+      "step": 803
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019195907319106392,
+      "loss": 0.833,
+      "step": 804
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019193457573667995,
+      "loss": 0.8848,
+      "step": 805
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001919100425901283,
+      "loss": 0.813,
+      "step": 806
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019188547376093354,
+      "loss": 0.7932,
+      "step": 807
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001918608692586342,
+      "loss": 0.9404,
+      "step": 808
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001918362290927825,
+      "loss": 0.8708,
+      "step": 809
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001918115532729447,
+      "loss": 0.848,
+      "step": 810
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019178684180870073,
+      "loss": 0.8762,
+      "step": 811
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019176209470964445,
+      "loss": 0.8423,
+      "step": 812
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019173731198538353,
+      "loss": 0.8867,
+      "step": 813
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019171249364553954,
+      "loss": 0.8196,
+      "step": 814
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019168763969974772,
+      "loss": 0.8721,
+      "step": 815
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001916627501576573,
+      "loss": 0.8857,
+      "step": 816
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001916378250289312,
+      "loss": 0.8474,
+      "step": 817
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019161286432324624,
+      "loss": 0.9119,
+      "step": 818
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019158786805029306,
+      "loss": 0.885,
+      "step": 819
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019156283621977603,
+      "loss": 0.8904,
+      "step": 820
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019153776884141336,
+      "loss": 0.8667,
+      "step": 821
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019151266592493712,
+      "loss": 0.8486,
+      "step": 822
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019148752748009306,
+      "loss": 0.8323,
+      "step": 823
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001914623535166408,
+      "loss": 0.8857,
+      "step": 824
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019143714404435381,
+      "loss": 0.9167,
+      "step": 825
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019141189907301921,
+      "loss": 0.8994,
+      "step": 826
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000191386618612438,
+      "loss": 0.9216,
+      "step": 827
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019136130267242488,
+      "loss": 0.8408,
+      "step": 828
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019133595126280846,
+      "loss": 0.8806,
+      "step": 829
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019131056439343093,
+      "loss": 0.8413,
+      "step": 830
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001912851420741484,
+      "loss": 0.8564,
+      "step": 831
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019125968431483065,
+      "loss": 0.8411,
+      "step": 832
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001912341911253613,
+      "loss": 0.8042,
+      "step": 833
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001912086625156377,
+      "loss": 0.8894,
+      "step": 834
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019118309849557087,
+      "loss": 0.9348,
+      "step": 835
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001911574990750857,
+      "loss": 0.8279,
+      "step": 836
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019113186426412073,
+      "loss": 0.8867,
+      "step": 837
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001911061940726283,
+      "loss": 0.8723,
+      "step": 838
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019108048851057446,
+      "loss": 0.3297,
+      "step": 839
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019105474758793895,
+      "loss": 0.8997,
+      "step": 840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019102897131471536,
+      "loss": 0.8672,
+      "step": 841
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019100315970091086,
+      "loss": 0.9012,
+      "step": 842
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019097731275654643,
+      "loss": 0.8467,
+      "step": 843
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019095143049165678,
+      "loss": 0.8691,
+      "step": 844
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019092551291629027,
+      "loss": 0.8557,
+      "step": 845
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019089956004050894,
+      "loss": 0.8823,
+      "step": 846
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019087357187438868,
+      "loss": 0.3196,
+      "step": 847
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001908475484280189,
+      "loss": 0.3181,
+      "step": 848
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001908214897115029,
+      "loss": 0.9148,
+      "step": 849
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019079539573495748,
+      "loss": 0.8867,
+      "step": 850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001907692665085133,
+      "loss": 0.8398,
+      "step": 851
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019074310204231457,
+      "loss": 0.9031,
+      "step": 852
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001907169023465192,
+      "loss": 0.8928,
+      "step": 853
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019069066743129893,
+      "loss": 0.9058,
+      "step": 854
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000190664397306839,
+      "loss": 0.7695,
+      "step": 855
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019063809198333832,
+      "loss": 0.8542,
+      "step": 856
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001906117514710096,
+      "loss": 0.8391,
+      "step": 857
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019058537578007907,
+      "loss": 0.8674,
+      "step": 858
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019055896492078676,
+      "loss": 0.9148,
+      "step": 859
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001905325189033862,
+      "loss": 0.8706,
+      "step": 860
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019050603773814468,
+      "loss": 0.8462,
+      "step": 861
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001904795214353431,
+      "loss": 0.8792,
+      "step": 862
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000190452970005276,
+      "loss": 0.9458,
+      "step": 863
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019042638345825156,
+      "loss": 0.8374,
+      "step": 864
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019039976180459158,
+      "loss": 0.9075,
+      "step": 865
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019037310505463153,
+      "loss": 0.873,
+      "step": 866
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019034641321872045,
+      "loss": 0.2896,
+      "step": 867
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019031968630722103,
+      "loss": 0.886,
+      "step": 868
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001902929243305096,
+      "loss": 0.8884,
+      "step": 869
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019026612729897607,
+      "loss": 0.8555,
+      "step": 870
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019023929522302396,
+      "loss": 0.8552,
+      "step": 871
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019021242811307043,
+      "loss": 0.8431,
+      "step": 872
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001901855259795462,
+      "loss": 0.8264,
+      "step": 873
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019015858883289556,
+      "loss": 0.7719,
+      "step": 874
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019013161668357653,
+      "loss": 0.8505,
+      "step": 875
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019010460954206056,
+      "loss": 0.8462,
+      "step": 876
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019007756741883282,
+      "loss": 0.8289,
+      "step": 877
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001900504903243919,
+      "loss": 0.8572,
+      "step": 878
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019002337826925012,
+      "loss": 0.8713,
+      "step": 879
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001899962312639333,
+      "loss": 0.8721,
+      "step": 880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018996904931898082,
+      "loss": 0.8557,
+      "step": 881
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001899418324449457,
+      "loss": 0.3129,
+      "step": 882
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018991458065239445,
+      "loss": 0.8203,
+      "step": 883
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001898872939519071,
+      "loss": 0.884,
+      "step": 884
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018985997235407734,
+      "loss": 0.8923,
+      "step": 885
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018983261586951237,
+      "loss": 0.8801,
+      "step": 886
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018980522450883286,
+      "loss": 0.9131,
+      "step": 887
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018977779828267311,
+      "loss": 0.8335,
+      "step": 888
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018975033720168094,
+      "loss": 0.9009,
+      "step": 889
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018972284127651768,
+      "loss": 0.8179,
+      "step": 890
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018969531051785818,
+      "loss": 0.8572,
+      "step": 891
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018966774493639084,
+      "loss": 0.8733,
+      "step": 892
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018964014454281758,
+      "loss": 0.8884,
+      "step": 893
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018961250934785377,
+      "loss": 0.937,
+      "step": 894
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018958483936222842,
+      "loss": 0.7991,
+      "step": 895
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001895571345966839,
+      "loss": 0.8396,
+      "step": 896
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001895293950619762,
+      "loss": 0.8354,
+      "step": 897
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018950162076887477,
+      "loss": 0.3253,
+      "step": 898
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018947381172816248,
+      "loss": 0.8453,
+      "step": 899
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018944596795063582,
+      "loss": 0.8447,
+      "step": 900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001894180894471047,
+      "loss": 0.8718,
+      "step": 901
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018939017622839253,
+      "loss": 0.8748,
+      "step": 902
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001893622283053361,
+      "loss": 0.8657,
+      "step": 903
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018933424568878584,
+      "loss": 0.8508,
+      "step": 904
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018930622838960555,
+      "loss": 0.8521,
+      "step": 905
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018927817641867245,
+      "loss": 0.9143,
+      "step": 906
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018925008978687736,
+      "loss": 0.8809,
+      "step": 907
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018922196850512446,
+      "loss": 0.823,
+      "step": 908
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018919381258433133,
+      "loss": 0.3397,
+      "step": 909
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018916562203542917,
+      "loss": 0.9004,
+      "step": 910
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018913739686936246,
+      "loss": 0.8792,
+      "step": 911
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018910913709708918,
+      "loss": 0.8668,
+      "step": 912
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018908084272958078,
+      "loss": 0.9001,
+      "step": 913
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018905251377782206,
+      "loss": 0.8523,
+      "step": 914
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018902415025281134,
+      "loss": 0.8384,
+      "step": 915
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018899575216556032,
+      "loss": 0.9016,
+      "step": 916
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018896731952709408,
+      "loss": 0.8569,
+      "step": 917
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018893885234845116,
+      "loss": 0.8358,
+      "step": 918
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001889103506406835,
+      "loss": 0.8884,
+      "step": 919
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018888181441485647,
+      "loss": 0.8569,
+      "step": 920
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001888532436820488,
+      "loss": 0.8696,
+      "step": 921
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001888246384533526,
+      "loss": 0.7944,
+      "step": 922
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018879599873987345,
+      "loss": 0.814,
+      "step": 923
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018876732455273024,
+      "loss": 0.8318,
+      "step": 924
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018873861590305526,
+      "loss": 0.8335,
+      "step": 925
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018870987280199426,
+      "loss": 0.8706,
+      "step": 926
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018868109526070628,
+      "loss": 0.347,
+      "step": 927
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018865228329036372,
+      "loss": 0.8943,
+      "step": 928
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001886234369021524,
+      "loss": 0.8378,
+      "step": 929
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001885945561072715,
+      "loss": 0.9043,
+      "step": 930
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001885656409169335,
+      "loss": 0.8595,
+      "step": 931
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001885366913423643,
+      "loss": 0.8389,
+      "step": 932
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001885077073948031,
+      "loss": 0.8916,
+      "step": 933
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001884786890855025,
+      "loss": 0.907,
+      "step": 934
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018844963642572837,
+      "loss": 0.8076,
+      "step": 935
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018842054942675996,
+      "loss": 0.8345,
+      "step": 936
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018839142809988987,
+      "loss": 0.8772,
+      "step": 937
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018836227245642398,
+      "loss": 0.8004,
+      "step": 938
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018833308250768154,
+      "loss": 0.7991,
+      "step": 939
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018830385826499506,
+      "loss": 0.8479,
+      "step": 940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001882745997397104,
+      "loss": 0.8408,
+      "step": 941
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018824530694318674,
+      "loss": 0.8533,
+      "step": 942
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018821597988679658,
+      "loss": 0.8555,
+      "step": 943
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001881866185819256,
+      "loss": 0.8657,
+      "step": 944
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000188157223039973,
+      "loss": 0.9204,
+      "step": 945
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018812779327235103,
+      "loss": 0.8745,
+      "step": 946
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001880983292904854,
+      "loss": 0.3369,
+      "step": 947
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000188068831105815,
+      "loss": 0.8706,
+      "step": 948
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018803929872979213,
+      "loss": 0.8667,
+      "step": 949
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018800973217388215,
+      "loss": 0.8818,
+      "step": 950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001879801314495639,
+      "loss": 0.8633,
+      "step": 951
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018795049656832936,
+      "loss": 0.8735,
+      "step": 952
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018792082754168386,
+      "loss": 0.8992,
+      "step": 953
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018789112438114588,
+      "loss": 0.8132,
+      "step": 954
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018786138709824725,
+      "loss": 0.8047,
+      "step": 955
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018783161570453297,
+      "loss": 0.7689,
+      "step": 956
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001878018102115614,
+      "loss": 0.8376,
+      "step": 957
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018777197063090393,
+      "loss": 0.8518,
+      "step": 958
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001877420969741454,
+      "loss": 0.8362,
+      "step": 959
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001877121892528838,
+      "loss": 0.8425,
+      "step": 960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001876822474787303,
+      "loss": 0.8638,
+      "step": 961
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018765227166330935,
+      "loss": 0.855,
+      "step": 962
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018762226181825856,
+      "loss": 0.8633,
+      "step": 963
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001875922179552288,
+      "loss": 0.8606,
+      "step": 964
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018756214008588417,
+      "loss": 0.9187,
+      "step": 965
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001875320282219019,
+      "loss": 0.8906,
+      "step": 966
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018750188237497245,
+      "loss": 0.8562,
+      "step": 967
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001874717025567995,
+      "loss": 0.8645,
+      "step": 968
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018744148877909987,
+      "loss": 0.8054,
+      "step": 969
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001874112410536036,
+      "loss": 0.9092,
+      "step": 970
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001873809593920539,
+      "loss": 0.9153,
+      "step": 971
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018735064380620716,
+      "loss": 0.8057,
+      "step": 972
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001873202943078329,
+      "loss": 0.8997,
+      "step": 973
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018728991090871387,
+      "loss": 0.3295,
+      "step": 974
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018725949362064594,
+      "loss": 0.8445,
+      "step": 975
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018722904245543816,
+      "loss": 0.8547,
+      "step": 976
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001871985574249127,
+      "loss": 0.9456,
+      "step": 977
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018716803854090493,
+      "loss": 0.8616,
+      "step": 978
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018713748581526332,
+      "loss": 0.8311,
+      "step": 979
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018710689925984948,
+      "loss": 0.8268,
+      "step": 980
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018707627888653815,
+      "loss": 0.8362,
+      "step": 981
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018704562470721727,
+      "loss": 0.8364,
+      "step": 982
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018701493673378778,
+      "loss": 0.8372,
+      "step": 983
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018698421497816386,
+      "loss": 0.8398,
+      "step": 984
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001869534594522727,
+      "loss": 0.8691,
+      "step": 985
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018692267016805471,
+      "loss": 0.8596,
+      "step": 986
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018689184713746332,
+      "loss": 0.9143,
+      "step": 987
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001868609903724651,
+      "loss": 0.9194,
+      "step": 988
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001868300998850397,
+      "loss": 0.8386,
+      "step": 989
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001867991756871799,
+      "loss": 0.8752,
+      "step": 990
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001867682177908915,
+      "loss": 0.8582,
+      "step": 991
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001867372262081934,
+      "loss": 0.9255,
+      "step": 992
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018670620095111767,
+      "loss": 0.8196,
+      "step": 993
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018667514203170936,
+      "loss": 0.8721,
+      "step": 994
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018664404946202657,
+      "loss": 0.7776,
+      "step": 995
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018661292325414057,
+      "loss": 0.9241,
+      "step": 996
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018658176342013557,
+      "loss": 0.8298,
+      "step": 997
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001865505699721089,
+      "loss": 0.8745,
+      "step": 998
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018651934292217097,
+      "loss": 0.8303,
+      "step": 999
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018648808228244516,
+      "loss": 0.8281,
+      "step": 1000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018645678806506795,
+      "loss": 0.8479,
+      "step": 1001
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001864254602821888,
+      "loss": 0.8354,
+      "step": 1002
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018639409894597025,
+      "loss": 0.3209,
+      "step": 1003
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018636270406858786,
+      "loss": 0.8975,
+      "step": 1004
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018633127566223021,
+      "loss": 0.8076,
+      "step": 1005
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018629981373909885,
+      "loss": 0.8794,
+      "step": 1006
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018626831831140844,
+      "loss": 0.9111,
+      "step": 1007
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001862367893913865,
+      "loss": 0.8625,
+      "step": 1008
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018620522699127373,
+      "loss": 0.8633,
+      "step": 1009
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018617363112332375,
+      "loss": 0.9016,
+      "step": 1010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018614200179980307,
+      "loss": 0.8506,
+      "step": 1011
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018611033903299134,
+      "loss": 0.8818,
+      "step": 1012
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018607864283518114,
+      "loss": 0.8794,
+      "step": 1013
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018604691321867805,
+      "loss": 0.3715,
+      "step": 1014
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018601515019580053,
+      "loss": 0.8038,
+      "step": 1015
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018598335377888013,
+      "loss": 0.9119,
+      "step": 1016
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001859515239802613,
+      "loss": 0.8792,
+      "step": 1017
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018591966081230144,
+      "loss": 0.9065,
+      "step": 1018
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018588776428737097,
+      "loss": 0.8535,
+      "step": 1019
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001858558344178532,
+      "loss": 0.8708,
+      "step": 1020
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018582387121614437,
+      "loss": 0.8794,
+      "step": 1021
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018579187469465377,
+      "loss": 0.9058,
+      "step": 1022
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001857598448658035,
+      "loss": 0.8562,
+      "step": 1023
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001857277817420287,
+      "loss": 0.89,
+      "step": 1024
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018569568533577727,
+      "loss": 0.8718,
+      "step": 1025
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018566355565951024,
+      "loss": 0.936,
+      "step": 1026
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001856313927257014,
+      "loss": 0.8317,
+      "step": 1027
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018559919654683756,
+      "loss": 0.8573,
+      "step": 1028
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018556696713541832,
+      "loss": 0.9429,
+      "step": 1029
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001855347045039563,
+      "loss": 0.8875,
+      "step": 1030
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018550240866497695,
+      "loss": 0.8992,
+      "step": 1031
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001854700796310186,
+      "loss": 0.8721,
+      "step": 1032
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018543771741463255,
+      "loss": 0.9055,
+      "step": 1033
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018540532202838286,
+      "loss": 0.9172,
+      "step": 1034
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018537289348484657,
+      "loss": 0.9241,
+      "step": 1035
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018534043179661356,
+      "loss": 0.8596,
+      "step": 1036
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018530793697628658,
+      "loss": 0.8687,
+      "step": 1037
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018527540903648122,
+      "loss": 0.823,
+      "step": 1038
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018524284798982594,
+      "loss": 0.8982,
+      "step": 1039
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018521025384896208,
+      "loss": 0.8752,
+      "step": 1040
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001851776266265438,
+      "loss": 0.8601,
+      "step": 1041
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018514496633523813,
+      "loss": 0.9609,
+      "step": 1042
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001851122729877249,
+      "loss": 0.8867,
+      "step": 1043
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018507954659669677,
+      "loss": 0.8586,
+      "step": 1044
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001850467871748593,
+      "loss": 0.8311,
+      "step": 1045
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001850139947349308,
+      "loss": 0.8186,
+      "step": 1046
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001849811692896424,
+      "loss": 0.9144,
+      "step": 1047
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001849483108517381,
+      "loss": 0.3151,
+      "step": 1048
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001849154194339747,
+      "loss": 0.7812,
+      "step": 1049
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018488249504912174,
+      "loss": 0.8494,
+      "step": 1050
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018484953770996163,
+      "loss": 0.855,
+      "step": 1051
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018481654742928948,
+      "loss": 0.9014,
+      "step": 1052
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018478352421991335,
+      "loss": 0.9075,
+      "step": 1053
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001847504680946539,
+      "loss": 0.8296,
+      "step": 1054
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001847173790663447,
+      "loss": 0.3206,
+      "step": 1055
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018468425714783204,
+      "loss": 0.8943,
+      "step": 1056
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000184651102351975,
+      "loss": 0.7976,
+      "step": 1057
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001846179146916454,
+      "loss": 0.8799,
+      "step": 1058
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018458469417972784,
+      "loss": 0.8866,
+      "step": 1059
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018455144082911966,
+      "loss": 0.8457,
+      "step": 1060
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018451815465273094,
+      "loss": 0.8799,
+      "step": 1061
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018448483566348457,
+      "loss": 0.8894,
+      "step": 1062
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018445148387431605,
+      "loss": 0.9099,
+      "step": 1063
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001844180992981738,
+      "loss": 0.3201,
+      "step": 1064
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018438468194801875,
+      "loss": 0.3289,
+      "step": 1065
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018435123183682475,
+      "loss": 0.8718,
+      "step": 1066
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018431774897757826,
+      "loss": 0.8977,
+      "step": 1067
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018428423338327844,
+      "loss": 0.8416,
+      "step": 1068
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018425068506693727,
+      "loss": 0.8501,
+      "step": 1069
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018421710404157928,
+      "loss": 0.8555,
+      "step": 1070
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018418349032024183,
+      "loss": 0.8557,
+      "step": 1071
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001841498439159749,
+      "loss": 0.8564,
+      "step": 1072
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018411616484184125,
+      "loss": 0.7893,
+      "step": 1073
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018408245311091616,
+      "loss": 0.8918,
+      "step": 1074
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018404870873628775,
+      "loss": 0.9067,
+      "step": 1075
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018401493173105673,
+      "loss": 0.8794,
+      "step": 1076
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001839811221083365,
+      "loss": 0.8098,
+      "step": 1077
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018394727988125308,
+      "loss": 0.9221,
+      "step": 1078
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018391340506294522,
+      "loss": 0.8242,
+      "step": 1079
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018387949766656434,
+      "loss": 0.8936,
+      "step": 1080
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018384555770527435,
+      "loss": 0.8477,
+      "step": 1081
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018381158519225203,
+      "loss": 0.8994,
+      "step": 1082
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001837775801406866,
+      "loss": 0.8796,
+      "step": 1083
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018374354256378,
+      "loss": 0.8452,
+      "step": 1084
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001837094724747468,
+      "loss": 0.8345,
+      "step": 1085
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001836753698868142,
+      "loss": 0.8503,
+      "step": 1086
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018364123481322197,
+      "loss": 0.8376,
+      "step": 1087
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018360706726722252,
+      "loss": 0.8877,
+      "step": 1088
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001835728672620809,
+      "loss": 0.8682,
+      "step": 1089
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001835386348110747,
+      "loss": 0.7688,
+      "step": 1090
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018350436992749418,
+      "loss": 0.8665,
+      "step": 1091
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018347007262464206,
+      "loss": 0.8416,
+      "step": 1092
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018343574291583383,
+      "loss": 0.8175,
+      "step": 1093
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001834013808143974,
+      "loss": 0.886,
+      "step": 1094
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018336698633367337,
+      "loss": 0.8467,
+      "step": 1095
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018333255948701482,
+      "loss": 0.835,
+      "step": 1096
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018329810028778747,
+      "loss": 0.8186,
+      "step": 1097
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001832636087493695,
+      "loss": 0.8267,
+      "step": 1098
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001832290848851518,
+      "loss": 0.8628,
+      "step": 1099
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001831945287085377,
+      "loss": 0.9224,
+      "step": 1100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018315994023294306,
+      "loss": 0.8579,
+      "step": 1101
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018312531947179634,
+      "loss": 0.9265,
+      "step": 1102
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018309066643853853,
+      "loss": 0.8613,
+      "step": 1103
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018305598114662312,
+      "loss": 0.8389,
+      "step": 1104
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018302126360951611,
+      "loss": 0.8784,
+      "step": 1105
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018298651384069604,
+      "loss": 0.8618,
+      "step": 1106
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018295173185365403,
+      "loss": 0.8997,
+      "step": 1107
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018291691766189358,
+      "loss": 0.8511,
+      "step": 1108
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001828820712789308,
+      "loss": 0.9185,
+      "step": 1109
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001828471927182942,
+      "loss": 0.8589,
+      "step": 1110
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001828122819935249,
+      "loss": 0.793,
+      "step": 1111
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001827773391181764,
+      "loss": 0.8462,
+      "step": 1112
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001827423641058148,
+      "loss": 0.3491,
+      "step": 1113
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018270735697001848,
+      "loss": 0.8774,
+      "step": 1114
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018267231772437853,
+      "loss": 0.8586,
+      "step": 1115
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018263724638249834,
+      "loss": 0.8506,
+      "step": 1116
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018260214295799383,
+      "loss": 0.3105,
+      "step": 1117
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001825670074644933,
+      "loss": 0.3307,
+      "step": 1118
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018253183991563768,
+      "loss": 0.8256,
+      "step": 1119
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001824966403250801,
+      "loss": 0.3347,
+      "step": 1120
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018246140870648633,
+      "loss": 0.8303,
+      "step": 1121
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018242614507353445,
+      "loss": 0.9053,
+      "step": 1122
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018239084943991505,
+      "loss": 0.8718,
+      "step": 1123
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018235552181933108,
+      "loss": 0.7654,
+      "step": 1124
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018232016222549796,
+      "loss": 0.8114,
+      "step": 1125
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001822847706721435,
+      "loss": 0.8784,
+      "step": 1126
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018224934717300792,
+      "loss": 0.8311,
+      "step": 1127
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018221389174184386,
+      "loss": 0.8525,
+      "step": 1128
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018217840439241633,
+      "loss": 0.7943,
+      "step": 1129
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018214288513850266,
+      "loss": 0.8892,
+      "step": 1130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018210733399389277,
+      "loss": 0.3983,
+      "step": 1131
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018207175097238878,
+      "loss": 0.8591,
+      "step": 1132
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018203613608780523,
+      "loss": 0.3376,
+      "step": 1133
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018200048935396908,
+      "loss": 0.9041,
+      "step": 1134
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018196481078471962,
+      "loss": 0.788,
+      "step": 1135
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018192910039390844,
+      "loss": 0.8749,
+      "step": 1136
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001818933581953996,
+      "loss": 0.8601,
+      "step": 1137
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018185758420306947,
+      "loss": 0.8289,
+      "step": 1138
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001818217784308067,
+      "loss": 0.8206,
+      "step": 1139
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001817859408925123,
+      "loss": 0.8928,
+      "step": 1140
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018175007160209968,
+      "loss": 0.865,
+      "step": 1141
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018171417057349457,
+      "loss": 0.8596,
+      "step": 1142
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018167823782063488,
+      "loss": 0.8422,
+      "step": 1143
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018164227335747105,
+      "loss": 0.33,
+      "step": 1144
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018160627719796568,
+      "loss": 0.8085,
+      "step": 1145
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001815702493560937,
+      "loss": 0.8889,
+      "step": 1146
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018153418984584238,
+      "loss": 0.8818,
+      "step": 1147
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018149809868121125,
+      "loss": 0.8689,
+      "step": 1148
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018146197587621217,
+      "loss": 0.8357,
+      "step": 1149
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001814258214448692,
+      "loss": 0.3439,
+      "step": 1150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001813896354012188,
+      "loss": 0.8875,
+      "step": 1151
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001813534177593096,
+      "loss": 0.7953,
+      "step": 1152
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018131716853320254,
+      "loss": 0.3434,
+      "step": 1153
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018128088773697085,
+      "loss": 0.9155,
+      "step": 1154
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018124457538469996,
+      "loss": 0.8882,
+      "step": 1155
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018120823149048754,
+      "loss": 0.8606,
+      "step": 1156
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018117185606844358,
+      "loss": 0.8513,
+      "step": 1157
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018113544913269026,
+      "loss": 0.9155,
+      "step": 1158
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018109901069736202,
+      "loss": 0.821,
+      "step": 1159
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018106254077660552,
+      "loss": 0.8242,
+      "step": 1160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001810260393845796,
+      "loss": 0.9084,
+      "step": 1161
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001809895065354554,
+      "loss": 0.3105,
+      "step": 1162
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018095294224341622,
+      "loss": 0.3075,
+      "step": 1163
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018091634652265753,
+      "loss": 0.9058,
+      "step": 1164
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018087971938738714,
+      "loss": 0.8657,
+      "step": 1165
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018084306085182488,
+      "loss": 0.8616,
+      "step": 1166
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018080637093020288,
+      "loss": 0.8882,
+      "step": 1167
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001807696496367655,
+      "loss": 0.8235,
+      "step": 1168
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018073289698576913,
+      "loss": 0.8774,
+      "step": 1169
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018069611299148237,
+      "loss": 0.3415,
+      "step": 1170
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018065929766818617,
+      "loss": 0.8855,
+      "step": 1171
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001806224510301734,
+      "loss": 0.9326,
+      "step": 1172
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018058557309174925,
+      "loss": 0.8169,
+      "step": 1173
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018054866386723096,
+      "loss": 0.8008,
+      "step": 1174
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000180511723370948,
+      "loss": 0.8074,
+      "step": 1175
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001804747516172419,
+      "loss": 0.3314,
+      "step": 1176
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018043774862046643,
+      "loss": 0.811,
+      "step": 1177
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001804007143949874,
+      "loss": 0.8547,
+      "step": 1178
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018036364895518273,
+      "loss": 0.863,
+      "step": 1179
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018032655231544253,
+      "loss": 0.8862,
+      "step": 1180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018028942449016904,
+      "loss": 0.8169,
+      "step": 1181
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018025226549377648,
+      "loss": 0.8728,
+      "step": 1182
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001802150753406913,
+      "loss": 0.8911,
+      "step": 1183
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000180177854045352,
+      "loss": 0.8467,
+      "step": 1184
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018014060162220913,
+      "loss": 0.3282,
+      "step": 1185
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001801033180857254,
+      "loss": 0.8428,
+      "step": 1186
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018006600345037556,
+      "loss": 0.8997,
+      "step": 1187
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018002865773064644,
+      "loss": 0.8059,
+      "step": 1188
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017999128094103688,
+      "loss": 0.8828,
+      "step": 1189
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017995387309605788,
+      "loss": 0.8293,
+      "step": 1190
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001799164342102325,
+      "loss": 0.8821,
+      "step": 1191
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001798789642980957,
+      "loss": 0.8406,
+      "step": 1192
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017984146337419467,
+      "loss": 0.822,
+      "step": 1193
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017980393145308857,
+      "loss": 0.8447,
+      "step": 1194
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017976636854934852,
+      "loss": 0.8779,
+      "step": 1195
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017972877467755776,
+      "loss": 0.8411,
+      "step": 1196
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017969114985231152,
+      "loss": 0.8499,
+      "step": 1197
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017965349408821709,
+      "loss": 0.8146,
+      "step": 1198
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017961580739989365,
+      "loss": 0.835,
+      "step": 1199
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001795780898019726,
+      "loss": 0.8562,
+      "step": 1200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017954034130909707,
+      "loss": 0.8594,
+      "step": 1201
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017950256193592244,
+      "loss": 0.7659,
+      "step": 1202
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017946475169711588,
+      "loss": 0.3369,
+      "step": 1203
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017942691060735666,
+      "loss": 0.8447,
+      "step": 1204
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017938903868133598,
+      "loss": 0.8239,
+      "step": 1205
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017935113593375707,
+      "loss": 0.897,
+      "step": 1206
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017931320237933503,
+      "loss": 0.8879,
+      "step": 1207
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000179275238032797,
+      "loss": 0.8776,
+      "step": 1208
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017923724290888206,
+      "loss": 0.8574,
+      "step": 1209
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017919921702234118,
+      "loss": 0.886,
+      "step": 1210
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001791611603879374,
+      "loss": 0.8604,
+      "step": 1211
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017912307302044548,
+      "loss": 0.8376,
+      "step": 1212
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017908495493465236,
+      "loss": 0.9238,
+      "step": 1213
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017904680614535675,
+      "loss": 0.8198,
+      "step": 1214
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017900862666736934,
+      "loss": 0.8032,
+      "step": 1215
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001789704165155127,
+      "loss": 0.8267,
+      "step": 1216
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017893217570462134,
+      "loss": 0.3302,
+      "step": 1217
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017889390424954166,
+      "loss": 0.9395,
+      "step": 1218
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017885560216513197,
+      "loss": 0.8069,
+      "step": 1219
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017881726946626243,
+      "loss": 0.8162,
+      "step": 1220
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017877890616781512,
+      "loss": 0.8711,
+      "step": 1221
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000178740512284684,
+      "loss": 0.8259,
+      "step": 1222
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001787020878317749,
+      "loss": 0.8374,
+      "step": 1223
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017866363282400554,
+      "loss": 0.8169,
+      "step": 1224
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017862514727630543,
+      "loss": 0.8367,
+      "step": 1225
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017858663120361596,
+      "loss": 0.3323,
+      "step": 1226
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017854808462089048,
+      "loss": 0.8027,
+      "step": 1227
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017850950754309407,
+      "loss": 0.3328,
+      "step": 1228
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017847089998520365,
+      "loss": 0.8408,
+      "step": 1229
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017843226196220802,
+      "loss": 0.9275,
+      "step": 1230
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001783935934891078,
+      "loss": 0.9175,
+      "step": 1231
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017835489458091543,
+      "loss": 0.8357,
+      "step": 1232
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017831616525265513,
+      "loss": 0.8488,
+      "step": 1233
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017827740551936295,
+      "loss": 0.8442,
+      "step": 1234
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017823861539608684,
+      "loss": 0.8409,
+      "step": 1235
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017819979489788637,
+      "loss": 0.8533,
+      "step": 1236
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017816094403983298,
+      "loss": 0.7969,
+      "step": 1237
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017812206283701002,
+      "loss": 0.845,
+      "step": 1238
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017808315130451243,
+      "loss": 0.8785,
+      "step": 1239
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000178044209457447,
+      "loss": 0.8677,
+      "step": 1240
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017800523731093234,
+      "loss": 0.8738,
+      "step": 1241
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017796623488009874,
+      "loss": 0.8003,
+      "step": 1242
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017792720218008827,
+      "loss": 0.3282,
+      "step": 1243
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017788813922605488,
+      "loss": 0.8159,
+      "step": 1244
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000177849046033164,
+      "loss": 0.7977,
+      "step": 1245
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017780992261659306,
+      "loss": 0.8672,
+      "step": 1246
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017777076899153112,
+      "loss": 0.865,
+      "step": 1247
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017773158517317888,
+      "loss": 0.8291,
+      "step": 1248
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017769237117674892,
+      "loss": 0.8362,
+      "step": 1249
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017765312701746545,
+      "loss": 0.8931,
+      "step": 1250
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017761385271056437,
+      "loss": 0.8223,
+      "step": 1251
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001775745482712934,
+      "loss": 0.8505,
+      "step": 1252
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017753521371491175,
+      "loss": 0.9131,
+      "step": 1253
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017749584905669058,
+      "loss": 0.8071,
+      "step": 1254
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017745645431191252,
+      "loss": 0.8271,
+      "step": 1255
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017741702949587194,
+      "loss": 0.31,
+      "step": 1256
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017737757462387506,
+      "loss": 0.8391,
+      "step": 1257
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017733808971123947,
+      "loss": 0.8562,
+      "step": 1258
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017729857477329463,
+      "loss": 0.8247,
+      "step": 1259
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017725902982538162,
+      "loss": 0.8499,
+      "step": 1260
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001772194548828531,
+      "loss": 0.8503,
+      "step": 1261
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017717984996107344,
+      "loss": 0.8752,
+      "step": 1262
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017714021507541866,
+      "loss": 0.8743,
+      "step": 1263
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017710055024127637,
+      "loss": 0.8495,
+      "step": 1264
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017706085547404584,
+      "loss": 0.8582,
+      "step": 1265
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001770211307891379,
+      "loss": 0.8535,
+      "step": 1266
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001769813762019751,
+      "loss": 0.8639,
+      "step": 1267
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017694159172799152,
+      "loss": 0.9346,
+      "step": 1268
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017690177738263283,
+      "loss": 0.8313,
+      "step": 1269
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017686193318135635,
+      "loss": 0.7771,
+      "step": 1270
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017682205913963103,
+      "loss": 0.8879,
+      "step": 1271
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017678215527293722,
+      "loss": 0.3329,
+      "step": 1272
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017674222159676708,
+      "loss": 0.8623,
+      "step": 1273
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001767022581266242,
+      "loss": 0.8401,
+      "step": 1274
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001766622648780238,
+      "loss": 0.8296,
+      "step": 1275
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017662224186649258,
+      "loss": 0.8643,
+      "step": 1276
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017658218910756893,
+      "loss": 0.8352,
+      "step": 1277
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017654210661680262,
+      "loss": 0.7949,
+      "step": 1278
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001765019944097551,
+      "loss": 0.8345,
+      "step": 1279
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017646185250199935,
+      "loss": 0.835,
+      "step": 1280
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017642168090911973,
+      "loss": 0.8232,
+      "step": 1281
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017638147964671238,
+      "loss": 0.823,
+      "step": 1282
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001763412487303847,
+      "loss": 0.8796,
+      "step": 1283
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017630098817575577,
+      "loss": 0.8572,
+      "step": 1284
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001762606979984561,
+      "loss": 0.8401,
+      "step": 1285
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017622037821412776,
+      "loss": 0.8672,
+      "step": 1286
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017618002883842423,
+      "loss": 0.8586,
+      "step": 1287
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017613964988701057,
+      "loss": 0.8696,
+      "step": 1288
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017609924137556324,
+      "loss": 0.8578,
+      "step": 1289
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017605880331977023,
+      "loss": 0.3367,
+      "step": 1290
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000176018335735331,
+      "loss": 0.8315,
+      "step": 1291
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017597783863795643,
+      "loss": 0.875,
+      "step": 1292
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017593731204336893,
+      "loss": 0.8667,
+      "step": 1293
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001758967559673023,
+      "loss": 0.9111,
+      "step": 1294
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001758561704255018,
+      "loss": 0.342,
+      "step": 1295
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017581555543372412,
+      "loss": 0.8447,
+      "step": 1296
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017577491100773744,
+      "loss": 0.9001,
+      "step": 1297
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017573423716332127,
+      "loss": 0.9087,
+      "step": 1298
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017569353391626663,
+      "loss": 0.3333,
+      "step": 1299
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017565280128237595,
+      "loss": 0.885,
+      "step": 1300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000175612039277463,
+      "loss": 0.8151,
+      "step": 1301
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017557124791735298,
+      "loss": 0.3467,
+      "step": 1302
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017553042721788254,
+      "loss": 0.8628,
+      "step": 1303
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017548957719489968,
+      "loss": 0.8118,
+      "step": 1304
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017544869786426372,
+      "loss": 0.8882,
+      "step": 1305
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017540778924184553,
+      "loss": 0.8208,
+      "step": 1306
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017536685134352716,
+      "loss": 0.8738,
+      "step": 1307
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017532588418520215,
+      "loss": 0.82,
+      "step": 1308
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017528488778277535,
+      "loss": 0.3329,
+      "step": 1309
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000175243862152163,
+      "loss": 0.9182,
+      "step": 1310
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001752028073092926,
+      "loss": 0.8298,
+      "step": 1311
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001751617232701031,
+      "loss": 0.8558,
+      "step": 1312
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001751206100505448,
+      "loss": 0.8447,
+      "step": 1313
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017507946766657912,
+      "loss": 0.7933,
+      "step": 1314
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017503829613417905,
+      "loss": 0.8342,
+      "step": 1315
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001749970954693288,
+      "loss": 0.8503,
+      "step": 1316
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017495586568802384,
+      "loss": 0.8525,
+      "step": 1317
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017491460680627103,
+      "loss": 0.885,
+      "step": 1318
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017487331884008845,
+      "loss": 0.8372,
+      "step": 1319
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017483200180550554,
+      "loss": 0.8877,
+      "step": 1320
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017479065571856302,
+      "loss": 0.866,
+      "step": 1321
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001747492805953128,
+      "loss": 0.8372,
+      "step": 1322
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017470787645181815,
+      "loss": 0.8328,
+      "step": 1323
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017466644330415364,
+      "loss": 0.8755,
+      "step": 1324
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017462498116840495,
+      "loss": 0.3289,
+      "step": 1325
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017458349006066917,
+      "loss": 0.8076,
+      "step": 1326
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017454196999705456,
+      "loss": 0.8113,
+      "step": 1327
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017450042099368066,
+      "loss": 0.8041,
+      "step": 1328
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001744588430666782,
+      "loss": 0.8979,
+      "step": 1329
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017441723623218915,
+      "loss": 0.8364,
+      "step": 1330
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017437560050636677,
+      "loss": 0.8298,
+      "step": 1331
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001743339359053754,
+      "loss": 0.9358,
+      "step": 1332
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017429224244539077,
+      "loss": 0.8123,
+      "step": 1333
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017425052014259963,
+      "loss": 0.8187,
+      "step": 1334
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017420876901320007,
+      "loss": 0.8992,
+      "step": 1335
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017416698907340126,
+      "loss": 0.866,
+      "step": 1336
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017412518033942372,
+      "loss": 0.8608,
+      "step": 1337
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001740833428274989,
+      "loss": 0.8838,
+      "step": 1338
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017404147655386963,
+      "loss": 0.8334,
+      "step": 1339
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001739995815347899,
+      "loss": 0.8442,
+      "step": 1340
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001739576577865247,
+      "loss": 0.8337,
+      "step": 1341
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001739157053253503,
+      "loss": 0.8831,
+      "step": 1342
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001738737241675541,
+      "loss": 0.8394,
+      "step": 1343
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017383171432943466,
+      "loss": 0.8315,
+      "step": 1344
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017378967582730158,
+      "loss": 0.855,
+      "step": 1345
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017374760867747572,
+      "loss": 0.8372,
+      "step": 1346
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017370551289628896,
+      "loss": 0.8962,
+      "step": 1347
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017366338850008432,
+      "loss": 0.8508,
+      "step": 1348
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000173621235505216,
+      "loss": 0.8347,
+      "step": 1349
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017357905392804916,
+      "loss": 0.8152,
+      "step": 1350
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001735368437849602,
+      "loss": 0.8789,
+      "step": 1351
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017349460509233654,
+      "loss": 0.8411,
+      "step": 1352
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001734523378665767,
+      "loss": 0.3187,
+      "step": 1353
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017341004212409025,
+      "loss": 0.8384,
+      "step": 1354
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017336771788129785,
+      "loss": 0.8582,
+      "step": 1355
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017332536515463125,
+      "loss": 0.9089,
+      "step": 1356
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017328298396053325,
+      "loss": 0.9456,
+      "step": 1357
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017324057431545765,
+      "loss": 0.9128,
+      "step": 1358
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017319813623586935,
+      "loss": 0.8601,
+      "step": 1359
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001731556697382443,
+      "loss": 0.885,
+      "step": 1360
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017311317483906943,
+      "loss": 0.8804,
+      "step": 1361
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017307065155484273,
+      "loss": 0.823,
+      "step": 1362
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017302809990207317,
+      "loss": 0.8611,
+      "step": 1363
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017298551989728084,
+      "loss": 0.8618,
+      "step": 1364
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001729429115569967,
+      "loss": 0.8318,
+      "step": 1365
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017290027489776278,
+      "loss": 0.8538,
+      "step": 1366
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017285760993613215,
+      "loss": 0.8821,
+      "step": 1367
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017281491668866874,
+      "loss": 0.7874,
+      "step": 1368
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001727721951719476,
+      "loss": 0.8938,
+      "step": 1369
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001727294454025547,
+      "loss": 0.8313,
+      "step": 1370
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001726866673970869,
+      "loss": 0.8475,
+      "step": 1371
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017264386117215216,
+      "loss": 0.8787,
+      "step": 1372
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017260102674436932,
+      "loss": 0.8757,
+      "step": 1373
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001725581641303682,
+      "loss": 0.8064,
+      "step": 1374
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017251527334678944,
+      "loss": 0.3466,
+      "step": 1375
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017247235441028485,
+      "loss": 0.9045,
+      "step": 1376
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017242940733751695,
+      "loss": 0.3324,
+      "step": 1377
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017238643214515932,
+      "loss": 0.8635,
+      "step": 1378
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017234342884989642,
+      "loss": 0.8508,
+      "step": 1379
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017230039746842353,
+      "loss": 0.8906,
+      "step": 1380
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000172257338017447,
+      "loss": 0.895,
+      "step": 1381
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017221425051368395,
+      "loss": 0.8794,
+      "step": 1382
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017217113497386242,
+      "loss": 0.7888,
+      "step": 1383
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001721279914147214,
+      "loss": 0.826,
+      "step": 1384
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017208481985301065,
+      "loss": 0.875,
+      "step": 1385
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017204162030549092,
+      "loss": 0.8477,
+      "step": 1386
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017199839278893368,
+      "loss": 0.8618,
+      "step": 1387
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001719551373201214,
+      "loss": 0.8367,
+      "step": 1388
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017191185391584733,
+      "loss": 0.8613,
+      "step": 1389
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017186854259291557,
+      "loss": 0.8115,
+      "step": 1390
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017182520336814106,
+      "loss": 0.3361,
+      "step": 1391
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017178183625834957,
+      "loss": 0.8359,
+      "step": 1392
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017173844128037775,
+      "loss": 0.7809,
+      "step": 1393
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000171695018451073,
+      "loss": 0.8584,
+      "step": 1394
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017165156778729354,
+      "loss": 0.3596,
+      "step": 1395
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017160808930590844,
+      "loss": 0.8481,
+      "step": 1396
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017156458302379753,
+      "loss": 0.9138,
+      "step": 1397
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017152104895785147,
+      "loss": 0.309,
+      "step": 1398
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017147748712497164,
+      "loss": 0.8525,
+      "step": 1399
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017143389754207026,
+      "loss": 0.8711,
+      "step": 1400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017139028022607032,
+      "loss": 0.8318,
+      "step": 1401
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017134663519390557,
+      "loss": 0.8667,
+      "step": 1402
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017130296246252046,
+      "loss": 0.8442,
+      "step": 1403
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017125926204887034,
+      "loss": 0.8025,
+      "step": 1404
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017121553396992113,
+      "loss": 0.8586,
+      "step": 1405
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017117177824264962,
+      "loss": 0.8257,
+      "step": 1406
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017112799488404326,
+      "loss": 0.8188,
+      "step": 1407
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017108418391110033,
+      "loss": 0.843,
+      "step": 1408
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017104034534082968,
+      "loss": 0.8364,
+      "step": 1409
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017099647919025096,
+      "loss": 0.8417,
+      "step": 1410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017095258547639454,
+      "loss": 0.8591,
+      "step": 1411
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001709086642163015,
+      "loss": 0.8691,
+      "step": 1412
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017086471542702354,
+      "loss": 0.8218,
+      "step": 1413
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017082073912562309,
+      "loss": 0.8538,
+      "step": 1414
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001707767353291733,
+      "loss": 0.3314,
+      "step": 1415
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017073270405475795,
+      "loss": 0.8704,
+      "step": 1416
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017068864531947147,
+      "loss": 0.842,
+      "step": 1417
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017064455914041896,
+      "loss": 0.8135,
+      "step": 1418
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017060044553471627,
+      "loss": 0.8439,
+      "step": 1419
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001705563045194898,
+      "loss": 0.822,
+      "step": 1420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017051213611187654,
+      "loss": 0.7996,
+      "step": 1421
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001704679403290243,
+      "loss": 0.3026,
+      "step": 1422
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001704237171880913,
+      "loss": 0.8184,
+      "step": 1423
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017037946670624653,
+      "loss": 0.8279,
+      "step": 1424
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017033518890066955,
+      "loss": 0.8406,
+      "step": 1425
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017029088378855053,
+      "loss": 0.3511,
+      "step": 1426
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017024655138709024,
+      "loss": 0.3484,
+      "step": 1427
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017020219171350002,
+      "loss": 0.865,
+      "step": 1428
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017015780478500187,
+      "loss": 0.8486,
+      "step": 1429
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001701133906188283,
+      "loss": 0.8263,
+      "step": 1430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017006894923222242,
+      "loss": 0.8596,
+      "step": 1431
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001700244806424379,
+      "loss": 0.9243,
+      "step": 1432
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016997998486673895,
+      "loss": 0.8547,
+      "step": 1433
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016993546192240041,
+      "loss": 0.8525,
+      "step": 1434
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016989091182670763,
+      "loss": 0.8176,
+      "step": 1435
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016984633459695645,
+      "loss": 0.8162,
+      "step": 1436
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016980173025045328,
+      "loss": 0.8364,
+      "step": 1437
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001697570988045151,
+      "loss": 0.936,
+      "step": 1438
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016971244027646937,
+      "loss": 0.3392,
+      "step": 1439
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000169667754683654,
+      "loss": 0.8806,
+      "step": 1440
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001696230420434176,
+      "loss": 0.3613,
+      "step": 1441
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016957830237311904,
+      "loss": 0.8721,
+      "step": 1442
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016953353569012784,
+      "loss": 0.9243,
+      "step": 1443
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016948874201182403,
+      "loss": 0.809,
+      "step": 1444
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016944392135559796,
+      "loss": 0.863,
+      "step": 1445
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016939907373885062,
+      "loss": 0.9246,
+      "step": 1446
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016935419917899335,
+      "loss": 0.8525,
+      "step": 1447
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016930929769344805,
+      "loss": 0.8982,
+      "step": 1448
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016926436929964699,
+      "loss": 0.863,
+      "step": 1449
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001692194140150329,
+      "loss": 0.8733,
+      "step": 1450
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000169174431857059,
+      "loss": 0.9146,
+      "step": 1451
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016912942284318894,
+      "loss": 0.8684,
+      "step": 1452
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016908438699089672,
+      "loss": 0.8562,
+      "step": 1453
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001690393243176668,
+      "loss": 0.8784,
+      "step": 1454
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016899423484099412,
+      "loss": 0.8082,
+      "step": 1455
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016894911857838393,
+      "loss": 0.9141,
+      "step": 1456
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016890397554735192,
+      "loss": 0.8687,
+      "step": 1457
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016885880576542417,
+      "loss": 0.8075,
+      "step": 1458
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016881360925013713,
+      "loss": 0.7521,
+      "step": 1459
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016876838601903767,
+      "loss": 0.856,
+      "step": 1460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016872313608968294,
+      "loss": 0.8112,
+      "step": 1461
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016867785947964064,
+      "loss": 0.7866,
+      "step": 1462
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016863255620648863,
+      "loss": 0.8689,
+      "step": 1463
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001685872262878152,
+      "loss": 0.8293,
+      "step": 1464
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000168541869741219,
+      "loss": 0.8489,
+      "step": 1465
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000168496486584309,
+      "loss": 0.9233,
+      "step": 1466
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016845107683470452,
+      "loss": 0.885,
+      "step": 1467
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016840564051003514,
+      "loss": 0.3306,
+      "step": 1468
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016836017762794086,
+      "loss": 0.8245,
+      "step": 1469
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001683146882060719,
+      "loss": 0.3431,
+      "step": 1470
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016826917226208887,
+      "loss": 0.8105,
+      "step": 1471
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016822362981366257,
+      "loss": 0.9136,
+      "step": 1472
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016817806087847417,
+      "loss": 0.7979,
+      "step": 1473
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001681324654742151,
+      "loss": 0.7703,
+      "step": 1474
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016808684361858703,
+      "loss": 0.7828,
+      "step": 1475
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000168041195329302,
+      "loss": 0.8767,
+      "step": 1476
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016799552062408225,
+      "loss": 0.8564,
+      "step": 1477
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016794981952066018,
+      "loss": 0.8779,
+      "step": 1478
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001679040920367786,
+      "loss": 0.8562,
+      "step": 1479
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001678583381901905,
+      "loss": 0.3502,
+      "step": 1480
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001678125579986591,
+      "loss": 0.8525,
+      "step": 1481
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001677667514799578,
+      "loss": 0.8501,
+      "step": 1482
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001677209186518703,
+      "loss": 0.8293,
+      "step": 1483
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016767505953219048,
+      "loss": 0.853,
+      "step": 1484
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016762917413872246,
+      "loss": 0.8569,
+      "step": 1485
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016758326248928049,
+      "loss": 0.7998,
+      "step": 1486
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016753732460168905,
+      "loss": 0.8567,
+      "step": 1487
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001674913604937828,
+      "loss": 0.7898,
+      "step": 1488
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001674453701834066,
+      "loss": 0.8164,
+      "step": 1489
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016739935368841554,
+      "loss": 0.8784,
+      "step": 1490
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016735331102667473,
+      "loss": 0.8303,
+      "step": 1491
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016730724221605952,
+      "loss": 0.8699,
+      "step": 1492
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016726114727445544,
+      "loss": 0.8865,
+      "step": 1493
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016721502621975813,
+      "loss": 0.8623,
+      "step": 1494
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016716887906987333,
+      "loss": 0.8019,
+      "step": 1495
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000167122705842717,
+      "loss": 0.897,
+      "step": 1496
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016707650655621516,
+      "loss": 0.844,
+      "step": 1497
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000167030281228304,
+      "loss": 0.8328,
+      "step": 1498
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016698402987692966,
+      "loss": 0.3353,
+      "step": 1499
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016693775252004867,
+      "loss": 0.7937,
+      "step": 1500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001668914491756274,
+      "loss": 0.855,
+      "step": 1501
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001668451198616424,
+      "loss": 0.8257,
+      "step": 1502
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016679876459608034,
+      "loss": 0.8401,
+      "step": 1503
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016675238339693791,
+      "loss": 0.8591,
+      "step": 1504
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016670597628222188,
+      "loss": 0.8516,
+      "step": 1505
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001666595432699491,
+      "loss": 0.8782,
+      "step": 1506
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001666130843781465,
+      "loss": 0.7686,
+      "step": 1507
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016656659962485098,
+      "loss": 0.8906,
+      "step": 1508
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001665200890281095,
+      "loss": 0.8293,
+      "step": 1509
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016647355260597914,
+      "loss": 0.7587,
+      "step": 1510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016642699037652692,
+      "loss": 0.8557,
+      "step": 1511
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016638040235782982,
+      "loss": 0.8219,
+      "step": 1512
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016633378856797503,
+      "loss": 0.7852,
+      "step": 1513
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016628714902505956,
+      "loss": 0.8567,
+      "step": 1514
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016624048374719053,
+      "loss": 0.8562,
+      "step": 1515
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000166193792752485,
+      "loss": 0.908,
+      "step": 1516
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016614707605906996,
+      "loss": 0.8555,
+      "step": 1517
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001661003336850825,
+      "loss": 0.906,
+      "step": 1518
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016605356564866958,
+      "loss": 0.873,
+      "step": 1519
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016600677196798821,
+      "loss": 0.8438,
+      "step": 1520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016595995266120527,
+      "loss": 0.905,
+      "step": 1521
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016591310774649766,
+      "loss": 0.8953,
+      "step": 1522
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016586623724205215,
+      "loss": 0.8677,
+      "step": 1523
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016581934116606552,
+      "loss": 0.8904,
+      "step": 1524
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016577241953674439,
+      "loss": 0.8521,
+      "step": 1525
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001657254723723054,
+      "loss": 0.8127,
+      "step": 1526
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016567849969097503,
+      "loss": 0.8604,
+      "step": 1527
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001656315015109897,
+      "loss": 0.8535,
+      "step": 1528
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016558447785059577,
+      "loss": 0.8938,
+      "step": 1529
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016553742872804937,
+      "loss": 0.8037,
+      "step": 1530
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016549035416161664,
+      "loss": 0.8298,
+      "step": 1531
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001654432541695735,
+      "loss": 0.8037,
+      "step": 1532
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016539612877020582,
+      "loss": 0.8774,
+      "step": 1533
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016534897798180928,
+      "loss": 0.8713,
+      "step": 1534
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016530180182268946,
+      "loss": 0.336,
+      "step": 1535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016525460031116177,
+      "loss": 0.873,
+      "step": 1536
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016520737346555149,
+      "loss": 0.9092,
+      "step": 1537
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016516012130419366,
+      "loss": 0.8569,
+      "step": 1538
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016511284384543317,
+      "loss": 0.8152,
+      "step": 1539
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016506554110762482,
+      "loss": 0.8057,
+      "step": 1540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016501821310913316,
+      "loss": 0.3331,
+      "step": 1541
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016497085986833253,
+      "loss": 0.8396,
+      "step": 1542
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016492348140360706,
+      "loss": 0.8818,
+      "step": 1543
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016487607773335073,
+      "loss": 0.3272,
+      "step": 1544
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001648286488759673,
+      "loss": 0.832,
+      "step": 1545
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016478119484987025,
+      "loss": 0.8811,
+      "step": 1546
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016473371567348285,
+      "loss": 0.8611,
+      "step": 1547
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016468621136523823,
+      "loss": 0.823,
+      "step": 1548
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001646386819435791,
+      "loss": 0.8691,
+      "step": 1549
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016459112742695807,
+      "loss": 0.8247,
+      "step": 1550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016454354783383747,
+      "loss": 0.8002,
+      "step": 1551
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016449594318268927,
+      "loss": 0.8828,
+      "step": 1552
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016444831349199528,
+      "loss": 0.7981,
+      "step": 1553
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016440065878024697,
+      "loss": 0.8291,
+      "step": 1554
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016435297906594555,
+      "loss": 0.8706,
+      "step": 1555
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016430527436760191,
+      "loss": 0.8875,
+      "step": 1556
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016425754470373666,
+      "loss": 0.8241,
+      "step": 1557
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001642097900928801,
+      "loss": 0.8369,
+      "step": 1558
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016416201055357224,
+      "loss": 0.7839,
+      "step": 1559
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001641142061043627,
+      "loss": 0.832,
+      "step": 1560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016406637676381083,
+      "loss": 0.854,
+      "step": 1561
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016401852255048564,
+      "loss": 0.8635,
+      "step": 1562
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016397064348296577,
+      "loss": 0.8723,
+      "step": 1563
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016392273957983956,
+      "loss": 0.8699,
+      "step": 1564
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001638748108597049,
+      "loss": 0.8611,
+      "step": 1565
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016382685734116933,
+      "loss": 0.894,
+      "step": 1566
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016377887904285018,
+      "loss": 0.9021,
+      "step": 1567
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016373087598337422,
+      "loss": 0.8948,
+      "step": 1568
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016368284818137787,
+      "loss": 0.8057,
+      "step": 1569
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016363479565550718,
+      "loss": 0.9226,
+      "step": 1570
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001635867184244178,
+      "loss": 0.7943,
+      "step": 1571
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000163538616506775,
+      "loss": 0.8936,
+      "step": 1572
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016349048992125357,
+      "loss": 0.8838,
+      "step": 1573
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016344233868653788,
+      "loss": 0.8002,
+      "step": 1574
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016339416282132196,
+      "loss": 0.7946,
+      "step": 1575
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001633459623443093,
+      "loss": 0.8237,
+      "step": 1576
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016329773727421297,
+      "loss": 0.8677,
+      "step": 1577
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016324948762975566,
+      "loss": 0.8667,
+      "step": 1578
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001632012134296695,
+      "loss": 0.8206,
+      "step": 1579
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001631529146926962,
+      "loss": 0.8545,
+      "step": 1580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000163104591437587,
+      "loss": 0.8933,
+      "step": 1581
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016305624368310263,
+      "loss": 0.8572,
+      "step": 1582
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001630078714480134,
+      "loss": 0.8728,
+      "step": 1583
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016295947475109904,
+      "loss": 0.8391,
+      "step": 1584
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001629110536111488,
+      "loss": 0.368,
+      "step": 1585
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001628626080469615,
+      "loss": 0.8997,
+      "step": 1586
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001628141380773453,
+      "loss": 0.8503,
+      "step": 1587
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016276564372111796,
+      "loss": 0.3329,
+      "step": 1588
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016271712499710663,
+      "loss": 0.9131,
+      "step": 1589
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016266858192414798,
+      "loss": 0.3538,
+      "step": 1590
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016262001452108807,
+      "loss": 0.7963,
+      "step": 1591
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016257142280678245,
+      "loss": 0.8755,
+      "step": 1592
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016252280680009612,
+      "loss": 0.885,
+      "step": 1593
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016247416651990343,
+      "loss": 0.855,
+      "step": 1594
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001624255019850883,
+      "loss": 0.8586,
+      "step": 1595
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016237681321454387,
+      "loss": 0.842,
+      "step": 1596
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001623281002271729,
+      "loss": 0.96,
+      "step": 1597
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001622793630418874,
+      "loss": 0.8314,
+      "step": 1598
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001622306016776088,
+      "loss": 0.3307,
+      "step": 1599
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016218181615326795,
+      "loss": 0.875,
+      "step": 1600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016213300648780513,
+      "loss": 0.8816,
+      "step": 1601
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001620841727001699,
+      "loss": 0.8245,
+      "step": 1602
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016203531480932115,
+      "loss": 0.8577,
+      "step": 1603
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016198643283422729,
+      "loss": 0.7977,
+      "step": 1604
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016193752679386593,
+      "loss": 0.825,
+      "step": 1605
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001618885967072241,
+      "loss": 0.7946,
+      "step": 1606
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016183964259329815,
+      "loss": 0.767,
+      "step": 1607
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016179066447109377,
+      "loss": 0.8406,
+      "step": 1608
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016174166235962586,
+      "loss": 0.843,
+      "step": 1609
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016169263627791887,
+      "loss": 0.8101,
+      "step": 1610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001616435862450063,
+      "loss": 0.8572,
+      "step": 1611
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001615945122799311,
+      "loss": 0.8862,
+      "step": 1612
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016154541440174548,
+      "loss": 0.8735,
+      "step": 1613
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001614962926295109,
+      "loss": 0.8083,
+      "step": 1614
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016144714698229816,
+      "loss": 0.8801,
+      "step": 1615
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016139797747918725,
+      "loss": 0.8362,
+      "step": 1616
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001613487841392675,
+      "loss": 0.908,
+      "step": 1617
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016129956698163745,
+      "loss": 0.8716,
+      "step": 1618
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001612503260254049,
+      "loss": 0.8489,
+      "step": 1619
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016120106128968687,
+      "loss": 0.8418,
+      "step": 1620
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016115177279360965,
+      "loss": 0.755,
+      "step": 1621
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016110246055630868,
+      "loss": 0.8018,
+      "step": 1622
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016105312459692873,
+      "loss": 0.8347,
+      "step": 1623
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016100376493462368,
+      "loss": 0.8689,
+      "step": 1624
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016095438158855667,
+      "loss": 0.8479,
+      "step": 1625
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016090497457789998,
+      "loss": 0.8936,
+      "step": 1626
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016085554392183517,
+      "loss": 0.8547,
+      "step": 1627
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016080608963955288,
+      "loss": 0.8447,
+      "step": 1628
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016075661175025298,
+      "loss": 0.8469,
+      "step": 1629
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016070711027314446,
+      "loss": 0.8124,
+      "step": 1630
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016065758522744558,
+      "loss": 0.8145,
+      "step": 1631
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001606080366323836,
+      "loss": 0.8984,
+      "step": 1632
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016055846450719496,
+      "loss": 0.8154,
+      "step": 1633
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016050886887112532,
+      "loss": 0.8567,
+      "step": 1634
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016045924974342942,
+      "loss": 0.8147,
+      "step": 1635
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001604096071433711,
+      "loss": 0.9336,
+      "step": 1636
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016035994109022333,
+      "loss": 0.8447,
+      "step": 1637
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016031025160326813,
+      "loss": 0.886,
+      "step": 1638
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016026053870179675,
+      "loss": 0.8757,
+      "step": 1639
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016021080240510944,
+      "loss": 0.906,
+      "step": 1640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001601610427325155,
+      "loss": 0.8359,
+      "step": 1641
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016011125970333333,
+      "loss": 0.3237,
+      "step": 1642
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016006145333689049,
+      "loss": 0.7959,
+      "step": 1643
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016001162365252347,
+      "loss": 0.7878,
+      "step": 1644
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015996177066957787,
+      "loss": 0.9263,
+      "step": 1645
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001599118944074084,
+      "loss": 0.8398,
+      "step": 1646
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015986199488537866,
+      "loss": 0.8093,
+      "step": 1647
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001598120721228614,
+      "loss": 0.8401,
+      "step": 1648
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015976212613923836,
+      "loss": 0.8926,
+      "step": 1649
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015971215695390027,
+      "loss": 0.8623,
+      "step": 1650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015966216458624692,
+      "loss": 0.8171,
+      "step": 1651
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015961214905568704,
+      "loss": 0.8652,
+      "step": 1652
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001595621103816384,
+      "loss": 0.8743,
+      "step": 1653
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015951204858352772,
+      "loss": 0.8413,
+      "step": 1654
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001594619636807907,
+      "loss": 0.8381,
+      "step": 1655
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015941185569287204,
+      "loss": 0.8154,
+      "step": 1656
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015936172463922542,
+      "loss": 0.8691,
+      "step": 1657
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001593115705393134,
+      "loss": 0.8494,
+      "step": 1658
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015926139341260756,
+      "loss": 0.812,
+      "step": 1659
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015921119327858834,
+      "loss": 0.8892,
+      "step": 1660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015916097015674518,
+      "loss": 0.8389,
+      "step": 1661
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015911072406657646,
+      "loss": 0.8682,
+      "step": 1662
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001590604550275894,
+      "loss": 0.8088,
+      "step": 1663
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001590101630593002,
+      "loss": 0.7888,
+      "step": 1664
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001589598481812339,
+      "loss": 0.8643,
+      "step": 1665
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001589095104129245,
+      "loss": 0.832,
+      "step": 1666
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015885914977391486,
+      "loss": 0.3516,
+      "step": 1667
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015880876628375666,
+      "loss": 0.8347,
+      "step": 1668
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015875835996201058,
+      "loss": 0.8223,
+      "step": 1669
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015870793082824604,
+      "loss": 0.8381,
+      "step": 1670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015865747890204138,
+      "loss": 0.895,
+      "step": 1671
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015860700420298378,
+      "loss": 0.8054,
+      "step": 1672
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015855650675066925,
+      "loss": 0.8389,
+      "step": 1673
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015850598656470263,
+      "loss": 0.8716,
+      "step": 1674
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001584554436646976,
+      "loss": 0.8638,
+      "step": 1675
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015840487807027665,
+      "loss": 0.8279,
+      "step": 1676
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015835428980107112,
+      "loss": 0.8699,
+      "step": 1677
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001583036788767211,
+      "loss": 0.8582,
+      "step": 1678
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001582530453168755,
+      "loss": 0.8989,
+      "step": 1679
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015820238914119195,
+      "loss": 0.8584,
+      "step": 1680
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000158151710369337,
+      "loss": 0.8484,
+      "step": 1681
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015810100902098582,
+      "loss": 0.8381,
+      "step": 1682
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001580502851158225,
+      "loss": 0.3417,
+      "step": 1683
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015799953867353975,
+      "loss": 0.863,
+      "step": 1684
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001579487697138391,
+      "loss": 0.7969,
+      "step": 1685
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015789797825643085,
+      "loss": 0.8999,
+      "step": 1686
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015784716432103393,
+      "loss": 0.8528,
+      "step": 1687
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015779632792737607,
+      "loss": 0.8374,
+      "step": 1688
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015774546909519375,
+      "loss": 0.8174,
+      "step": 1689
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015769458784423206,
+      "loss": 0.3434,
+      "step": 1690
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001576436841942449,
+      "loss": 0.8936,
+      "step": 1691
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001575927581649948,
+      "loss": 0.8804,
+      "step": 1692
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015754180977625303,
+      "loss": 0.864,
+      "step": 1693
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015749083904779946,
+      "loss": 0.8735,
+      "step": 1694
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015743984599942272,
+      "loss": 0.8208,
+      "step": 1695
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015738883065092004,
+      "loss": 0.8282,
+      "step": 1696
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015733779302209736,
+      "loss": 0.8696,
+      "step": 1697
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001572867331327692,
+      "loss": 0.8816,
+      "step": 1698
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015723565100275884,
+      "loss": 0.7905,
+      "step": 1699
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015718454665189806,
+      "loss": 0.916,
+      "step": 1700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015713342010002733,
+      "loss": 0.8682,
+      "step": 1701
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015708227136699575,
+      "loss": 0.8218,
+      "step": 1702
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015703110047266106,
+      "loss": 0.3376,
+      "step": 1703
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001569799074368895,
+      "loss": 0.3558,
+      "step": 1704
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015692869227955603,
+      "loss": 0.796,
+      "step": 1705
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015687745502054407,
+      "loss": 0.8787,
+      "step": 1706
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015682619567974576,
+      "loss": 0.8083,
+      "step": 1707
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015677491427706168,
+      "loss": 0.8218,
+      "step": 1708
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015672361083240107,
+      "loss": 0.8376,
+      "step": 1709
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015667228536568167,
+      "loss": 0.8115,
+      "step": 1710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015662093789682978,
+      "loss": 0.7827,
+      "step": 1711
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001565695684457803,
+      "loss": 0.8014,
+      "step": 1712
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015651817703247667,
+      "loss": 0.7896,
+      "step": 1713
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015646676367687067,
+      "loss": 0.8469,
+      "step": 1714
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001564153283989228,
+      "loss": 0.8613,
+      "step": 1715
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015636387121860206,
+      "loss": 0.8442,
+      "step": 1716
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015631239215588578,
+      "loss": 0.8486,
+      "step": 1717
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015626089123076,
+      "loss": 0.8522,
+      "step": 1718
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015620936846321917,
+      "loss": 0.8926,
+      "step": 1719
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001561578238732661,
+      "loss": 0.8325,
+      "step": 1720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015610625748091228,
+      "loss": 0.8672,
+      "step": 1721
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015605466930617746,
+      "loss": 0.8579,
+      "step": 1722
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015600305936909004,
+      "loss": 0.8524,
+      "step": 1723
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015595142768968668,
+      "loss": 0.842,
+      "step": 1724
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015589977428801267,
+      "loss": 0.8391,
+      "step": 1725
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015584809918412157,
+      "loss": 0.804,
+      "step": 1726
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015579640239807548,
+      "loss": 0.8745,
+      "step": 1727
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015574468394994486,
+      "loss": 0.8326,
+      "step": 1728
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015569294385980856,
+      "loss": 0.8506,
+      "step": 1729
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001556411821477539,
+      "loss": 0.9011,
+      "step": 1730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015558939883387655,
+      "loss": 0.8335,
+      "step": 1731
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015553759393828058,
+      "loss": 0.8972,
+      "step": 1732
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015548576748107838,
+      "loss": 0.8181,
+      "step": 1733
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015543391948239088,
+      "loss": 0.8267,
+      "step": 1734
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015538204996234716,
+      "loss": 0.8086,
+      "step": 1735
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001553301589410848,
+      "loss": 0.7678,
+      "step": 1736
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015527824643874966,
+      "loss": 0.8418,
+      "step": 1737
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015522631247549598,
+      "loss": 0.8389,
+      "step": 1738
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015517435707148626,
+      "loss": 0.7937,
+      "step": 1739
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015512238024689142,
+      "loss": 0.8635,
+      "step": 1740
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001550703820218907,
+      "loss": 0.8914,
+      "step": 1741
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001550183624166715,
+      "loss": 0.8394,
+      "step": 1742
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001549663214514297,
+      "loss": 0.8662,
+      "step": 1743
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015491425914636935,
+      "loss": 0.865,
+      "step": 1744
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015486217552170283,
+      "loss": 0.8811,
+      "step": 1745
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015481007059765082,
+      "loss": 0.8796,
+      "step": 1746
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015475794439444226,
+      "loss": 0.8323,
+      "step": 1747
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015470579693231432,
+      "loss": 0.8354,
+      "step": 1748
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015465362823151245,
+      "loss": 0.7885,
+      "step": 1749
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015460143831229025,
+      "loss": 0.8145,
+      "step": 1750
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015454922719490978,
+      "loss": 0.8352,
+      "step": 1751
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001544969948996411,
+      "loss": 0.8123,
+      "step": 1752
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001544447414467626,
+      "loss": 0.8711,
+      "step": 1753
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015439246685656092,
+      "loss": 0.8152,
+      "step": 1754
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001543401711493308,
+      "loss": 0.3445,
+      "step": 1755
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015428785434537526,
+      "loss": 0.3306,
+      "step": 1756
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015423551646500546,
+      "loss": 0.8853,
+      "step": 1757
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001541831575285408,
+      "loss": 0.8254,
+      "step": 1758
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015413077755630877,
+      "loss": 0.865,
+      "step": 1759
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001540783765686452,
+      "loss": 0.3312,
+      "step": 1760
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001540259545858938,
+      "loss": 0.8256,
+      "step": 1761
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001539735116284067,
+      "loss": 0.8081,
+      "step": 1762
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000153921047716544,
+      "loss": 0.8406,
+      "step": 1763
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015386856287067402,
+      "loss": 0.8491,
+      "step": 1764
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015381605711117316,
+      "loss": 0.8777,
+      "step": 1765
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015376353045842603,
+      "loss": 0.8442,
+      "step": 1766
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015371098293282524,
+      "loss": 0.8137,
+      "step": 1767
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015365841455477157,
+      "loss": 0.8955,
+      "step": 1768
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001536058253446738,
+      "loss": 0.8542,
+      "step": 1769
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015355321532294896,
+      "loss": 0.8317,
+      "step": 1770
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015350058451002205,
+      "loss": 0.8748,
+      "step": 1771
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015344793292632614,
+      "loss": 0.876,
+      "step": 1772
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001533952605923024,
+      "loss": 0.8472,
+      "step": 1773
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015334256752840006,
+      "loss": 0.8379,
+      "step": 1774
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015328985375507638,
+      "loss": 0.8375,
+      "step": 1775
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001532371192927966,
+      "loss": 0.8281,
+      "step": 1776
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015318436416203413,
+      "loss": 0.8782,
+      "step": 1777
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001531315883832703,
+      "loss": 0.8092,
+      "step": 1778
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001530787919769945,
+      "loss": 0.8228,
+      "step": 1779
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015302597496370408,
+      "loss": 0.821,
+      "step": 1780
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015297313736390447,
+      "loss": 0.8274,
+      "step": 1781
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015292027919810896,
+      "loss": 0.3448,
+      "step": 1782
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015286740048683898,
+      "loss": 0.8682,
+      "step": 1783
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001528145012506239,
+      "loss": 0.8008,
+      "step": 1784
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015276158151000094,
+      "loss": 0.7966,
+      "step": 1785
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015270864128551543,
+      "loss": 0.8804,
+      "step": 1786
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015265568059772054,
+      "loss": 0.7603,
+      "step": 1787
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015260269946717746,
+      "loss": 0.8729,
+      "step": 1788
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015254969791445526,
+      "loss": 0.8464,
+      "step": 1789
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000152496675960131,
+      "loss": 0.8328,
+      "step": 1790
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015244363362478967,
+      "loss": 0.8401,
+      "step": 1791
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015239057092902405,
+      "loss": 0.8638,
+      "step": 1792
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015233748789343488,
+      "loss": 0.7992,
+      "step": 1793
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015228438453863095,
+      "loss": 0.9008,
+      "step": 1794
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001522312608852287,
+      "loss": 0.8064,
+      "step": 1795
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015217811695385263,
+      "loss": 0.8784,
+      "step": 1796
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000152124952765135,
+      "loss": 0.7964,
+      "step": 1797
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015207176833971598,
+      "loss": 0.8933,
+      "step": 1798
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001520185636982436,
+      "loss": 0.8608,
+      "step": 1799
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015196533886137377,
+      "loss": 0.8297,
+      "step": 1800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015191209384977014,
+      "loss": 0.8267,
+      "step": 1801
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001518588286841043,
+      "loss": 0.8652,
+      "step": 1802
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015180554338505565,
+      "loss": 0.7744,
+      "step": 1803
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001517522379733113,
+      "loss": 0.8813,
+      "step": 1804
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001516989124695663,
+      "loss": 0.7716,
+      "step": 1805
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015164556689452346,
+      "loss": 0.8198,
+      "step": 1806
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001515922012688933,
+      "loss": 0.822,
+      "step": 1807
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015153881561339426,
+      "loss": 0.8394,
+      "step": 1808
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001514854099487524,
+      "loss": 0.8561,
+      "step": 1809
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001514319842957018,
+      "loss": 0.826,
+      "step": 1810
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000151378538674984,
+      "loss": 0.3392,
+      "step": 1811
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015132507310734847,
+      "loss": 0.8376,
+      "step": 1812
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001512715876135524,
+      "loss": 0.3523,
+      "step": 1813
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001512180822143607,
+      "loss": 0.8347,
+      "step": 1814
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015116455693054594,
+      "loss": 0.8167,
+      "step": 1815
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015111101178288857,
+      "loss": 0.8259,
+      "step": 1816
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001510574467921766,
+      "loss": 0.8433,
+      "step": 1817
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015100386197920584,
+      "loss": 0.7966,
+      "step": 1818
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015095025736477978,
+      "loss": 0.8704,
+      "step": 1819
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001508966329697095,
+      "loss": 0.7302,
+      "step": 1820
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015084298881481387,
+      "loss": 0.8596,
+      "step": 1821
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015078932492091943,
+      "loss": 0.3295,
+      "step": 1822
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001507356413088603,
+      "loss": 0.7815,
+      "step": 1823
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015068193799947839,
+      "loss": 0.8025,
+      "step": 1824
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015062821501362307,
+      "loss": 0.8186,
+      "step": 1825
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001505744723721515,
+      "loss": 0.8152,
+      "step": 1826
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015052071009592844,
+      "loss": 0.8708,
+      "step": 1827
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015046692820582625,
+      "loss": 0.3392,
+      "step": 1828
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001504131267227249,
+      "loss": 0.8779,
+      "step": 1829
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015035930566751197,
+      "loss": 0.9062,
+      "step": 1830
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015030546506108268,
+      "loss": 0.8058,
+      "step": 1831
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015025160492433977,
+      "loss": 0.8071,
+      "step": 1832
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001501977252781936,
+      "loss": 0.8071,
+      "step": 1833
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015014382614356213,
+      "loss": 0.8242,
+      "step": 1834
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015008990754137086,
+      "loss": 0.9441,
+      "step": 1835
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015003596949255283,
+      "loss": 0.8193,
+      "step": 1836
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014998201201804867,
+      "loss": 0.8389,
+      "step": 1837
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001499280351388065,
+      "loss": 0.327,
+      "step": 1838
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000149874038875782,
+      "loss": 0.8745,
+      "step": 1839
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001498200232499384,
+      "loss": 0.8101,
+      "step": 1840
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001497659882822464,
+      "loss": 0.8752,
+      "step": 1841
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001497119339936843,
+      "loss": 0.8567,
+      "step": 1842
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001496578604052378,
+      "loss": 0.8357,
+      "step": 1843
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001496037675379001,
+      "loss": 0.7692,
+      "step": 1844
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014954965541267193,
+      "loss": 0.8955,
+      "step": 1845
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001494955240505615,
+      "loss": 0.8691,
+      "step": 1846
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001494413734725844,
+      "loss": 0.8708,
+      "step": 1847
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014938720369976386,
+      "loss": 0.8098,
+      "step": 1848
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014933301475313036,
+      "loss": 0.8059,
+      "step": 1849
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014927880665372196,
+      "loss": 0.8506,
+      "step": 1850
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001492245794225841,
+      "loss": 0.3572,
+      "step": 1851
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014917033308076965,
+      "loss": 0.8328,
+      "step": 1852
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014911606764933893,
+      "loss": 0.8118,
+      "step": 1853
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001490617831493596,
+      "loss": 0.8289,
+      "step": 1854
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014900747960190682,
+      "loss": 0.7942,
+      "step": 1855
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001489531570280631,
+      "loss": 0.8301,
+      "step": 1856
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001488988154489183,
+      "loss": 0.7629,
+      "step": 1857
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001488444548855697,
+      "loss": 0.8174,
+      "step": 1858
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014879007535912198,
+      "loss": 0.7986,
+      "step": 1859
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014873567689068707,
+      "loss": 0.7742,
+      "step": 1860
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014868125950138442,
+      "loss": 0.8389,
+      "step": 1861
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014862682321234062,
+      "loss": 0.8097,
+      "step": 1862
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014857236804468983,
+      "loss": 0.8296,
+      "step": 1863
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014851789401957338,
+      "loss": 0.8669,
+      "step": 1864
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014846340115813993,
+      "loss": 0.7573,
+      "step": 1865
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014840888948154549,
+      "loss": 0.8542,
+      "step": 1866
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001483543590109534,
+      "loss": 0.8357,
+      "step": 1867
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014829980976753426,
+      "loss": 0.8098,
+      "step": 1868
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014824524177246598,
+      "loss": 0.8472,
+      "step": 1869
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014819065504693364,
+      "loss": 0.8024,
+      "step": 1870
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014813604961212983,
+      "loss": 0.8689,
+      "step": 1871
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014808142548925418,
+      "loss": 0.8304,
+      "step": 1872
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014802678269951365,
+      "loss": 0.7888,
+      "step": 1873
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014797212126412243,
+      "loss": 0.8582,
+      "step": 1874
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014791744120430203,
+      "loss": 0.8374,
+      "step": 1875
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001478627425412811,
+      "loss": 0.7881,
+      "step": 1876
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014780802529629558,
+      "loss": 0.8767,
+      "step": 1877
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014775328949058854,
+      "loss": 0.8149,
+      "step": 1878
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014769853514541036,
+      "loss": 0.8525,
+      "step": 1879
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001476437622820185,
+      "loss": 0.8796,
+      "step": 1880
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001475889709216777,
+      "loss": 0.8079,
+      "step": 1881
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014753416108565984,
+      "loss": 0.844,
+      "step": 1882
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000147479332795244,
+      "loss": 0.7815,
+      "step": 1883
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014742448607171644,
+      "loss": 0.8591,
+      "step": 1884
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001473696209363705,
+      "loss": 0.8341,
+      "step": 1885
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014731473741050671,
+      "loss": 0.8713,
+      "step": 1886
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014725983551543278,
+      "loss": 0.8594,
+      "step": 1887
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001472049152724635,
+      "loss": 0.8258,
+      "step": 1888
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014714997670292078,
+      "loss": 0.8381,
+      "step": 1889
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014709501982813368,
+      "loss": 0.8325,
+      "step": 1890
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001470400446694384,
+      "loss": 0.8401,
+      "step": 1891
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014698505124817812,
+      "loss": 0.3611,
+      "step": 1892
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014693003958570318,
+      "loss": 0.7971,
+      "step": 1893
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014687500970337103,
+      "loss": 0.8206,
+      "step": 1894
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014681996162254618,
+      "loss": 0.8494,
+      "step": 1895
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014676489536460014,
+      "loss": 0.8616,
+      "step": 1896
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001467098109509116,
+      "loss": 0.8015,
+      "step": 1897
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014665470840286615,
+      "loss": 0.8259,
+      "step": 1898
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014659958774185654,
+      "loss": 0.8127,
+      "step": 1899
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001465444489892825,
+      "loss": 0.8704,
+      "step": 1900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014648929216655077,
+      "loss": 0.8782,
+      "step": 1901
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014643411729507516,
+      "loss": 0.8491,
+      "step": 1902
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014637892439627642,
+      "loss": 0.8591,
+      "step": 1903
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001463237134915824,
+      "loss": 0.8125,
+      "step": 1904
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001462684846024278,
+      "loss": 0.8689,
+      "step": 1905
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014621323775025445,
+      "loss": 0.8469,
+      "step": 1906
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000146157972956511,
+      "loss": 0.8235,
+      "step": 1907
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014610269024265316,
+      "loss": 0.8601,
+      "step": 1908
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014604738963014364,
+      "loss": 0.3474,
+      "step": 1909
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014599207114045202,
+      "loss": 0.8752,
+      "step": 1910
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014593673479505482,
+      "loss": 0.8572,
+      "step": 1911
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001458813806154355,
+      "loss": 0.3195,
+      "step": 1912
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014582600862308447,
+      "loss": 0.771,
+      "step": 1913
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014577061883949913,
+      "loss": 0.8909,
+      "step": 1914
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014571521128618357,
+      "loss": 0.8494,
+      "step": 1915
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014565978598464894,
+      "loss": 0.8682,
+      "step": 1916
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014560434295641337,
+      "loss": 0.7775,
+      "step": 1917
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001455488822230016,
+      "loss": 0.855,
+      "step": 1918
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014549340380594545,
+      "loss": 0.8076,
+      "step": 1919
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014543790772678357,
+      "loss": 0.8236,
+      "step": 1920
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014538239400706146,
+      "loss": 0.8186,
+      "step": 1921
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014532686266833142,
+      "loss": 0.8479,
+      "step": 1922
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014527131373215263,
+      "loss": 0.8271,
+      "step": 1923
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014521574722009114,
+      "loss": 0.8945,
+      "step": 1924
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014516016315371975,
+      "loss": 0.8239,
+      "step": 1925
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014510456155461806,
+      "loss": 0.8469,
+      "step": 1926
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014504894244437265,
+      "loss": 0.7749,
+      "step": 1927
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014499330584457666,
+      "loss": 0.9028,
+      "step": 1928
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014493765177683016,
+      "loss": 0.8042,
+      "step": 1929
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014488198026274006,
+      "loss": 0.8884,
+      "step": 1930
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014482629132391985,
+      "loss": 0.8167,
+      "step": 1931
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014477058498198992,
+      "loss": 0.8638,
+      "step": 1932
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014471486125857742,
+      "loss": 0.7859,
+      "step": 1933
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001446591201753162,
+      "loss": 0.8638,
+      "step": 1934
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001446033617538469,
+      "loss": 0.843,
+      "step": 1935
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014454758601581676,
+      "loss": 0.3323,
+      "step": 1936
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014449179298287997,
+      "loss": 0.8923,
+      "step": 1937
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014443598267669723,
+      "loss": 0.827,
+      "step": 1938
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000144380155118936,
+      "loss": 0.8931,
+      "step": 1939
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014432431033127056,
+      "loss": 0.8469,
+      "step": 1940
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001442684483353817,
+      "loss": 0.8101,
+      "step": 1941
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014421256915295695,
+      "loss": 0.7865,
+      "step": 1942
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014415667280569063,
+      "loss": 0.7971,
+      "step": 1943
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014410075931528356,
+      "loss": 0.8242,
+      "step": 1944
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014404482870344323,
+      "loss": 0.8158,
+      "step": 1945
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014398888099188394,
+      "loss": 0.8344,
+      "step": 1946
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014393291620232645,
+      "loss": 0.8879,
+      "step": 1947
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014387693435649826,
+      "loss": 0.8357,
+      "step": 1948
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014382093547613337,
+      "loss": 0.802,
+      "step": 1949
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014376491958297262,
+      "loss": 0.8843,
+      "step": 1950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014370888669876317,
+      "loss": 0.3348,
+      "step": 1951
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014365283684525894,
+      "loss": 0.8577,
+      "step": 1952
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014359677004422046,
+      "loss": 0.8413,
+      "step": 1953
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014354068631741476,
+      "loss": 0.8484,
+      "step": 1954
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014348458568661548,
+      "loss": 0.7869,
+      "step": 1955
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001434284681736028,
+      "loss": 0.8518,
+      "step": 1956
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014337233380016352,
+      "loss": 0.8604,
+      "step": 1957
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001433161825880909,
+      "loss": 0.3306,
+      "step": 1958
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014326001455918478,
+      "loss": 0.9314,
+      "step": 1959
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001432038297352515,
+      "loss": 0.8455,
+      "step": 1960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000143147628138104,
+      "loss": 0.8252,
+      "step": 1961
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001430914097895616,
+      "loss": 0.8274,
+      "step": 1962
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001430351747114503,
+      "loss": 0.7769,
+      "step": 1963
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001429789229256024,
+      "loss": 0.8396,
+      "step": 1964
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001429226544538568,
+      "loss": 0.8379,
+      "step": 1965
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014286636931805887,
+      "loss": 0.8381,
+      "step": 1966
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014281006754006045,
+      "loss": 0.8179,
+      "step": 1967
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001427537491417198,
+      "loss": 0.3367,
+      "step": 1968
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001426974141449017,
+      "loss": 0.793,
+      "step": 1969
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014264106257147732,
+      "loss": 0.8167,
+      "step": 1970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014258469444332423,
+      "loss": 0.8479,
+      "step": 1971
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014252830978232656,
+      "loss": 0.3184,
+      "step": 1972
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014247190861037474,
+      "loss": 0.8579,
+      "step": 1973
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014241549094936567,
+      "loss": 0.8186,
+      "step": 1974
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014235905682120255,
+      "loss": 0.8557,
+      "step": 1975
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014230260624779512,
+      "loss": 0.7821,
+      "step": 1976
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014224613925105947,
+      "loss": 0.8574,
+      "step": 1977
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014218965585291793,
+      "loss": 0.8843,
+      "step": 1978
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014213315607529938,
+      "loss": 0.7719,
+      "step": 1979
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014207663994013895,
+      "loss": 0.885,
+      "step": 1980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014202010746937815,
+      "loss": 0.8608,
+      "step": 1981
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014196355868496485,
+      "loss": 0.8162,
+      "step": 1982
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014190699360885323,
+      "loss": 0.8511,
+      "step": 1983
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014185041226300374,
+      "loss": 0.8691,
+      "step": 1984
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001417938146693833,
+      "loss": 0.8104,
+      "step": 1985
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000141737200849965,
+      "loss": 0.8584,
+      "step": 1986
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014168057082672826,
+      "loss": 0.8784,
+      "step": 1987
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014162392462165884,
+      "loss": 0.9036,
+      "step": 1988
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014156726225674873,
+      "loss": 0.7913,
+      "step": 1989
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001415105837539962,
+      "loss": 0.8596,
+      "step": 1990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001414538891354058,
+      "loss": 0.8142,
+      "step": 1991
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014139717842298835,
+      "loss": 0.7592,
+      "step": 1992
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014134045163876087,
+      "loss": 0.361,
+      "step": 1993
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014128370880474666,
+      "loss": 0.8445,
+      "step": 1994
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001412269499429753,
+      "loss": 0.8413,
+      "step": 1995
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014117017507548244,
+      "loss": 0.8552,
+      "step": 1996
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014111338422431014,
+      "loss": 0.8473,
+      "step": 1997
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014105657741150647,
+      "loss": 0.845,
+      "step": 1998
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014099975465912582,
+      "loss": 0.3361,
+      "step": 1999
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014094291598922877,
+      "loss": 0.8584,
+      "step": 2000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014088606142388202,
+      "loss": 0.8574,
+      "step": 2001
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014082919098515846,
+      "loss": 0.834,
+      "step": 2002
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001407723046951372,
+      "loss": 0.305,
+      "step": 2003
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001407154025759034,
+      "loss": 0.8435,
+      "step": 2004
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014065848464954846,
+      "loss": 0.8577,
+      "step": 2005
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014060155093816986,
+      "loss": 0.3557,
+      "step": 2006
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014054460146387124,
+      "loss": 0.8267,
+      "step": 2007
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014048763624876235,
+      "loss": 0.8315,
+      "step": 2008
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014043065531495902,
+      "loss": 0.8575,
+      "step": 2009
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014037365868458326,
+      "loss": 0.3496,
+      "step": 2010
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014031664637976306,
+      "loss": 0.834,
+      "step": 2011
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001402596184226326,
+      "loss": 0.7703,
+      "step": 2012
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014020257483533208,
+      "loss": 0.8558,
+      "step": 2013
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001401455156400078,
+      "loss": 0.8406,
+      "step": 2014
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001400884408588121,
+      "loss": 0.8433,
+      "step": 2015
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001400313505139034,
+      "loss": 0.7922,
+      "step": 2016
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013997424462744605,
+      "loss": 0.8262,
+      "step": 2017
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013991712322161063,
+      "loss": 0.8723,
+      "step": 2018
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013985998631857357,
+      "loss": 0.8699,
+      "step": 2019
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001398028339405174,
+      "loss": 0.8369,
+      "step": 2020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013974566610963068,
+      "loss": 0.7894,
+      "step": 2021
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013968848284810786,
+      "loss": 0.8645,
+      "step": 2022
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001396312841781495,
+      "loss": 0.7937,
+      "step": 2023
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013957407012196203,
+      "loss": 0.8055,
+      "step": 2024
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000139516840701758,
+      "loss": 0.8396,
+      "step": 2025
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001394595959397558,
+      "loss": 0.7555,
+      "step": 2026
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013940233585817984,
+      "loss": 0.8601,
+      "step": 2027
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001393450604792604,
+      "loss": 0.8313,
+      "step": 2028
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013928776982523384,
+      "loss": 0.8048,
+      "step": 2029
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013923046391834228,
+      "loss": 0.8176,
+      "step": 2030
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001391731427808339,
+      "loss": 0.8369,
+      "step": 2031
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001391158064349627,
+      "loss": 0.9043,
+      "step": 2032
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013905845490298866,
+      "loss": 0.835,
+      "step": 2033
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001390010882071776,
+      "loss": 0.8716,
+      "step": 2034
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013894370636980126,
+      "loss": 0.7408,
+      "step": 2035
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013888630941313727,
+      "loss": 0.8195,
+      "step": 2036
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000138828897359469,
+      "loss": 0.906,
+      "step": 2037
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001387714702310859,
+      "loss": 0.8075,
+      "step": 2038
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013871402805028314,
+      "loss": 0.8169,
+      "step": 2039
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013865657083936164,
+      "loss": 0.3643,
+      "step": 2040
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013859909862062842,
+      "loss": 0.8613,
+      "step": 2041
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001385416114163961,
+      "loss": 0.2872,
+      "step": 2042
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001384841092489832,
+      "loss": 0.8767,
+      "step": 2043
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013842659214071404,
+      "loss": 0.8435,
+      "step": 2044
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013836906011391878,
+      "loss": 0.8325,
+      "step": 2045
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013831151319093323,
+      "loss": 0.8635,
+      "step": 2046
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001382539513940992,
+      "loss": 0.7883,
+      "step": 2047
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001381963747457641,
+      "loss": 0.8596,
+      "step": 2048
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001381387832682812,
+      "loss": 0.7998,
+      "step": 2049
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013808117698400947,
+      "loss": 0.8923,
+      "step": 2050
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013802355591531365,
+      "loss": 0.8313,
+      "step": 2051
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013796592008456427,
+      "loss": 0.8787,
+      "step": 2052
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013790826951413747,
+      "loss": 0.8152,
+      "step": 2053
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013785060422641526,
+      "loss": 0.3541,
+      "step": 2054
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001377929242437852,
+      "loss": 0.8398,
+      "step": 2055
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013773522958864076,
+      "loss": 0.8438,
+      "step": 2056
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001376775202833809,
+      "loss": 0.8464,
+      "step": 2057
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001376197963504104,
+      "loss": 0.8052,
+      "step": 2058
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013756205781213963,
+      "loss": 0.3188,
+      "step": 2059
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013750430469098478,
+      "loss": 0.8257,
+      "step": 2060
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013744653700936753,
+      "loss": 0.8477,
+      "step": 2061
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013738875478971525,
+      "loss": 0.8467,
+      "step": 2062
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013733095805446106,
+      "loss": 0.865,
+      "step": 2063
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001372731468260436,
+      "loss": 0.8259,
+      "step": 2064
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001372153211269072,
+      "loss": 0.8389,
+      "step": 2065
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013715748097950175,
+      "loss": 0.8263,
+      "step": 2066
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013709962640628284,
+      "loss": 0.8425,
+      "step": 2067
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013704175742971158,
+      "loss": 0.8323,
+      "step": 2068
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001369838740722547,
+      "loss": 0.8254,
+      "step": 2069
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001369259763563845,
+      "loss": 0.802,
+      "step": 2070
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001368680643045789,
+      "loss": 0.3277,
+      "step": 2071
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001368101379393213,
+      "loss": 0.3408,
+      "step": 2072
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013675219728310077,
+      "loss": 0.8477,
+      "step": 2073
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013669424235841185,
+      "loss": 0.7795,
+      "step": 2074
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013663627318775457,
+      "loss": 0.8474,
+      "step": 2075
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013657828979363467,
+      "loss": 0.8301,
+      "step": 2076
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013652029219856322,
+      "loss": 0.8076,
+      "step": 2077
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013646228042505694,
+      "loss": 0.7883,
+      "step": 2078
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001364042544956379,
+      "loss": 0.8464,
+      "step": 2079
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013634621443283387,
+      "loss": 0.8713,
+      "step": 2080
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013628816025917798,
+      "loss": 0.8259,
+      "step": 2081
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013623009199720883,
+      "loss": 0.771,
+      "step": 2082
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013617200966947052,
+      "loss": 0.8442,
+      "step": 2083
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013611391329851263,
+      "loss": 0.7952,
+      "step": 2084
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013605580290689014,
+      "loss": 0.8269,
+      "step": 2085
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013599767851716352,
+      "loss": 0.8308,
+      "step": 2086
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013593954015189867,
+      "loss": 0.8374,
+      "step": 2087
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001358813878336669,
+      "loss": 0.8591,
+      "step": 2088
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013582322158504494,
+      "loss": 0.8579,
+      "step": 2089
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013576504142861495,
+      "loss": 0.8311,
+      "step": 2090
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013570684738696445,
+      "loss": 0.8362,
+      "step": 2091
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001356486394826863,
+      "loss": 0.8486,
+      "step": 2092
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013559041773837898,
+      "loss": 0.8787,
+      "step": 2093
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013553218217664603,
+      "loss": 0.8745,
+      "step": 2094
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013547393282009655,
+      "loss": 0.8059,
+      "step": 2095
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013541566969134497,
+      "loss": 0.863,
+      "step": 2096
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000135357392813011,
+      "loss": 0.8326,
+      "step": 2097
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013529910220771977,
+      "loss": 0.7942,
+      "step": 2098
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013524079789810163,
+      "loss": 0.8489,
+      "step": 2099
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013518247990679241,
+      "loss": 0.8789,
+      "step": 2100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013512414825643312,
+      "loss": 0.8511,
+      "step": 2101
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001350658029696701,
+      "loss": 0.7913,
+      "step": 2102
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013500744406915504,
+      "loss": 0.7944,
+      "step": 2103
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013494907157754485,
+      "loss": 0.8029,
+      "step": 2104
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001348906855175017,
+      "loss": 0.9084,
+      "step": 2105
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013483228591169315,
+      "loss": 0.9109,
+      "step": 2106
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001347738727827919,
+      "loss": 0.8046,
+      "step": 2107
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001347154461534759,
+      "loss": 0.8232,
+      "step": 2108
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013465700604642846,
+      "loss": 0.8032,
+      "step": 2109
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000134598552484338,
+      "loss": 0.8744,
+      "step": 2110
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013454008548989817,
+      "loss": 0.8077,
+      "step": 2111
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001344816050858079,
+      "loss": 0.7959,
+      "step": 2112
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013442311129477133,
+      "loss": 0.8459,
+      "step": 2113
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001343646041394977,
+      "loss": 0.356,
+      "step": 2114
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013430608364270157,
+      "loss": 0.8618,
+      "step": 2115
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013424754982710254,
+      "loss": 0.8098,
+      "step": 2116
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001341890027154255,
+      "loss": 0.3601,
+      "step": 2117
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013413044233040043,
+      "loss": 0.8655,
+      "step": 2118
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013407186869476254,
+      "loss": 0.8186,
+      "step": 2119
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013401328183125207,
+      "loss": 0.765,
+      "step": 2120
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001339546817626145,
+      "loss": 0.8118,
+      "step": 2121
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013389606851160036,
+      "loss": 0.8596,
+      "step": 2122
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013383744210096536,
+      "loss": 0.8738,
+      "step": 2123
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013377880255347027,
+      "loss": 0.7893,
+      "step": 2124
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013372014989188099,
+      "loss": 0.8513,
+      "step": 2125
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001336614841389685,
+      "loss": 0.811,
+      "step": 2126
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013360280531750886,
+      "loss": 0.3475,
+      "step": 2127
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013354411345028325,
+      "loss": 0.8712,
+      "step": 2128
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001334854085600778,
+      "loss": 0.7888,
+      "step": 2129
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013342669066968385,
+      "loss": 0.8633,
+      "step": 2130
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001333679598018976,
+      "loss": 0.9006,
+      "step": 2131
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013330921597952056,
+      "loss": 0.8386,
+      "step": 2132
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013325045922535895,
+      "loss": 0.8433,
+      "step": 2133
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013319168956222422,
+      "loss": 0.8684,
+      "step": 2134
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001331329070129328,
+      "loss": 0.8149,
+      "step": 2135
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013307411160030607,
+      "loss": 0.8586,
+      "step": 2136
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013301530334717046,
+      "loss": 0.834,
+      "step": 2137
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001329564822763573,
+      "loss": 0.8557,
+      "step": 2138
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000132897648410703,
+      "loss": 0.8816,
+      "step": 2139
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013283880177304892,
+      "loss": 0.8408,
+      "step": 2140
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001327799423862413,
+      "loss": 0.8125,
+      "step": 2141
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013272107027313142,
+      "loss": 0.8501,
+      "step": 2142
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013266218545657541,
+      "loss": 0.8564,
+      "step": 2143
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001326032879594344,
+      "loss": 0.8462,
+      "step": 2144
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013254437780457447,
+      "loss": 0.748,
+      "step": 2145
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013248545501486653,
+      "loss": 0.8208,
+      "step": 2146
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013242651961318646,
+      "loss": 0.8232,
+      "step": 2147
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000132367571622415,
+      "loss": 0.8503,
+      "step": 2148
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013230861106543776,
+      "loss": 0.9006,
+      "step": 2149
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001322496379651453,
+      "loss": 0.8215,
+      "step": 2150
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000132190652344433,
+      "loss": 0.8484,
+      "step": 2151
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001321316542262011,
+      "loss": 0.7915,
+      "step": 2152
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001320726436333547,
+      "loss": 0.8669,
+      "step": 2153
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013201362058880376,
+      "loss": 0.8206,
+      "step": 2154
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013195458511546307,
+      "loss": 0.8447,
+      "step": 2155
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013189553723625216,
+      "loss": 0.785,
+      "step": 2156
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001318364769740955,
+      "loss": 0.8267,
+      "step": 2157
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013177740435192233,
+      "loss": 0.8566,
+      "step": 2158
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013171831939266668,
+      "loss": 0.8342,
+      "step": 2159
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013165922211926732,
+      "loss": 0.764,
+      "step": 2160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001316001125546679,
+      "loss": 0.8423,
+      "step": 2161
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013154099072181676,
+      "loss": 0.9121,
+      "step": 2162
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013148185664366702,
+      "loss": 0.856,
+      "step": 2163
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001314227103431766,
+      "loss": 0.821,
+      "step": 2164
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001313635518433081,
+      "loss": 0.8135,
+      "step": 2165
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013130438116702888,
+      "loss": 0.8145,
+      "step": 2166
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013124519833731106,
+      "loss": 0.771,
+      "step": 2167
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013118600337713146,
+      "loss": 0.8308,
+      "step": 2168
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013112679630947156,
+      "loss": 0.845,
+      "step": 2169
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013106757715731758,
+      "loss": 0.8391,
+      "step": 2170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001310083459436605,
+      "loss": 0.8054,
+      "step": 2171
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013094910269149587,
+      "loss": 0.7778,
+      "step": 2172
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013088984742382396,
+      "loss": 0.8232,
+      "step": 2173
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001308305801636497,
+      "loss": 0.8362,
+      "step": 2174
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013077130093398274,
+      "loss": 0.8381,
+      "step": 2175
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013071200975783723,
+      "loss": 0.7882,
+      "step": 2176
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013065270665823205,
+      "loss": 0.7905,
+      "step": 2177
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013059339165819083,
+      "loss": 0.8325,
+      "step": 2178
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013053406478074156,
+      "loss": 0.8843,
+      "step": 2179
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013047472604891703,
+      "loss": 0.7971,
+      "step": 2180
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013041537548575457,
+      "loss": 0.866,
+      "step": 2181
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013035601311429612,
+      "loss": 0.8101,
+      "step": 2182
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013029663895758813,
+      "loss": 0.8457,
+      "step": 2183
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001302372530386818,
+      "loss": 0.8511,
+      "step": 2184
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013017785538063276,
+      "loss": 0.8625,
+      "step": 2185
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001301184460065012,
+      "loss": 0.8108,
+      "step": 2186
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001300590249393519,
+      "loss": 0.8615,
+      "step": 2187
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012999959220225414,
+      "loss": 0.8492,
+      "step": 2188
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012994014781828178,
+      "loss": 0.8713,
+      "step": 2189
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012988069181051314,
+      "loss": 0.8193,
+      "step": 2190
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012982122420203114,
+      "loss": 0.8195,
+      "step": 2191
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012976174501592313,
+      "loss": 0.7969,
+      "step": 2192
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012970225427528097,
+      "loss": 0.832,
+      "step": 2193
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012964275200320105,
+      "loss": 0.7812,
+      "step": 2194
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012958323822278412,
+      "loss": 0.8225,
+      "step": 2195
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012952371295713556,
+      "loss": 0.8538,
+      "step": 2196
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012946417622936512,
+      "loss": 0.8354,
+      "step": 2197
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012940462806258695,
+      "loss": 0.8623,
+      "step": 2198
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012934506847991976,
+      "loss": 0.3492,
+      "step": 2199
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001292854975044866,
+      "loss": 0.8308,
+      "step": 2200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012922591515941498,
+      "loss": 0.356,
+      "step": 2201
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012916632146783684,
+      "loss": 0.7644,
+      "step": 2202
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001291067164528884,
+      "loss": 0.8379,
+      "step": 2203
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012904710013771054,
+      "loss": 0.7808,
+      "step": 2204
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012898747254544825,
+      "loss": 0.8037,
+      "step": 2205
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012892783369925105,
+      "loss": 0.8523,
+      "step": 2206
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012886818362227282,
+      "loss": 0.8181,
+      "step": 2207
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012880852233767174,
+      "loss": 0.338,
+      "step": 2208
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001287488498686104,
+      "loss": 0.8057,
+      "step": 2209
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001286891662382556,
+      "loss": 0.812,
+      "step": 2210
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012862947146977875,
+      "loss": 0.8101,
+      "step": 2211
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001285697655863553,
+      "loss": 0.3361,
+      "step": 2212
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012851004861116517,
+      "loss": 0.8386,
+      "step": 2213
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012845032056739257,
+      "loss": 0.77,
+      "step": 2214
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012839058147822593,
+      "loss": 0.8269,
+      "step": 2215
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012833083136685802,
+      "loss": 0.8159,
+      "step": 2216
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012827107025648594,
+      "loss": 0.7803,
+      "step": 2217
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000128211298170311,
+      "loss": 0.8,
+      "step": 2218
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012815151513153873,
+      "loss": 0.8538,
+      "step": 2219
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012809172116337903,
+      "loss": 0.8872,
+      "step": 2220
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012803191628904593,
+      "loss": 0.8398,
+      "step": 2221
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001279721005317578,
+      "loss": 0.854,
+      "step": 2222
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012791227391473705,
+      "loss": 0.3381,
+      "step": 2223
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001278524364612106,
+      "loss": 0.8303,
+      "step": 2224
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001277925881944093,
+      "loss": 0.8782,
+      "step": 2225
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012773272913756833,
+      "loss": 0.8059,
+      "step": 2226
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012767285931392704,
+      "loss": 0.8154,
+      "step": 2227
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000127612978746729,
+      "loss": 0.8108,
+      "step": 2228
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001275530874592218,
+      "loss": 0.8472,
+      "step": 2229
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001274931854746574,
+      "loss": 0.7534,
+      "step": 2230
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001274332728162918,
+      "loss": 0.7935,
+      "step": 2231
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001273733495073851,
+      "loss": 0.811,
+      "step": 2232
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012731341557120168,
+      "loss": 0.8245,
+      "step": 2233
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012725347103100987,
+      "loss": 0.8118,
+      "step": 2234
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001271935159100823,
+      "loss": 0.816,
+      "step": 2235
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012713355023169547,
+      "loss": 0.8669,
+      "step": 2236
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012707357401913022,
+      "loss": 0.8865,
+      "step": 2237
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001270135872956714,
+      "loss": 0.8262,
+      "step": 2238
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012695359008460785,
+      "loss": 0.8726,
+      "step": 2239
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012689358240923262,
+      "loss": 0.7993,
+      "step": 2240
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012683356429284273,
+      "loss": 0.8156,
+      "step": 2241
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012677353575873924,
+      "loss": 0.8323,
+      "step": 2242
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012671349683022735,
+      "loss": 0.7817,
+      "step": 2243
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001266534475306162,
+      "loss": 0.8125,
+      "step": 2244
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012659338788321905,
+      "loss": 0.8052,
+      "step": 2245
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012653331791135308,
+      "loss": 0.7941,
+      "step": 2246
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001264732376383395,
+      "loss": 0.7827,
+      "step": 2247
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001264131470875036,
+      "loss": 0.8137,
+      "step": 2248
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012635304628217452,
+      "loss": 0.8502,
+      "step": 2249
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012629293524568555,
+      "loss": 0.7737,
+      "step": 2250
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012623281400137384,
+      "loss": 0.8862,
+      "step": 2251
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001261726825725805,
+      "loss": 0.8235,
+      "step": 2252
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001261125409826506,
+      "loss": 0.3362,
+      "step": 2253
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012605238925493325,
+      "loss": 0.7839,
+      "step": 2254
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012599222741278135,
+      "loss": 0.8512,
+      "step": 2255
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012593205547955185,
+      "loss": 0.8125,
+      "step": 2256
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012587187347860553,
+      "loss": 0.8264,
+      "step": 2257
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012581168143330714,
+      "loss": 0.8851,
+      "step": 2258
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001257514793670253,
+      "loss": 0.8145,
+      "step": 2259
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012569126730313255,
+      "loss": 0.8879,
+      "step": 2260
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012563104526500522,
+      "loss": 0.8254,
+      "step": 2261
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001255708132760236,
+      "loss": 0.8704,
+      "step": 2262
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012551057135957186,
+      "loss": 0.853,
+      "step": 2263
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012545031953903796,
+      "loss": 0.3281,
+      "step": 2264
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001253900578378137,
+      "loss": 0.8435,
+      "step": 2265
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012532978627929485,
+      "loss": 0.8551,
+      "step": 2266
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012526950488688082,
+      "loss": 0.772,
+      "step": 2267
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001252092136839749,
+      "loss": 0.8284,
+      "step": 2268
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012514891269398427,
+      "loss": 0.8784,
+      "step": 2269
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012508860194031986,
+      "loss": 0.8181,
+      "step": 2270
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012502828144639628,
+      "loss": 0.8357,
+      "step": 2271
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001249679512356322,
+      "loss": 0.8668,
+      "step": 2272
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001249076113314497,
+      "loss": 0.8582,
+      "step": 2273
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001248472617572749,
+      "loss": 0.8809,
+      "step": 2274
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012478690253653755,
+      "loss": 0.7847,
+      "step": 2275
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012472653369267122,
+      "loss": 0.7952,
+      "step": 2276
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012466615524911315,
+      "loss": 0.8081,
+      "step": 2277
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001246057672293043,
+      "loss": 0.8181,
+      "step": 2278
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012454536965668948,
+      "loss": 0.8035,
+      "step": 2279
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000124484962554717,
+      "loss": 0.8596,
+      "step": 2280
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000124424545946839,
+      "loss": 0.8396,
+      "step": 2281
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012436411985651132,
+      "loss": 0.8373,
+      "step": 2282
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001243036843071934,
+      "loss": 0.9062,
+      "step": 2283
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001242432393223485,
+      "loss": 0.8511,
+      "step": 2284
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001241827849254433,
+      "loss": 0.7755,
+      "step": 2285
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001241223211399484,
+      "loss": 0.8137,
+      "step": 2286
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012406184798933786,
+      "loss": 0.8975,
+      "step": 2287
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012400136549708946,
+      "loss": 0.8403,
+      "step": 2288
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012394087368668459,
+      "loss": 0.8345,
+      "step": 2289
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012388037258160824,
+      "loss": 0.8267,
+      "step": 2290
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000123819862205349,
+      "loss": 0.8129,
+      "step": 2291
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012375934258139915,
+      "loss": 0.8308,
+      "step": 2292
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012369881373325447,
+      "loss": 0.7938,
+      "step": 2293
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001236382756844143,
+      "loss": 0.8451,
+      "step": 2294
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012357772845838157,
+      "loss": 0.822,
+      "step": 2295
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012351717207866291,
+      "loss": 0.762,
+      "step": 2296
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001234566065687683,
+      "loss": 0.8225,
+      "step": 2297
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001233960319522114,
+      "loss": 0.798,
+      "step": 2298
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012333544825250938,
+      "loss": 0.3303,
+      "step": 2299
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012327485549318284,
+      "loss": 0.8894,
+      "step": 2300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012321425369775602,
+      "loss": 0.8345,
+      "step": 2301
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012315364288975663,
+      "loss": 0.865,
+      "step": 2302
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012309302309271587,
+      "loss": 0.8765,
+      "step": 2303
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012303239433016842,
+      "loss": 0.8818,
+      "step": 2304
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012297175662565247,
+      "loss": 0.8003,
+      "step": 2305
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001229111100027097,
+      "loss": 0.8574,
+      "step": 2306
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012285045448488511,
+      "loss": 0.8484,
+      "step": 2307
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012278979009572736,
+      "loss": 0.8445,
+      "step": 2308
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001227291168587884,
+      "loss": 0.7954,
+      "step": 2309
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001226684347976237,
+      "loss": 0.7924,
+      "step": 2310
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001226077439357921,
+      "loss": 0.8127,
+      "step": 2311
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012254704429685593,
+      "loss": 0.7769,
+      "step": 2312
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012248633590438083,
+      "loss": 0.7953,
+      "step": 2313
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012242561878193588,
+      "loss": 0.8606,
+      "step": 2314
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012236489295309362,
+      "loss": 0.8247,
+      "step": 2315
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012230415844142984,
+      "loss": 0.3474,
+      "step": 2316
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001222434152705238,
+      "loss": 0.8174,
+      "step": 2317
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001221826634639581,
+      "loss": 0.8542,
+      "step": 2318
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001221219030453187,
+      "loss": 0.8608,
+      "step": 2319
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012206113403819479,
+      "loss": 0.8384,
+      "step": 2320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012200035646617911,
+      "loss": 0.8625,
+      "step": 2321
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012193957035286755,
+      "loss": 0.8345,
+      "step": 2322
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012187877572185937,
+      "loss": 0.8115,
+      "step": 2323
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012181797259675714,
+      "loss": 0.812,
+      "step": 2324
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012175716100116676,
+      "loss": 0.897,
+      "step": 2325
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012169634095869735,
+      "loss": 0.8643,
+      "step": 2326
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012163551249296133,
+      "loss": 0.8574,
+      "step": 2327
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012157467562757443,
+      "loss": 0.7834,
+      "step": 2328
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012151383038615563,
+      "loss": 0.7854,
+      "step": 2329
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001214529767923271,
+      "loss": 0.8157,
+      "step": 2330
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012139211486971436,
+      "loss": 0.8254,
+      "step": 2331
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001213312446419461,
+      "loss": 0.8665,
+      "step": 2332
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012127036613265417,
+      "loss": 0.7888,
+      "step": 2333
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012120947936547376,
+      "loss": 0.8159,
+      "step": 2334
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012114858436404321,
+      "loss": 0.8103,
+      "step": 2335
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012108768115200404,
+      "loss": 0.811,
+      "step": 2336
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012102676975300094,
+      "loss": 0.2989,
+      "step": 2337
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001209658501906819,
+      "loss": 0.8059,
+      "step": 2338
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012090492248869793,
+      "loss": 0.7126,
+      "step": 2339
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012084398667070325,
+      "loss": 0.8323,
+      "step": 2340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012078304276035526,
+      "loss": 0.7893,
+      "step": 2341
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012072209078131451,
+      "loss": 0.8721,
+      "step": 2342
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001206611307572446,
+      "loss": 0.3482,
+      "step": 2343
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001206001627118124,
+      "loss": 0.855,
+      "step": 2344
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012053918666868776,
+      "loss": 0.8682,
+      "step": 2345
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012047820265154361,
+      "loss": 0.8423,
+      "step": 2346
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012041721068405613,
+      "loss": 0.848,
+      "step": 2347
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012035621078990449,
+      "loss": 0.8633,
+      "step": 2348
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012029520299277095,
+      "loss": 0.78,
+      "step": 2349
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012023418731634077,
+      "loss": 0.8682,
+      "step": 2350
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012017316378430243,
+      "loss": 0.8069,
+      "step": 2351
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012011213242034733,
+      "loss": 0.8147,
+      "step": 2352
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012005109324816992,
+      "loss": 0.814,
+      "step": 2353
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011999004629146775,
+      "loss": 0.8196,
+      "step": 2354
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011992899157394133,
+      "loss": 0.8032,
+      "step": 2355
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011986792911929419,
+      "loss": 0.8557,
+      "step": 2356
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011980685895123289,
+      "loss": 0.8264,
+      "step": 2357
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011974578109346702,
+      "loss": 0.8264,
+      "step": 2358
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011968469556970905,
+      "loss": 0.7554,
+      "step": 2359
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011962360240367445,
+      "loss": 0.8354,
+      "step": 2360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011956250161908178,
+      "loss": 0.8242,
+      "step": 2361
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001195013932396524,
+      "loss": 0.8867,
+      "step": 2362
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011944027728911071,
+      "loss": 0.8806,
+      "step": 2363
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011937915379118405,
+      "loss": 0.3563,
+      "step": 2364
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011931802276960265,
+      "loss": 0.7782,
+      "step": 2365
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011925688424809964,
+      "loss": 0.8486,
+      "step": 2366
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011919573825041115,
+      "loss": 0.8376,
+      "step": 2367
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011913458480027613,
+      "loss": 0.7607,
+      "step": 2368
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011907342392143645,
+      "loss": 0.7949,
+      "step": 2369
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011901225563763693,
+      "loss": 0.8035,
+      "step": 2370
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011895107997262515,
+      "loss": 0.8345,
+      "step": 2371
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011888989695015165,
+      "loss": 0.79,
+      "step": 2372
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011882870659396968,
+      "loss": 0.7781,
+      "step": 2373
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011876750892783557,
+      "loss": 0.8079,
+      "step": 2374
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011870630397550831,
+      "loss": 0.8474,
+      "step": 2375
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011864509176074974,
+      "loss": 0.7903,
+      "step": 2376
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001185838723073246,
+      "loss": 0.8079,
+      "step": 2377
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011852264563900038,
+      "loss": 0.8109,
+      "step": 2378
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011846141177954733,
+      "loss": 0.3349,
+      "step": 2379
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011840017075273861,
+      "loss": 0.8528,
+      "step": 2380
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011833892258235007,
+      "loss": 0.8323,
+      "step": 2381
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011827766729216035,
+      "loss": 0.8184,
+      "step": 2382
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011821640490595084,
+      "loss": 0.8516,
+      "step": 2383
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011815513544750579,
+      "loss": 0.3232,
+      "step": 2384
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011809385894061205,
+      "loss": 0.8044,
+      "step": 2385
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011803257540905925,
+      "loss": 0.8296,
+      "step": 2386
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011797128487663981,
+      "loss": 0.813,
+      "step": 2387
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011790998736714883,
+      "loss": 0.7776,
+      "step": 2388
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011784868290438404,
+      "loss": 0.8237,
+      "step": 2389
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011778737151214605,
+      "loss": 0.7983,
+      "step": 2390
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011772605321423799,
+      "loss": 0.8411,
+      "step": 2391
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011766472803446575,
+      "loss": 0.7719,
+      "step": 2392
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011760339599663787,
+      "loss": 0.8442,
+      "step": 2393
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011754205712456557,
+      "loss": 0.7798,
+      "step": 2394
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011748071144206266,
+      "loss": 0.7862,
+      "step": 2395
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011741935897294571,
+      "loss": 0.8088,
+      "step": 2396
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011735799974103388,
+      "loss": 0.3134,
+      "step": 2397
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011729663377014888,
+      "loss": 0.8594,
+      "step": 2398
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011723526108411509,
+      "loss": 0.7874,
+      "step": 2399
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011717388170675954,
+      "loss": 0.8596,
+      "step": 2400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011711249566191178,
+      "loss": 0.8257,
+      "step": 2401
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011705110297340397,
+      "loss": 0.7814,
+      "step": 2402
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011698970366507096,
+      "loss": 0.8667,
+      "step": 2403
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011692829776074998,
+      "loss": 0.7827,
+      "step": 2404
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011686688528428098,
+      "loss": 0.8011,
+      "step": 2405
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011680546625950635,
+      "loss": 0.8285,
+      "step": 2406
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001167440407102711,
+      "loss": 0.798,
+      "step": 2407
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011668260866042271,
+      "loss": 0.8384,
+      "step": 2408
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011662117013381126,
+      "loss": 0.8033,
+      "step": 2409
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011655972515428928,
+      "loss": 0.8806,
+      "step": 2410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001164982737457118,
+      "loss": 0.7687,
+      "step": 2411
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011643681593193643,
+      "loss": 0.3312,
+      "step": 2412
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011637535173682316,
+      "loss": 0.8239,
+      "step": 2413
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011631388118423456,
+      "loss": 0.8361,
+      "step": 2414
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011625240429803552,
+      "loss": 0.7681,
+      "step": 2415
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001161909211020936,
+      "loss": 0.8328,
+      "step": 2416
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011612943162027863,
+      "loss": 0.8726,
+      "step": 2417
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011606793587646294,
+      "loss": 0.79,
+      "step": 2418
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001160064338945213,
+      "loss": 0.795,
+      "step": 2419
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011594492569833094,
+      "loss": 0.8362,
+      "step": 2420
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011588341131177136,
+      "loss": 0.8486,
+      "step": 2421
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011582189075872468,
+      "loss": 0.824,
+      "step": 2422
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011576036406307522,
+      "loss": 0.8033,
+      "step": 2423
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011569883124870979,
+      "loss": 0.7979,
+      "step": 2424
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011563729233951756,
+      "loss": 0.8228,
+      "step": 2425
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011557574735939004,
+      "loss": 0.8362,
+      "step": 2426
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011551419633222108,
+      "loss": 0.8152,
+      "step": 2427
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011545263928190693,
+      "loss": 0.8596,
+      "step": 2428
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011539107623234617,
+      "loss": 0.8561,
+      "step": 2429
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001153295072074397,
+      "loss": 0.3407,
+      "step": 2430
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011526793223109071,
+      "loss": 0.8318,
+      "step": 2431
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011520635132720474,
+      "loss": 0.8318,
+      "step": 2432
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001151447645196896,
+      "loss": 0.8555,
+      "step": 2433
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011508317183245545,
+      "loss": 0.8008,
+      "step": 2434
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011502157328941465,
+      "loss": 0.8848,
+      "step": 2435
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011495996891448189,
+      "loss": 0.7833,
+      "step": 2436
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011489835873157413,
+      "loss": 0.8135,
+      "step": 2437
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011483674276461052,
+      "loss": 0.7982,
+      "step": 2438
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011477512103751253,
+      "loss": 0.8025,
+      "step": 2439
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011471349357420384,
+      "loss": 0.8105,
+      "step": 2440
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011465186039861033,
+      "loss": 0.3341,
+      "step": 2441
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011459022153466015,
+      "loss": 0.8247,
+      "step": 2442
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011452857700628361,
+      "loss": 0.8044,
+      "step": 2443
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011446692683741325,
+      "loss": 0.7686,
+      "step": 2444
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011440527105198378,
+      "loss": 0.8533,
+      "step": 2445
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001143436096739321,
+      "loss": 0.8457,
+      "step": 2446
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011428194272719728,
+      "loss": 0.8004,
+      "step": 2447
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011422027023572051,
+      "loss": 0.3713,
+      "step": 2448
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011415859222344525,
+      "loss": 0.8384,
+      "step": 2449
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000114096908714317,
+      "loss": 0.8464,
+      "step": 2450
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011403521973228341,
+      "loss": 0.8406,
+      "step": 2451
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011397352530129429,
+      "loss": 0.7727,
+      "step": 2452
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001139118254453015,
+      "loss": 0.8774,
+      "step": 2453
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011385012018825907,
+      "loss": 0.761,
+      "step": 2454
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011378840955412313,
+      "loss": 0.7937,
+      "step": 2455
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011372669356685183,
+      "loss": 0.8191,
+      "step": 2456
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011366497225040548,
+      "loss": 0.8774,
+      "step": 2457
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011360324562874643,
+      "loss": 0.7606,
+      "step": 2458
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011354151372583901,
+      "loss": 0.863,
+      "step": 2459
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011347977656564975,
+      "loss": 0.3307,
+      "step": 2460
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011341803417214705,
+      "loss": 0.8547,
+      "step": 2461
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011335628656930152,
+      "loss": 0.7856,
+      "step": 2462
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001132945337810857,
+      "loss": 0.79,
+      "step": 2463
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011323277583147409,
+      "loss": 0.7687,
+      "step": 2464
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011317101274444328,
+      "loss": 0.8259,
+      "step": 2465
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011310924454397187,
+      "loss": 0.7782,
+      "step": 2466
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011304747125404031,
+      "loss": 0.8643,
+      "step": 2467
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011298569289863119,
+      "loss": 0.8181,
+      "step": 2468
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011292390950172899,
+      "loss": 0.7871,
+      "step": 2469
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011286212108732015,
+      "loss": 0.8218,
+      "step": 2470
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011280032767939301,
+      "loss": 0.8467,
+      "step": 2471
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011273852930193798,
+      "loss": 0.7996,
+      "step": 2472
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011267672597894725,
+      "loss": 0.8491,
+      "step": 2473
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000112614917734415,
+      "loss": 0.8638,
+      "step": 2474
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011255310459233737,
+      "loss": 0.7771,
+      "step": 2475
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011249128657671233,
+      "loss": 0.8394,
+      "step": 2476
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011242946371153972,
+      "loss": 0.8069,
+      "step": 2477
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011236763602082137,
+      "loss": 0.7932,
+      "step": 2478
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011230580352856088,
+      "loss": 0.8177,
+      "step": 2479
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011224396625876375,
+      "loss": 0.7819,
+      "step": 2480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011218212423543734,
+      "loss": 0.7925,
+      "step": 2481
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011212027748259086,
+      "loss": 0.8472,
+      "step": 2482
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011205842602423537,
+      "loss": 0.7493,
+      "step": 2483
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011199656988438372,
+      "loss": 0.7264,
+      "step": 2484
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011193470908705055,
+      "loss": 0.7551,
+      "step": 2485
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011187284365625242,
+      "loss": 0.894,
+      "step": 2486
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011181097361600755,
+      "loss": 0.7778,
+      "step": 2487
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011174909899033607,
+      "loss": 0.8488,
+      "step": 2488
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011168721980325987,
+      "loss": 0.8777,
+      "step": 2489
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001116253360788025,
+      "loss": 0.7961,
+      "step": 2490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011156344784098942,
+      "loss": 0.8284,
+      "step": 2491
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011150155511384773,
+      "loss": 0.8318,
+      "step": 2492
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011143965792140632,
+      "loss": 0.8018,
+      "step": 2493
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011137775628769584,
+      "loss": 0.8188,
+      "step": 2494
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011131585023674861,
+      "loss": 0.8406,
+      "step": 2495
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011125393979259872,
+      "loss": 0.8567,
+      "step": 2496
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011119202497928192,
+      "loss": 0.8391,
+      "step": 2497
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011113010582083567,
+      "loss": 0.7773,
+      "step": 2498
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011106818234129912,
+      "loss": 0.8558,
+      "step": 2499
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011100625456471307,
+      "loss": 0.7842,
+      "step": 2500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011094432251512005,
+      "loss": 0.8347,
+      "step": 2501
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011088238621656421,
+      "loss": 0.7705,
+      "step": 2502
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011082044569309137,
+      "loss": 0.8345,
+      "step": 2503
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011075850096874893,
+      "loss": 0.7283,
+      "step": 2504
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011069655206758602,
+      "loss": 0.8196,
+      "step": 2505
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011063459901365325,
+      "loss": 0.814,
+      "step": 2506
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011057264183100302,
+      "loss": 0.3365,
+      "step": 2507
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001105106805436892,
+      "loss": 0.8115,
+      "step": 2508
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011044871517576729,
+      "loss": 0.7693,
+      "step": 2509
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011038674575129441,
+      "loss": 0.8875,
+      "step": 2510
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001103247722943292,
+      "loss": 0.8281,
+      "step": 2511
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011026279482893187,
+      "loss": 0.8867,
+      "step": 2512
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011020081337916424,
+      "loss": 0.8523,
+      "step": 2513
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011013882796908963,
+      "loss": 0.8354,
+      "step": 2514
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011007683862277292,
+      "loss": 0.823,
+      "step": 2515
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011001484536428051,
+      "loss": 0.8479,
+      "step": 2516
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010995284821768028,
+      "loss": 0.8055,
+      "step": 2517
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001098908472070417,
+      "loss": 0.7622,
+      "step": 2518
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010982884235643567,
+      "loss": 0.8364,
+      "step": 2519
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010976683368993464,
+      "loss": 0.884,
+      "step": 2520
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010970482123161248,
+      "loss": 0.8236,
+      "step": 2521
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010964280500554458,
+      "loss": 0.7214,
+      "step": 2522
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010958078503580775,
+      "loss": 0.8367,
+      "step": 2523
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010951876134648032,
+      "loss": 0.7404,
+      "step": 2524
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010945673396164198,
+      "loss": 0.8162,
+      "step": 2525
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010939470290537388,
+      "loss": 0.7957,
+      "step": 2526
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010933266820175867,
+      "loss": 0.8718,
+      "step": 2527
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010927062987488035,
+      "loss": 0.8511,
+      "step": 2528
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010920858794882429,
+      "loss": 0.8679,
+      "step": 2529
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010914654244767734,
+      "loss": 0.8728,
+      "step": 2530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010908449339552768,
+      "loss": 0.8225,
+      "step": 2531
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010902244081646489,
+      "loss": 0.8521,
+      "step": 2532
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010896038473457993,
+      "loss": 0.8345,
+      "step": 2533
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010889832517396509,
+      "loss": 0.8438,
+      "step": 2534
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010883626215871408,
+      "loss": 0.8638,
+      "step": 2535
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010877419571292182,
+      "loss": 0.8455,
+      "step": 2536
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010871212586068469,
+      "loss": 0.7852,
+      "step": 2537
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010865005262610033,
+      "loss": 0.7971,
+      "step": 2538
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010858797603326769,
+      "loss": 0.8477,
+      "step": 2539
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001085258961062871,
+      "loss": 0.8047,
+      "step": 2540
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010846381286926006,
+      "loss": 0.7844,
+      "step": 2541
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010840172634628946,
+      "loss": 0.8625,
+      "step": 2542
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010833963656147943,
+      "loss": 0.8186,
+      "step": 2543
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010827754353893532,
+      "loss": 0.8452,
+      "step": 2544
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010821544730276378,
+      "loss": 0.8789,
+      "step": 2545
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010815334787707276,
+      "loss": 0.8513,
+      "step": 2546
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010809124528597139,
+      "loss": 0.8369,
+      "step": 2547
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010802913955356997,
+      "loss": 0.7983,
+      "step": 2548
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010796703070398015,
+      "loss": 0.834,
+      "step": 2549
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001079049187613147,
+      "loss": 0.7771,
+      "step": 2550
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001078428037496876,
+      "loss": 0.8281,
+      "step": 2551
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010778068569321403,
+      "loss": 0.821,
+      "step": 2552
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001077185646160104,
+      "loss": 0.845,
+      "step": 2553
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010765644054219421,
+      "loss": 0.7493,
+      "step": 2554
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001075943134958842,
+      "loss": 0.8445,
+      "step": 2555
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010753218350120022,
+      "loss": 0.8435,
+      "step": 2556
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010747005058226324,
+      "loss": 0.8306,
+      "step": 2557
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010740791476319543,
+      "loss": 0.3339,
+      "step": 2558
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010734577606812007,
+      "loss": 0.834,
+      "step": 2559
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010728363452116148,
+      "loss": 0.7964,
+      "step": 2560
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010722149014644522,
+      "loss": 0.844,
+      "step": 2561
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010715934296809782,
+      "loss": 0.791,
+      "step": 2562
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000107097193010247,
+      "loss": 0.821,
+      "step": 2563
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010703504029702148,
+      "loss": 0.7301,
+      "step": 2564
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010697288485255106,
+      "loss": 0.8291,
+      "step": 2565
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010691072670096669,
+      "loss": 0.7717,
+      "step": 2566
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010684856586640026,
+      "loss": 0.7856,
+      "step": 2567
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010678640237298476,
+      "loss": 0.8787,
+      "step": 2568
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010672423624485421,
+      "loss": 0.8262,
+      "step": 2569
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010666206750614362,
+      "loss": 0.8811,
+      "step": 2570
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010659989618098905,
+      "loss": 0.8323,
+      "step": 2571
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010653772229352749,
+      "loss": 0.8091,
+      "step": 2572
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010647554586789708,
+      "loss": 0.833,
+      "step": 2573
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001064133669282368,
+      "loss": 0.8022,
+      "step": 2574
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010635118549868668,
+      "loss": 0.8074,
+      "step": 2575
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010628900160338764,
+      "loss": 0.8025,
+      "step": 2576
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010622681526648167,
+      "loss": 0.7849,
+      "step": 2577
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010616462651211157,
+      "loss": 0.8811,
+      "step": 2578
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010610243536442124,
+      "loss": 0.8176,
+      "step": 2579
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010604024184755539,
+      "loss": 0.8003,
+      "step": 2580
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010597804598565968,
+      "loss": 0.761,
+      "step": 2581
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010591584780288068,
+      "loss": 0.7692,
+      "step": 2582
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010585364732336587,
+      "loss": 0.8107,
+      "step": 2583
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010579144457126364,
+      "loss": 0.7734,
+      "step": 2584
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010572923957072321,
+      "loss": 0.8367,
+      "step": 2585
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001056670323458947,
+      "loss": 0.8276,
+      "step": 2586
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010560482292092913,
+      "loss": 0.835,
+      "step": 2587
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010554261131997834,
+      "loss": 0.7952,
+      "step": 2588
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010548039756719497,
+      "loss": 0.8311,
+      "step": 2589
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001054181816867326,
+      "loss": 0.7817,
+      "step": 2590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001053559637027455,
+      "loss": 0.7791,
+      "step": 2591
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010529374363938889,
+      "loss": 0.8033,
+      "step": 2592
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010523152152081874,
+      "loss": 0.8853,
+      "step": 2593
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010516929737119181,
+      "loss": 0.7621,
+      "step": 2594
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010510707121466567,
+      "loss": 0.8425,
+      "step": 2595
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010504484307539863,
+      "loss": 0.8586,
+      "step": 2596
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010498261297754983,
+      "loss": 0.8625,
+      "step": 2597
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010492038094527907,
+      "loss": 0.7994,
+      "step": 2598
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010485814700274706,
+      "loss": 0.8064,
+      "step": 2599
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001047959111741151,
+      "loss": 0.771,
+      "step": 2600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010473367348354527,
+      "loss": 0.8345,
+      "step": 2601
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010467143395520043,
+      "loss": 0.8328,
+      "step": 2602
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001046091926132441,
+      "loss": 0.3247,
+      "step": 2603
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010454694948184045,
+      "loss": 0.8909,
+      "step": 2604
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010448470458515449,
+      "loss": 0.8113,
+      "step": 2605
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010442245794735179,
+      "loss": 0.7842,
+      "step": 2606
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010436020959259862,
+      "loss": 0.8418,
+      "step": 2607
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010429795954506203,
+      "loss": 0.8367,
+      "step": 2608
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001042357078289095,
+      "loss": 0.8154,
+      "step": 2609
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010417345446830937,
+      "loss": 0.8984,
+      "step": 2610
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010411119948743051,
+      "loss": 0.7295,
+      "step": 2611
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010404894291044246,
+      "loss": 0.8025,
+      "step": 2612
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010398668476151538,
+      "loss": 0.8076,
+      "step": 2613
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010392442506481999,
+      "loss": 0.7653,
+      "step": 2614
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001038621638445277,
+      "loss": 0.8193,
+      "step": 2615
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010379990112481039,
+      "loss": 0.3253,
+      "step": 2616
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010373763692984062,
+      "loss": 0.8513,
+      "step": 2617
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010367537128379154,
+      "loss": 0.7816,
+      "step": 2618
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010361310421083676,
+      "loss": 0.8801,
+      "step": 2619
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010355083573515051,
+      "loss": 0.8921,
+      "step": 2620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010348856588090763,
+      "loss": 0.7732,
+      "step": 2621
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010342629467228331,
+      "loss": 0.7358,
+      "step": 2622
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010336402213345344,
+      "loss": 0.8057,
+      "step": 2623
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010330174828859433,
+      "loss": 0.834,
+      "step": 2624
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010323947316188288,
+      "loss": 0.8157,
+      "step": 2625
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001031771967774964,
+      "loss": 0.8162,
+      "step": 2626
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010311491915961271,
+      "loss": 0.8157,
+      "step": 2627
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001030526403324102,
+      "loss": 0.338,
+      "step": 2628
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001029903603200676,
+      "loss": 0.7952,
+      "step": 2629
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010292807914676412,
+      "loss": 0.7974,
+      "step": 2630
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010286579683667952,
+      "loss": 0.3013,
+      "step": 2631
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010280351341399392,
+      "loss": 0.764,
+      "step": 2632
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001027412289028879,
+      "loss": 0.8623,
+      "step": 2633
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010267894332754242,
+      "loss": 0.7902,
+      "step": 2634
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010261665671213891,
+      "loss": 0.7738,
+      "step": 2635
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010255436908085919,
+      "loss": 0.8035,
+      "step": 2636
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010249208045788538,
+      "loss": 0.7822,
+      "step": 2637
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010242979086740019,
+      "loss": 0.8519,
+      "step": 2638
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010236750033358648,
+      "loss": 0.8044,
+      "step": 2639
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010230520888062764,
+      "loss": 0.8057,
+      "step": 2640
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010224291653270738,
+      "loss": 0.7749,
+      "step": 2641
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010218062331400969,
+      "loss": 0.8982,
+      "step": 2642
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010211832924871889,
+      "loss": 0.8176,
+      "step": 2643
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010205603436101978,
+      "loss": 0.8788,
+      "step": 2644
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010199373867509734,
+      "loss": 0.7942,
+      "step": 2645
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001019314422151369,
+      "loss": 0.8096,
+      "step": 2646
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010186914500532407,
+      "loss": 0.7729,
+      "step": 2647
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010180684706984483,
+      "loss": 0.8469,
+      "step": 2648
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010174454843288532,
+      "loss": 0.781,
+      "step": 2649
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010168224911863205,
+      "loss": 0.793,
+      "step": 2650
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010161994915127173,
+      "loss": 0.7957,
+      "step": 2651
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010155764855499139,
+      "loss": 0.8037,
+      "step": 2652
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010149534735397823,
+      "loss": 0.8071,
+      "step": 2653
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001014330455724198,
+      "loss": 0.8284,
+      "step": 2654
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010137074323450371,
+      "loss": 0.8164,
+      "step": 2655
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010130844036441786,
+      "loss": 0.311,
+      "step": 2656
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010124613698635043,
+      "loss": 0.7266,
+      "step": 2657
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010118383312448974,
+      "loss": 0.8489,
+      "step": 2658
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010112152880302426,
+      "loss": 0.7993,
+      "step": 2659
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010105922404614265,
+      "loss": 0.3258,
+      "step": 2660
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010099691887803384,
+      "loss": 0.8435,
+      "step": 2661
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010093461332288679,
+      "loss": 0.7903,
+      "step": 2662
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010087230740489064,
+      "loss": 0.3539,
+      "step": 2663
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010081000114823474,
+      "loss": 0.8025,
+      "step": 2664
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001007476945771085,
+      "loss": 0.7487,
+      "step": 2665
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010068538771570148,
+      "loss": 0.8494,
+      "step": 2666
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010062308058820335,
+      "loss": 0.7961,
+      "step": 2667
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010056077321880393,
+      "loss": 0.8635,
+      "step": 2668
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010049846563169296,
+      "loss": 0.7915,
+      "step": 2669
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010043615785106051,
+      "loss": 0.833,
+      "step": 2670
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010037384990109657,
+      "loss": 0.7996,
+      "step": 2671
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010031154180599122,
+      "loss": 0.8445,
+      "step": 2672
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001002492335899346,
+      "loss": 0.345,
+      "step": 2673
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010018692527711695,
+      "loss": 0.7868,
+      "step": 2674
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010012461689172845,
+      "loss": 0.812,
+      "step": 2675
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010006230845795936,
+      "loss": 0.8433,
+      "step": 2676
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001,
+      "loss": 0.8665,
+      "step": 2677
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.993769154204063e-05,
+      "loss": 0.7955,
+      "step": 2678
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.987538310827159e-05,
+      "loss": 0.8311,
+      "step": 2679
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.981307472288309e-05,
+      "loss": 0.7637,
+      "step": 2680
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.975076641006542e-05,
+      "loss": 0.793,
+      "step": 2681
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.968845819400883e-05,
+      "loss": 0.8843,
+      "step": 2682
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.962615009890346e-05,
+      "loss": 0.7678,
+      "step": 2683
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.956384214893949e-05,
+      "loss": 0.9084,
+      "step": 2684
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.950153436830706e-05,
+      "loss": 0.8079,
+      "step": 2685
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.94392267811961e-05,
+      "loss": 0.8296,
+      "step": 2686
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.937691941179665e-05,
+      "loss": 0.8564,
+      "step": 2687
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.931461228429854e-05,
+      "loss": 0.793,
+      "step": 2688
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.925230542289152e-05,
+      "loss": 0.8406,
+      "step": 2689
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.91899988517653e-05,
+      "loss": 0.801,
+      "step": 2690
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.912769259510938e-05,
+      "loss": 0.8547,
+      "step": 2691
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.906538667711322e-05,
+      "loss": 0.7979,
+      "step": 2692
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.900308112196619e-05,
+      "loss": 0.8337,
+      "step": 2693
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.894077595385737e-05,
+      "loss": 0.8286,
+      "step": 2694
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.887847119697578e-05,
+      "loss": 0.7719,
+      "step": 2695
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.88161668755103e-05,
+      "loss": 0.8104,
+      "step": 2696
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.875386301364958e-05,
+      "loss": 0.7427,
+      "step": 2697
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.869155963558215e-05,
+      "loss": 0.7124,
+      "step": 2698
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.862925676549635e-05,
+      "loss": 0.8372,
+      "step": 2699
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.856695442758023e-05,
+      "loss": 0.8113,
+      "step": 2700
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.850465264602176e-05,
+      "loss": 0.866,
+      "step": 2701
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.844235144500864e-05,
+      "loss": 0.8547,
+      "step": 2702
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.83800508487283e-05,
+      "loss": 0.8108,
+      "step": 2703
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.831775088136796e-05,
+      "loss": 0.7964,
+      "step": 2704
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.825545156711472e-05,
+      "loss": 0.8098,
+      "step": 2705
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.819315293015518e-05,
+      "loss": 0.8081,
+      "step": 2706
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.813085499467594e-05,
+      "loss": 0.8506,
+      "step": 2707
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.806855778486315e-05,
+      "loss": 0.8376,
+      "step": 2708
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.800626132490269e-05,
+      "loss": 0.7772,
+      "step": 2709
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.794396563898022e-05,
+      "loss": 0.8169,
+      "step": 2710
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.788167075128113e-05,
+      "loss": 0.8191,
+      "step": 2711
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.781937668599035e-05,
+      "loss": 0.8291,
+      "step": 2712
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.775708346729263e-05,
+      "loss": 0.8225,
+      "step": 2713
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.769479111937238e-05,
+      "loss": 0.7744,
+      "step": 2714
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.763249966641353e-05,
+      "loss": 0.7983,
+      "step": 2715
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.757020913259987e-05,
+      "loss": 0.8066,
+      "step": 2716
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.750791954211464e-05,
+      "loss": 0.7733,
+      "step": 2717
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.744563091914085e-05,
+      "loss": 0.8093,
+      "step": 2718
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.738334328786112e-05,
+      "loss": 0.8145,
+      "step": 2719
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.732105667245759e-05,
+      "loss": 0.8342,
+      "step": 2720
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.725877109711212e-05,
+      "loss": 0.7781,
+      "step": 2721
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.71964865860061e-05,
+      "loss": 0.856,
+      "step": 2722
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.713420316332049e-05,
+      "loss": 0.814,
+      "step": 2723
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.707192085323589e-05,
+      "loss": 0.801,
+      "step": 2724
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.700963967993246e-05,
+      "loss": 0.7689,
+      "step": 2725
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.694735966758982e-05,
+      "loss": 0.7859,
+      "step": 2726
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.688508084038729e-05,
+      "loss": 0.8472,
+      "step": 2727
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.682280322250366e-05,
+      "loss": 0.8445,
+      "step": 2728
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.676052683811715e-05,
+      "loss": 0.793,
+      "step": 2729
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.669825171140568e-05,
+      "loss": 0.8689,
+      "step": 2730
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.66359778665466e-05,
+      "loss": 0.7959,
+      "step": 2731
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.657370532771671e-05,
+      "loss": 0.8589,
+      "step": 2732
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.65114341190924e-05,
+      "loss": 0.8054,
+      "step": 2733
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.64491642648495e-05,
+      "loss": 0.8394,
+      "step": 2734
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.638689578916326e-05,
+      "loss": 0.813,
+      "step": 2735
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.632462871620847e-05,
+      "loss": 0.7928,
+      "step": 2736
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.62623630701594e-05,
+      "loss": 0.7727,
+      "step": 2737
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.620009887518963e-05,
+      "loss": 0.7666,
+      "step": 2738
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.613783615547233e-05,
+      "loss": 0.7891,
+      "step": 2739
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.607557493518005e-05,
+      "loss": 0.7998,
+      "step": 2740
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.601331523848464e-05,
+      "loss": 0.8167,
+      "step": 2741
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.595105708955757e-05,
+      "loss": 0.7889,
+      "step": 2742
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.588880051256951e-05,
+      "loss": 0.3776,
+      "step": 2743
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.582654553169064e-05,
+      "loss": 0.8181,
+      "step": 2744
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.576429217109053e-05,
+      "loss": 0.8506,
+      "step": 2745
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.5702040454938e-05,
+      "loss": 0.8079,
+      "step": 2746
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.563979040740138e-05,
+      "loss": 0.8083,
+      "step": 2747
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.557754205264826e-05,
+      "loss": 0.8181,
+      "step": 2748
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.551529541484554e-05,
+      "loss": 0.8389,
+      "step": 2749
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.545305051815956e-05,
+      "loss": 0.8447,
+      "step": 2750
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.539080738675597e-05,
+      "loss": 0.7444,
+      "step": 2751
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.53285660447996e-05,
+      "loss": 0.8433,
+      "step": 2752
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.526632651645475e-05,
+      "loss": 0.7803,
+      "step": 2753
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.520408882588497e-05,
+      "loss": 0.7653,
+      "step": 2754
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.514185299725298e-05,
+      "loss": 0.8621,
+      "step": 2755
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.507961905472093e-05,
+      "loss": 0.8806,
+      "step": 2756
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.501738702245022e-05,
+      "loss": 0.8611,
+      "step": 2757
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.495515692460138e-05,
+      "loss": 0.7715,
+      "step": 2758
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.489292878533434e-05,
+      "loss": 0.7966,
+      "step": 2759
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.483070262880822e-05,
+      "loss": 0.824,
+      "step": 2760
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.476847847918127e-05,
+      "loss": 0.7729,
+      "step": 2761
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.47062563606111e-05,
+      "loss": 0.3372,
+      "step": 2762
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.464403629725453e-05,
+      "loss": 0.8401,
+      "step": 2763
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.458181831326743e-05,
+      "loss": 0.8635,
+      "step": 2764
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.451960243280506e-05,
+      "loss": 0.7461,
+      "step": 2765
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.44573886800217e-05,
+      "loss": 0.8157,
+      "step": 2766
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.439517707907088e-05,
+      "loss": 0.7892,
+      "step": 2767
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.433296765410534e-05,
+      "loss": 0.8579,
+      "step": 2768
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.427076042927683e-05,
+      "loss": 0.7959,
+      "step": 2769
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.420855542873637e-05,
+      "loss": 0.8291,
+      "step": 2770
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.414635267663417e-05,
+      "loss": 0.8105,
+      "step": 2771
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.408415219711934e-05,
+      "loss": 0.8184,
+      "step": 2772
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.402195401434036e-05,
+      "loss": 0.803,
+      "step": 2773
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.395975815244466e-05,
+      "loss": 0.8613,
+      "step": 2774
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.389756463557878e-05,
+      "loss": 0.8743,
+      "step": 2775
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.383537348788843e-05,
+      "loss": 0.3419,
+      "step": 2776
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.377318473351838e-05,
+      "loss": 0.7804,
+      "step": 2777
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.371099839661238e-05,
+      "loss": 0.757,
+      "step": 2778
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.364881450131336e-05,
+      "loss": 0.824,
+      "step": 2779
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.358663307176323e-05,
+      "loss": 0.8232,
+      "step": 2780
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.352445413210293e-05,
+      "loss": 0.8518,
+      "step": 2781
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.34622777064725e-05,
+      "loss": 0.7908,
+      "step": 2782
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.340010381901101e-05,
+      "loss": 0.8279,
+      "step": 2783
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.33379324938564e-05,
+      "loss": 0.7896,
+      "step": 2784
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.327576375514581e-05,
+      "loss": 0.8259,
+      "step": 2785
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.321359762701526e-05,
+      "loss": 0.8104,
+      "step": 2786
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.315143413359976e-05,
+      "loss": 0.77,
+      "step": 2787
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.308927329903332e-05,
+      "loss": 0.8079,
+      "step": 2788
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.302711514744896e-05,
+      "loss": 0.8442,
+      "step": 2789
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.296495970297854e-05,
+      "loss": 0.844,
+      "step": 2790
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.290280698975306e-05,
+      "loss": 0.7987,
+      "step": 2791
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.28406570319022e-05,
+      "loss": 0.8599,
+      "step": 2792
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.27785098535548e-05,
+      "loss": 0.7904,
+      "step": 2793
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.271636547883856e-05,
+      "loss": 0.8274,
+      "step": 2794
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.265422393187997e-05,
+      "loss": 0.8237,
+      "step": 2795
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.259208523680458e-05,
+      "loss": 0.7808,
+      "step": 2796
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.252994941773678e-05,
+      "loss": 0.8206,
+      "step": 2797
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.24678164987998e-05,
+      "loss": 0.8345,
+      "step": 2798
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.240568650411581e-05,
+      "loss": 0.8013,
+      "step": 2799
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.234355945780581e-05,
+      "loss": 0.8113,
+      "step": 2800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.228143538398962e-05,
+      "loss": 0.7683,
+      "step": 2801
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.221931430678598e-05,
+      "loss": 0.8381,
+      "step": 2802
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.215719625031245e-05,
+      "loss": 0.8032,
+      "step": 2803
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.209508123868533e-05,
+      "loss": 0.8149,
+      "step": 2804
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.203296929601986e-05,
+      "loss": 0.837,
+      "step": 2805
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.197086044643004e-05,
+      "loss": 0.7983,
+      "step": 2806
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.190875471402865e-05,
+      "loss": 0.3377,
+      "step": 2807
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.184665212292722e-05,
+      "loss": 0.7748,
+      "step": 2808
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.178455269723623e-05,
+      "loss": 0.7957,
+      "step": 2809
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.172245646106469e-05,
+      "loss": 0.8176,
+      "step": 2810
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.16603634385206e-05,
+      "loss": 0.8222,
+      "step": 2811
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.159827365371056e-05,
+      "loss": 0.7942,
+      "step": 2812
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.153618713073995e-05,
+      "loss": 0.7971,
+      "step": 2813
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.14741038937129e-05,
+      "loss": 0.7489,
+      "step": 2814
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.141202396673232e-05,
+      "loss": 0.7994,
+      "step": 2815
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.134994737389969e-05,
+      "loss": 0.8679,
+      "step": 2816
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.128787413931536e-05,
+      "loss": 0.8013,
+      "step": 2817
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.12258042870782e-05,
+      "loss": 0.8423,
+      "step": 2818
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.116373784128596e-05,
+      "loss": 0.7629,
+      "step": 2819
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.110167482603494e-05,
+      "loss": 0.7777,
+      "step": 2820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.10396152654201e-05,
+      "loss": 0.8263,
+      "step": 2821
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.097755918353512e-05,
+      "loss": 0.8237,
+      "step": 2822
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.091550660447235e-05,
+      "loss": 0.8831,
+      "step": 2823
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.085345755232268e-05,
+      "loss": 0.8733,
+      "step": 2824
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.079141205117572e-05,
+      "loss": 0.8701,
+      "step": 2825
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.072937012511969e-05,
+      "loss": 0.3157,
+      "step": 2826
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.066733179824134e-05,
+      "loss": 0.7686,
+      "step": 2827
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.060529709462611e-05,
+      "loss": 0.7104,
+      "step": 2828
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.054326603835807e-05,
+      "loss": 0.832,
+      "step": 2829
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.04812386535197e-05,
+      "loss": 0.7343,
+      "step": 2830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.041921496419226e-05,
+      "loss": 0.7891,
+      "step": 2831
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.035719499445544e-05,
+      "loss": 0.7703,
+      "step": 2832
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.029517876838755e-05,
+      "loss": 0.8062,
+      "step": 2833
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.023316631006537e-05,
+      "loss": 0.856,
+      "step": 2834
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.017115764356435e-05,
+      "loss": 0.8064,
+      "step": 2835
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.010915279295831e-05,
+      "loss": 0.7983,
+      "step": 2836
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.004715178231975e-05,
+      "loss": 0.8046,
+      "step": 2837
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.998515463571953e-05,
+      "loss": 0.8259,
+      "step": 2838
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.99231613772271e-05,
+      "loss": 0.8242,
+      "step": 2839
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.986117203091042e-05,
+      "loss": 0.7822,
+      "step": 2840
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.97991866208358e-05,
+      "loss": 0.8389,
+      "step": 2841
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.973720517106814e-05,
+      "loss": 0.8726,
+      "step": 2842
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.967522770567084e-05,
+      "loss": 0.8306,
+      "step": 2843
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.961325424870561e-05,
+      "loss": 0.8247,
+      "step": 2844
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.955128482423272e-05,
+      "loss": 0.7822,
+      "step": 2845
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.948931945631082e-05,
+      "loss": 0.8005,
+      "step": 2846
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.9427358168997e-05,
+      "loss": 0.8208,
+      "step": 2847
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.936540098634674e-05,
+      "loss": 0.7893,
+      "step": 2848
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.930344793241403e-05,
+      "loss": 0.8052,
+      "step": 2849
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.924149903125108e-05,
+      "loss": 0.7864,
+      "step": 2850
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.917955430690865e-05,
+      "loss": 0.8154,
+      "step": 2851
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.91176137834358e-05,
+      "loss": 0.8123,
+      "step": 2852
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.905567748487996e-05,
+      "loss": 0.8213,
+      "step": 2853
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.899374543528694e-05,
+      "loss": 0.8069,
+      "step": 2854
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.893181765870094e-05,
+      "loss": 0.7783,
+      "step": 2855
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.886989417916435e-05,
+      "loss": 0.8162,
+      "step": 2856
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.880797502071809e-05,
+      "loss": 0.7838,
+      "step": 2857
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.87460602074013e-05,
+      "loss": 0.8164,
+      "step": 2858
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.86841497632514e-05,
+      "loss": 0.7761,
+      "step": 2859
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.862224371230417e-05,
+      "loss": 0.3472,
+      "step": 2860
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.85603420785937e-05,
+      "loss": 0.8789,
+      "step": 2861
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.84984448861523e-05,
+      "loss": 0.7568,
+      "step": 2862
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.843655215901059e-05,
+      "loss": 0.8176,
+      "step": 2863
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.837466392119752e-05,
+      "loss": 0.7905,
+      "step": 2864
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.831278019674016e-05,
+      "loss": 0.8223,
+      "step": 2865
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.825090100966395e-05,
+      "loss": 0.7705,
+      "step": 2866
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.818902638399246e-05,
+      "loss": 0.8531,
+      "step": 2867
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.81271563437476e-05,
+      "loss": 0.834,
+      "step": 2868
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.806529091294948e-05,
+      "loss": 0.8135,
+      "step": 2869
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.800343011561632e-05,
+      "loss": 0.7896,
+      "step": 2870
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.794157397576464e-05,
+      "loss": 0.8083,
+      "step": 2871
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.787972251740915e-05,
+      "loss": 0.8016,
+      "step": 2872
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.781787576456268e-05,
+      "loss": 0.851,
+      "step": 2873
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.775603374123626e-05,
+      "loss": 0.7703,
+      "step": 2874
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.769419647143917e-05,
+      "loss": 0.8225,
+      "step": 2875
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.763236397917864e-05,
+      "loss": 0.8582,
+      "step": 2876
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.757053628846029e-05,
+      "loss": 0.8162,
+      "step": 2877
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.750871342328771e-05,
+      "loss": 0.8086,
+      "step": 2878
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.744689540766264e-05,
+      "loss": 0.8008,
+      "step": 2879
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.7385082265585e-05,
+      "loss": 0.791,
+      "step": 2880
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.732327402105279e-05,
+      "loss": 0.8557,
+      "step": 2881
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.726147069806205e-05,
+      "loss": 0.7949,
+      "step": 2882
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.719967232060697e-05,
+      "loss": 0.7843,
+      "step": 2883
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.713787891267989e-05,
+      "loss": 0.8384,
+      "step": 2884
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.707609049827102e-05,
+      "loss": 0.8635,
+      "step": 2885
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.70143071013688e-05,
+      "loss": 0.7388,
+      "step": 2886
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.695252874595971e-05,
+      "loss": 0.8176,
+      "step": 2887
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.689075545602815e-05,
+      "loss": 0.8584,
+      "step": 2888
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.68289872555567e-05,
+      "loss": 0.8354,
+      "step": 2889
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.676722416852594e-05,
+      "loss": 0.8154,
+      "step": 2890
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.670546621891434e-05,
+      "loss": 0.8118,
+      "step": 2891
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.664371343069849e-05,
+      "loss": 0.8064,
+      "step": 2892
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.658196582785296e-05,
+      "loss": 0.8237,
+      "step": 2893
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.652022343435028e-05,
+      "loss": 0.8672,
+      "step": 2894
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.645848627416103e-05,
+      "loss": 0.7422,
+      "step": 2895
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.63967543712536e-05,
+      "loss": 0.8499,
+      "step": 2896
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.633502774959453e-05,
+      "loss": 0.7869,
+      "step": 2897
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.627330643314818e-05,
+      "loss": 0.7874,
+      "step": 2898
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.62115904458769e-05,
+      "loss": 0.7584,
+      "step": 2899
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.614987981174094e-05,
+      "loss": 0.8198,
+      "step": 2900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.608817455469854e-05,
+      "loss": 0.8223,
+      "step": 2901
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.602647469870574e-05,
+      "loss": 0.7988,
+      "step": 2902
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.596478026771659e-05,
+      "loss": 0.7445,
+      "step": 2903
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.590309128568303e-05,
+      "loss": 0.8452,
+      "step": 2904
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.584140777655476e-05,
+      "loss": 0.3152,
+      "step": 2905
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.577972976427949e-05,
+      "loss": 0.8152,
+      "step": 2906
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.571805727280277e-05,
+      "loss": 0.8021,
+      "step": 2907
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.565639032606794e-05,
+      "loss": 0.8564,
+      "step": 2908
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.559472894801623e-05,
+      "loss": 0.8516,
+      "step": 2909
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.553307316258677e-05,
+      "loss": 0.8938,
+      "step": 2910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.547142299371641e-05,
+      "loss": 0.8025,
+      "step": 2911
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.540977846533985e-05,
+      "loss": 0.8667,
+      "step": 2912
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.534813960138969e-05,
+      "loss": 0.8591,
+      "step": 2913
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.528650642579617e-05,
+      "loss": 0.7961,
+      "step": 2914
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.522487896248749e-05,
+      "loss": 0.7917,
+      "step": 2915
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.51632572353895e-05,
+      "loss": 0.8369,
+      "step": 2916
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.51016412684259e-05,
+      "loss": 0.7856,
+      "step": 2917
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.504003108551814e-05,
+      "loss": 0.8059,
+      "step": 2918
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.497842671058538e-05,
+      "loss": 0.7988,
+      "step": 2919
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.491682816754456e-05,
+      "loss": 0.8284,
+      "step": 2920
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.485523548031042e-05,
+      "loss": 0.8303,
+      "step": 2921
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.479364867279529e-05,
+      "loss": 0.7605,
+      "step": 2922
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.47320677689093e-05,
+      "loss": 0.8098,
+      "step": 2923
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.467049279256033e-05,
+      "loss": 0.8167,
+      "step": 2924
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.460892376765386e-05,
+      "loss": 0.8608,
+      "step": 2925
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.454736071809308e-05,
+      "loss": 0.8453,
+      "step": 2926
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.448580366777897e-05,
+      "loss": 0.8298,
+      "step": 2927
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.442425264061e-05,
+      "loss": 0.8213,
+      "step": 2928
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.436270766048245e-05,
+      "loss": 0.7625,
+      "step": 2929
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.430116875129022e-05,
+      "loss": 0.8232,
+      "step": 2930
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.42396359369248e-05,
+      "loss": 0.7957,
+      "step": 2931
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.417810924127533e-05,
+      "loss": 0.8018,
+      "step": 2932
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.411658868822866e-05,
+      "loss": 0.8481,
+      "step": 2933
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.40550743016691e-05,
+      "loss": 0.7639,
+      "step": 2934
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.399356610547869e-05,
+      "loss": 0.8022,
+      "step": 2935
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.39320641235371e-05,
+      "loss": 0.8379,
+      "step": 2936
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.38705683797214e-05,
+      "loss": 0.8745,
+      "step": 2937
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.380907889790641e-05,
+      "loss": 0.8267,
+      "step": 2938
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.374759570196449e-05,
+      "loss": 0.8604,
+      "step": 2939
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.368611881576547e-05,
+      "loss": 0.8562,
+      "step": 2940
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.362464826317686e-05,
+      "loss": 0.8176,
+      "step": 2941
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.356318406806358e-05,
+      "loss": 0.8064,
+      "step": 2942
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.35017262542882e-05,
+      "loss": 0.9089,
+      "step": 2943
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.344027484571075e-05,
+      "loss": 0.7338,
+      "step": 2944
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.337882986618876e-05,
+      "loss": 0.8354,
+      "step": 2945
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.331739133957728e-05,
+      "loss": 0.3323,
+      "step": 2946
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.325595928972894e-05,
+      "loss": 0.8462,
+      "step": 2947
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.319453374049366e-05,
+      "loss": 0.8228,
+      "step": 2948
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.313311471571901e-05,
+      "loss": 0.8354,
+      "step": 2949
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.307170223925003e-05,
+      "loss": 0.8069,
+      "step": 2950
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.301029633492905e-05,
+      "loss": 0.8467,
+      "step": 2951
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.294889702659602e-05,
+      "loss": 0.3367,
+      "step": 2952
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.288750433808827e-05,
+      "loss": 0.7742,
+      "step": 2953
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.28261182932405e-05,
+      "loss": 0.811,
+      "step": 2954
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.276473891588492e-05,
+      "loss": 0.8369,
+      "step": 2955
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.270336622985116e-05,
+      "loss": 0.7793,
+      "step": 2956
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.264200025896616e-05,
+      "loss": 0.8574,
+      "step": 2957
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.258064102705427e-05,
+      "loss": 0.8151,
+      "step": 2958
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.251928855793735e-05,
+      "loss": 0.8513,
+      "step": 2959
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.245794287543447e-05,
+      "loss": 0.8071,
+      "step": 2960
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.239660400336214e-05,
+      "loss": 0.8098,
+      "step": 2961
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.233527196553427e-05,
+      "loss": 0.7753,
+      "step": 2962
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.227394678576203e-05,
+      "loss": 0.8599,
+      "step": 2963
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.221262848785395e-05,
+      "loss": 0.8031,
+      "step": 2964
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.215131709561598e-05,
+      "loss": 0.7847,
+      "step": 2965
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.20900126328512e-05,
+      "loss": 0.822,
+      "step": 2966
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.202871512336024e-05,
+      "loss": 0.7075,
+      "step": 2967
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.196742459094077e-05,
+      "loss": 0.7661,
+      "step": 2968
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.190614105938796e-05,
+      "loss": 0.8008,
+      "step": 2969
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.184486455249424e-05,
+      "loss": 0.7948,
+      "step": 2970
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.178359509404917e-05,
+      "loss": 0.8114,
+      "step": 2971
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.172233270783966e-05,
+      "loss": 0.8066,
+      "step": 2972
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.166107741764997e-05,
+      "loss": 0.774,
+      "step": 2973
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.159982924726141e-05,
+      "loss": 0.7817,
+      "step": 2974
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.153858822045266e-05,
+      "loss": 0.7839,
+      "step": 2975
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.147735436099967e-05,
+      "loss": 0.8145,
+      "step": 2976
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.141612769267544e-05,
+      "loss": 0.8352,
+      "step": 2977
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.135490823925027e-05,
+      "loss": 0.7856,
+      "step": 2978
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.129369602449175e-05,
+      "loss": 0.7625,
+      "step": 2979
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.123249107216446e-05,
+      "loss": 0.7618,
+      "step": 2980
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.117129340603032e-05,
+      "loss": 0.7839,
+      "step": 2981
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.111010304984841e-05,
+      "loss": 0.7896,
+      "step": 2982
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.104892002737487e-05,
+      "loss": 0.7781,
+      "step": 2983
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.098774436236308e-05,
+      "loss": 0.7886,
+      "step": 2984
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.092657607856356e-05,
+      "loss": 0.7698,
+      "step": 2985
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.086541519972388e-05,
+      "loss": 0.7661,
+      "step": 2986
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.080426174958886e-05,
+      "loss": 0.8301,
+      "step": 2987
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.074311575190038e-05,
+      "loss": 0.7725,
+      "step": 2988
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.068197723039737e-05,
+      "loss": 0.3463,
+      "step": 2989
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.062084620881597e-05,
+      "loss": 0.801,
+      "step": 2990
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.055972271088932e-05,
+      "loss": 0.7964,
+      "step": 2991
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.049860676034761e-05,
+      "loss": 0.7988,
+      "step": 2992
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.043749838091827e-05,
+      "loss": 0.8057,
+      "step": 2993
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.037639759632556e-05,
+      "loss": 0.8434,
+      "step": 2994
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.031530443029099e-05,
+      "loss": 0.7588,
+      "step": 2995
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.025421890653302e-05,
+      "loss": 0.7729,
+      "step": 2996
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.019314104876712e-05,
+      "loss": 0.8464,
+      "step": 2997
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.013207088070582e-05,
+      "loss": 0.7458,
+      "step": 2998
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.007100842605872e-05,
+      "loss": 0.8433,
+      "step": 2999
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.000995370853227e-05,
+      "loss": 0.8486,
+      "step": 3000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.994890675183008e-05,
+      "loss": 0.8337,
+      "step": 3001
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.98878675796527e-05,
+      "loss": 0.8127,
+      "step": 3002
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.98268362156976e-05,
+      "loss": 0.8235,
+      "step": 3003
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.976581268365923e-05,
+      "loss": 0.8097,
+      "step": 3004
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.97047970072291e-05,
+      "loss": 0.7689,
+      "step": 3005
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.964378921009552e-05,
+      "loss": 0.843,
+      "step": 3006
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.958278931594385e-05,
+      "loss": 0.8076,
+      "step": 3007
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.952179734845641e-05,
+      "loss": 0.3355,
+      "step": 3008
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.946081333131227e-05,
+      "loss": 0.7607,
+      "step": 3009
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.93998372881876e-05,
+      "loss": 0.8376,
+      "step": 3010
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.93388692427554e-05,
+      "loss": 0.7894,
+      "step": 3011
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.92779092186855e-05,
+      "loss": 0.7186,
+      "step": 3012
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.921695723964472e-05,
+      "loss": 0.7717,
+      "step": 3013
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.915601332929679e-05,
+      "loss": 0.8708,
+      "step": 3014
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.909507751130209e-05,
+      "loss": 0.764,
+      "step": 3015
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.903414980931814e-05,
+      "loss": 0.8141,
+      "step": 3016
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.897323024699907e-05,
+      "loss": 0.7651,
+      "step": 3017
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.891231884799599e-05,
+      "loss": 0.7858,
+      "step": 3018
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.885141563595684e-05,
+      "loss": 0.7874,
+      "step": 3019
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.879052063452625e-05,
+      "loss": 0.7888,
+      "step": 3020
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.872963386734583e-05,
+      "loss": 0.7687,
+      "step": 3021
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.866875535805394e-05,
+      "loss": 0.8203,
+      "step": 3022
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.860788513028565e-05,
+      "loss": 0.8171,
+      "step": 3023
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.854702320767289e-05,
+      "loss": 0.8335,
+      "step": 3024
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.848616961384442e-05,
+      "loss": 0.8254,
+      "step": 3025
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.842532437242559e-05,
+      "loss": 0.8635,
+      "step": 3026
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.836448750703868e-05,
+      "loss": 0.7957,
+      "step": 3027
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.83036590413027e-05,
+      "loss": 0.7915,
+      "step": 3028
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.824283899883326e-05,
+      "loss": 0.8523,
+      "step": 3029
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.818202740324286e-05,
+      "loss": 0.78,
+      "step": 3030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.812122427814068e-05,
+      "loss": 0.8101,
+      "step": 3031
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.806042964713247e-05,
+      "loss": 0.8149,
+      "step": 3032
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.799964353382089e-05,
+      "loss": 0.7886,
+      "step": 3033
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.793886596180522e-05,
+      "loss": 0.7692,
+      "step": 3034
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.787809695468133e-05,
+      "loss": 0.7874,
+      "step": 3035
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.78173365360419e-05,
+      "loss": 0.3358,
+      "step": 3036
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.775658472947623e-05,
+      "loss": 0.8318,
+      "step": 3037
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.769584155857018e-05,
+      "loss": 0.8137,
+      "step": 3038
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.763510704690644e-05,
+      "loss": 0.8282,
+      "step": 3039
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.757438121806414e-05,
+      "loss": 0.7969,
+      "step": 3040
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.75136640956192e-05,
+      "loss": 0.8198,
+      "step": 3041
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.745295570314411e-05,
+      "loss": 0.7634,
+      "step": 3042
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.739225606420793e-05,
+      "loss": 0.8293,
+      "step": 3043
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.733156520237632e-05,
+      "loss": 0.8169,
+      "step": 3044
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.727088314121165e-05,
+      "loss": 0.8345,
+      "step": 3045
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.721020990427268e-05,
+      "loss": 0.79,
+      "step": 3046
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.71495455151149e-05,
+      "loss": 0.8286,
+      "step": 3047
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.708888999729036e-05,
+      "loss": 0.8123,
+      "step": 3048
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.702824337434755e-05,
+      "loss": 0.8052,
+      "step": 3049
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.696760566983159e-05,
+      "loss": 0.3393,
+      "step": 3050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.690697690728416e-05,
+      "loss": 0.8279,
+      "step": 3051
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.68463571102434e-05,
+      "loss": 0.7869,
+      "step": 3052
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6785746302244e-05,
+      "loss": 0.7595,
+      "step": 3053
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.67251445068172e-05,
+      "loss": 0.7864,
+      "step": 3054
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.666455174749066e-05,
+      "loss": 0.8101,
+      "step": 3055
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.66039680477886e-05,
+      "loss": 0.775,
+      "step": 3056
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.654339343123173e-05,
+      "loss": 0.7842,
+      "step": 3057
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.648282792133711e-05,
+      "loss": 0.7727,
+      "step": 3058
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.642227154161841e-05,
+      "loss": 0.7535,
+      "step": 3059
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.636172431558575e-05,
+      "loss": 0.8418,
+      "step": 3060
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.630118626674556e-05,
+      "loss": 0.8685,
+      "step": 3061
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.624065741860084e-05,
+      "loss": 0.3328,
+      "step": 3062
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.618013779465101e-05,
+      "loss": 0.3403,
+      "step": 3063
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.611962741839178e-05,
+      "loss": 0.7957,
+      "step": 3064
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.605912631331546e-05,
+      "loss": 0.8207,
+      "step": 3065
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.599863450291055e-05,
+      "loss": 0.7935,
+      "step": 3066
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.593815201066215e-05,
+      "loss": 0.8311,
+      "step": 3067
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.587767886005163e-05,
+      "loss": 0.7787,
+      "step": 3068
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.581721507455673e-05,
+      "loss": 0.8357,
+      "step": 3069
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.575676067765154e-05,
+      "loss": 0.8345,
+      "step": 3070
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.569631569280661e-05,
+      "loss": 0.8145,
+      "step": 3071
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.56358801434887e-05,
+      "loss": 0.8077,
+      "step": 3072
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5575454053161e-05,
+      "loss": 0.8538,
+      "step": 3073
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.551503744528304e-05,
+      "loss": 0.8362,
+      "step": 3074
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.545463034331053e-05,
+      "loss": 0.7629,
+      "step": 3075
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.539423277069568e-05,
+      "loss": 0.8138,
+      "step": 3076
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.53338447508869e-05,
+      "loss": 0.3488,
+      "step": 3077
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.52734663073288e-05,
+      "loss": 0.783,
+      "step": 3078
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.521309746346246e-05,
+      "loss": 0.7664,
+      "step": 3079
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.515273824272514e-05,
+      "loss": 0.8499,
+      "step": 3080
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.509238866855033e-05,
+      "loss": 0.8313,
+      "step": 3081
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.503204876436785e-05,
+      "loss": 0.3486,
+      "step": 3082
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.497171855360371e-05,
+      "loss": 0.8191,
+      "step": 3083
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.491139805968018e-05,
+      "loss": 0.8127,
+      "step": 3084
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.485108730601571e-05,
+      "loss": 0.7799,
+      "step": 3085
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.479078631602512e-05,
+      "loss": 0.7786,
+      "step": 3086
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.47304951131192e-05,
+      "loss": 0.8301,
+      "step": 3087
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.467021372070516e-05,
+      "loss": 0.8301,
+      "step": 3088
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.46099421621863e-05,
+      "loss": 0.3638,
+      "step": 3089
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.454968046096204e-05,
+      "loss": 0.3172,
+      "step": 3090
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.448942864042818e-05,
+      "loss": 0.8486,
+      "step": 3091
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.442918672397642e-05,
+      "loss": 0.3624,
+      "step": 3092
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.43689547349948e-05,
+      "loss": 0.8325,
+      "step": 3093
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.43087326968675e-05,
+      "loss": 0.8052,
+      "step": 3094
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.424852063297469e-05,
+      "loss": 0.7961,
+      "step": 3095
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.418831856669286e-05,
+      "loss": 0.8142,
+      "step": 3096
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.41281265213945e-05,
+      "loss": 0.8083,
+      "step": 3097
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.406794452044816e-05,
+      "loss": 0.8293,
+      "step": 3098
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.400777258721865e-05,
+      "loss": 0.8462,
+      "step": 3099
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.394761074506679e-05,
+      "loss": 0.8281,
+      "step": 3100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.38874590173494e-05,
+      "loss": 0.8772,
+      "step": 3101
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.382731742741952e-05,
+      "loss": 0.8033,
+      "step": 3102
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.376718599862621e-05,
+      "loss": 0.7976,
+      "step": 3103
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.370706475431446e-05,
+      "loss": 0.814,
+      "step": 3104
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.364695371782546e-05,
+      "loss": 0.8578,
+      "step": 3105
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.358685291249644e-05,
+      "loss": 0.7974,
+      "step": 3106
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.352676236166051e-05,
+      "loss": 0.863,
+      "step": 3107
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.346668208864695e-05,
+      "loss": 0.7808,
+      "step": 3108
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.340661211678099e-05,
+      "loss": 0.8865,
+      "step": 3109
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.33465524693838e-05,
+      "loss": 0.75,
+      "step": 3110
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.328650316977264e-05,
+      "loss": 0.7783,
+      "step": 3111
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.322646424126078e-05,
+      "loss": 0.8049,
+      "step": 3112
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.31664357071573e-05,
+      "loss": 0.7971,
+      "step": 3113
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.310641759076741e-05,
+      "loss": 0.814,
+      "step": 3114
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.304640991539215e-05,
+      "loss": 0.7686,
+      "step": 3115
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.298641270432862e-05,
+      "loss": 0.7981,
+      "step": 3116
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.292642598086981e-05,
+      "loss": 0.8184,
+      "step": 3117
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.286644976830457e-05,
+      "loss": 0.8152,
+      "step": 3118
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.280648408991775e-05,
+      "loss": 0.787,
+      "step": 3119
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.274652896899015e-05,
+      "loss": 0.7659,
+      "step": 3120
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.268658442879833e-05,
+      "loss": 0.8752,
+      "step": 3121
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.262665049261488e-05,
+      "loss": 0.7778,
+      "step": 3122
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.256672718370824e-05,
+      "loss": 0.853,
+      "step": 3123
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.25068145253426e-05,
+      "loss": 0.7911,
+      "step": 3124
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.244691254077819e-05,
+      "loss": 0.8374,
+      "step": 3125
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.238702125327106e-05,
+      "loss": 0.7715,
+      "step": 3126
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.232714068607295e-05,
+      "loss": 0.8242,
+      "step": 3127
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.226727086243168e-05,
+      "loss": 0.3315,
+      "step": 3128
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.220741180559073e-05,
+      "loss": 0.7721,
+      "step": 3129
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.214756353878942e-05,
+      "loss": 0.8413,
+      "step": 3130
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.208772608526293e-05,
+      "loss": 0.7676,
+      "step": 3131
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.202789946824227e-05,
+      "loss": 0.811,
+      "step": 3132
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.196808371095408e-05,
+      "loss": 0.8374,
+      "step": 3133
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.190827883662099e-05,
+      "loss": 0.7914,
+      "step": 3134
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.184848486846128e-05,
+      "loss": 0.8225,
+      "step": 3135
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.178870182968903e-05,
+      "loss": 0.8518,
+      "step": 3136
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.172892974351406e-05,
+      "loss": 0.8354,
+      "step": 3137
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.166916863314199e-05,
+      "loss": 0.8403,
+      "step": 3138
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.160941852177409e-05,
+      "loss": 0.8391,
+      "step": 3139
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.154967943260748e-05,
+      "loss": 0.7656,
+      "step": 3140
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.148995138883483e-05,
+      "loss": 0.772,
+      "step": 3141
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.143023441364472e-05,
+      "loss": 0.7578,
+      "step": 3142
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.13705285302213e-05,
+      "loss": 0.7874,
+      "step": 3143
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.131083376174441e-05,
+      "loss": 0.8315,
+      "step": 3144
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.125115013138966e-05,
+      "loss": 0.8101,
+      "step": 3145
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.119147766232833e-05,
+      "loss": 0.7838,
+      "step": 3146
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.113181637772721e-05,
+      "loss": 0.8215,
+      "step": 3147
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.107216630074896e-05,
+      "loss": 0.8534,
+      "step": 3148
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.101252745455179e-05,
+      "loss": 0.7322,
+      "step": 3149
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.09528998622895e-05,
+      "loss": 0.8584,
+      "step": 3150
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.089328354711159e-05,
+      "loss": 0.8091,
+      "step": 3151
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.083367853216323e-05,
+      "loss": 0.8477,
+      "step": 3152
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.077408484058505e-05,
+      "loss": 0.8317,
+      "step": 3153
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.071450249551341e-05,
+      "loss": 0.8232,
+      "step": 3154
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.065493152008026e-05,
+      "loss": 0.8009,
+      "step": 3155
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.059537193741306e-05,
+      "loss": 0.7874,
+      "step": 3156
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.053582377063489e-05,
+      "loss": 0.8142,
+      "step": 3157
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.047628704286446e-05,
+      "loss": 0.811,
+      "step": 3158
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.041676177721588e-05,
+      "loss": 0.822,
+      "step": 3159
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.035724799679898e-05,
+      "loss": 0.8199,
+      "step": 3160
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.029774572471903e-05,
+      "loss": 0.7812,
+      "step": 3161
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.023825498407689e-05,
+      "loss": 0.7781,
+      "step": 3162
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.017877579796885e-05,
+      "loss": 0.8555,
+      "step": 3163
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.011930818948688e-05,
+      "loss": 0.78,
+      "step": 3164
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.005985218171825e-05,
+      "loss": 0.3374,
+      "step": 3165
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.000040779774591e-05,
+      "loss": 0.7898,
+      "step": 3166
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.994097506064812e-05,
+      "loss": 0.8467,
+      "step": 3167
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.988155399349881e-05,
+      "loss": 0.8257,
+      "step": 3168
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.982214461936727e-05,
+      "loss": 0.3435,
+      "step": 3169
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.97627469613182e-05,
+      "loss": 0.7913,
+      "step": 3170
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.970336104241186e-05,
+      "loss": 0.8066,
+      "step": 3171
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.964398688570395e-05,
+      "loss": 0.7913,
+      "step": 3172
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.958462451424547e-05,
+      "loss": 0.8286,
+      "step": 3173
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.952527395108302e-05,
+      "loss": 0.332,
+      "step": 3174
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.946593521925851e-05,
+      "loss": 0.8438,
+      "step": 3175
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.940660834180921e-05,
+      "loss": 0.7498,
+      "step": 3176
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.934729334176792e-05,
+      "loss": 0.845,
+      "step": 3177
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.928799024216282e-05,
+      "loss": 0.7954,
+      "step": 3178
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.92286990660173e-05,
+      "loss": 0.7277,
+      "step": 3179
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.91694198363503e-05,
+      "loss": 0.8054,
+      "step": 3180
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.911015257617605e-05,
+      "loss": 0.7396,
+      "step": 3181
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.905089730850416e-05,
+      "loss": 0.7993,
+      "step": 3182
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.89916540563395e-05,
+      "loss": 0.7701,
+      "step": 3183
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.893242284268244e-05,
+      "loss": 0.8196,
+      "step": 3184
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.887320369052848e-05,
+      "loss": 0.825,
+      "step": 3185
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.881399662286858e-05,
+      "loss": 0.8364,
+      "step": 3186
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.875480166268896e-05,
+      "loss": 0.8362,
+      "step": 3187
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.869561883297116e-05,
+      "loss": 0.7727,
+      "step": 3188
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.863644815669197e-05,
+      "loss": 0.7775,
+      "step": 3189
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.857728965682345e-05,
+      "loss": 0.8218,
+      "step": 3190
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.851814335633297e-05,
+      "loss": 0.8376,
+      "step": 3191
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.845900927818327e-05,
+      "loss": 0.7728,
+      "step": 3192
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.839988744533211e-05,
+      "loss": 0.8096,
+      "step": 3193
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.834077788073267e-05,
+      "loss": 0.3231,
+      "step": 3194
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.828168060733337e-05,
+      "loss": 0.8157,
+      "step": 3195
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.822259564807768e-05,
+      "loss": 0.8489,
+      "step": 3196
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.816352302590449e-05,
+      "loss": 0.8301,
+      "step": 3197
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.810446276374789e-05,
+      "loss": 0.7749,
+      "step": 3198
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.804541488453697e-05,
+      "loss": 0.8269,
+      "step": 3199
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.798637941119625e-05,
+      "loss": 0.3475,
+      "step": 3200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.792735636664532e-05,
+      "loss": 0.8264,
+      "step": 3201
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.786834577379892e-05,
+      "loss": 0.8132,
+      "step": 3202
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.780934765556701e-05,
+      "loss": 0.8284,
+      "step": 3203
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.775036203485473e-05,
+      "loss": 0.8254,
+      "step": 3204
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.769138893456225e-05,
+      "loss": 0.7977,
+      "step": 3205
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.763242837758504e-05,
+      "loss": 0.8088,
+      "step": 3206
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.757348038681357e-05,
+      "loss": 0.8003,
+      "step": 3207
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.751454498513348e-05,
+      "loss": 0.8303,
+      "step": 3208
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.745562219542554e-05,
+      "loss": 0.7236,
+      "step": 3209
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.739671204056562e-05,
+      "loss": 0.7881,
+      "step": 3210
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.733781454342462e-05,
+      "loss": 0.7401,
+      "step": 3211
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.72789297268686e-05,
+      "loss": 0.7871,
+      "step": 3212
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.722005761375873e-05,
+      "loss": 0.7905,
+      "step": 3213
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.71611982269511e-05,
+      "loss": 0.7988,
+      "step": 3214
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.710235158929703e-05,
+      "loss": 0.8523,
+      "step": 3215
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.704351772364274e-05,
+      "loss": 0.8276,
+      "step": 3216
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.698469665282957e-05,
+      "loss": 0.8284,
+      "step": 3217
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.692588839969397e-05,
+      "loss": 0.7614,
+      "step": 3218
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.686709298706722e-05,
+      "loss": 0.7993,
+      "step": 3219
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.680831043777579e-05,
+      "loss": 0.7861,
+      "step": 3220
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.674954077464107e-05,
+      "loss": 0.7881,
+      "step": 3221
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.669078402047948e-05,
+      "loss": 0.7917,
+      "step": 3222
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.663204019810237e-05,
+      "loss": 0.7346,
+      "step": 3223
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.657330933031619e-05,
+      "loss": 0.7638,
+      "step": 3224
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.65145914399222e-05,
+      "loss": 0.8135,
+      "step": 3225
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.645588654971678e-05,
+      "loss": 0.327,
+      "step": 3226
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.639719468249115e-05,
+      "loss": 0.8096,
+      "step": 3227
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.633851586103153e-05,
+      "loss": 0.8259,
+      "step": 3228
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.627985010811902e-05,
+      "loss": 0.8091,
+      "step": 3229
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.622119744652978e-05,
+      "loss": 0.8018,
+      "step": 3230
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.616255789903466e-05,
+      "loss": 0.7822,
+      "step": 3231
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.610393148839964e-05,
+      "loss": 0.8076,
+      "step": 3232
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.604531823738552e-05,
+      "loss": 0.8062,
+      "step": 3233
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.598671816874794e-05,
+      "loss": 0.8809,
+      "step": 3234
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.592813130523746e-05,
+      "loss": 0.8889,
+      "step": 3235
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.586955766959957e-05,
+      "loss": 0.8132,
+      "step": 3236
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.581099728457451e-05,
+      "loss": 0.8406,
+      "step": 3237
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.575245017289746e-05,
+      "loss": 0.8359,
+      "step": 3238
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.569391635729847e-05,
+      "loss": 0.7628,
+      "step": 3239
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.563539586050233e-05,
+      "loss": 0.7832,
+      "step": 3240
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.557688870522871e-05,
+      "loss": 0.8198,
+      "step": 3241
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.551839491419213e-05,
+      "loss": 0.8208,
+      "step": 3242
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.545991451010185e-05,
+      "loss": 0.792,
+      "step": 3243
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.540144751566207e-05,
+      "loss": 0.8032,
+      "step": 3244
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.534299395357155e-05,
+      "loss": 0.8164,
+      "step": 3245
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.52845538465241e-05,
+      "loss": 0.8477,
+      "step": 3246
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.522612721720814e-05,
+      "loss": 0.7446,
+      "step": 3247
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.516771408830686e-05,
+      "loss": 0.7695,
+      "step": 3248
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.51093144824983e-05,
+      "loss": 0.8547,
+      "step": 3249
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.505092842245519e-05,
+      "loss": 0.7905,
+      "step": 3250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.499255593084498e-05,
+      "loss": 0.7554,
+      "step": 3251
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.49341970303299e-05,
+      "loss": 0.7741,
+      "step": 3252
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.48758517435669e-05,
+      "loss": 0.8533,
+      "step": 3253
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.481752009320761e-05,
+      "loss": 0.3131,
+      "step": 3254
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.475920210189836e-05,
+      "loss": 0.8574,
+      "step": 3255
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.470089779228028e-05,
+      "loss": 0.7675,
+      "step": 3256
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.464260718698902e-05,
+      "loss": 0.8046,
+      "step": 3257
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.458433030865502e-05,
+      "loss": 0.7617,
+      "step": 3258
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.452606717990346e-05,
+      "loss": 0.8,
+      "step": 3259
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4467817823354e-05,
+      "loss": 0.8357,
+      "step": 3260
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.440958226162104e-05,
+      "loss": 0.8076,
+      "step": 3261
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.43513605173137e-05,
+      "loss": 0.8444,
+      "step": 3262
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.429315261303559e-05,
+      "loss": 0.823,
+      "step": 3263
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.423495857138509e-05,
+      "loss": 0.8517,
+      "step": 3264
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.417677841495507e-05,
+      "loss": 0.713,
+      "step": 3265
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.411861216633313e-05,
+      "loss": 0.8694,
+      "step": 3266
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.406045984810135e-05,
+      "loss": 0.8088,
+      "step": 3267
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.400232148283651e-05,
+      "loss": 0.765,
+      "step": 3268
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.394419709310988e-05,
+      "loss": 0.3539,
+      "step": 3269
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.388608670148741e-05,
+      "loss": 0.7845,
+      "step": 3270
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.38279903305295e-05,
+      "loss": 0.7749,
+      "step": 3271
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.376990800279119e-05,
+      "loss": 0.8098,
+      "step": 3272
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.371183974082203e-05,
+      "loss": 0.8335,
+      "step": 3273
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.365378556716612e-05,
+      "loss": 0.7711,
+      "step": 3274
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.359574550436208e-05,
+      "loss": 0.7974,
+      "step": 3275
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.35377195749431e-05,
+      "loss": 0.8081,
+      "step": 3276
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.347970780143678e-05,
+      "loss": 0.8315,
+      "step": 3277
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.342171020636533e-05,
+      "loss": 0.8262,
+      "step": 3278
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.336372681224542e-05,
+      "loss": 0.7516,
+      "step": 3279
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.33057576415882e-05,
+      "loss": 0.8867,
+      "step": 3280
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.324780271689923e-05,
+      "loss": 0.7634,
+      "step": 3281
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.318986206067872e-05,
+      "loss": 0.8271,
+      "step": 3282
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.313193569542113e-05,
+      "loss": 0.8696,
+      "step": 3283
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.30740236436155e-05,
+      "loss": 0.7838,
+      "step": 3284
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.301612592774533e-05,
+      "loss": 0.8306,
+      "step": 3285
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.295824257028844e-05,
+      "loss": 0.8511,
+      "step": 3286
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.290037359371717e-05,
+      "loss": 0.7854,
+      "step": 3287
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.284251902049827e-05,
+      "loss": 0.7781,
+      "step": 3288
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.278467887309283e-05,
+      "loss": 0.8,
+      "step": 3289
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.272685317395643e-05,
+      "loss": 0.8105,
+      "step": 3290
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.266904194553895e-05,
+      "loss": 0.8085,
+      "step": 3291
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.261124521028476e-05,
+      "loss": 0.8306,
+      "step": 3292
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.255346299063252e-05,
+      "loss": 0.7292,
+      "step": 3293
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.249569530901524e-05,
+      "loss": 0.8063,
+      "step": 3294
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.243794218786035e-05,
+      "loss": 0.8101,
+      "step": 3295
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.238020364958965e-05,
+      "loss": 0.8381,
+      "step": 3296
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.232247971661912e-05,
+      "loss": 0.8113,
+      "step": 3297
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.226477041135926e-05,
+      "loss": 0.781,
+      "step": 3298
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.22070757562148e-05,
+      "loss": 0.8428,
+      "step": 3299
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.214939577358478e-05,
+      "loss": 0.8213,
+      "step": 3300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.209173048586253e-05,
+      "loss": 0.8367,
+      "step": 3301
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.203407991543577e-05,
+      "loss": 0.7905,
+      "step": 3302
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.197644408468635e-05,
+      "loss": 0.8116,
+      "step": 3303
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.191882301599052e-05,
+      "loss": 0.8108,
+      "step": 3304
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.186121673171882e-05,
+      "loss": 0.7993,
+      "step": 3305
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.180362525423591e-05,
+      "loss": 0.7976,
+      "step": 3306
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.174604860590081e-05,
+      "loss": 0.7852,
+      "step": 3307
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.168848680906678e-05,
+      "loss": 0.7954,
+      "step": 3308
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.163093988608127e-05,
+      "loss": 0.8269,
+      "step": 3309
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.157340785928595e-05,
+      "loss": 0.799,
+      "step": 3310
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.151589075101682e-05,
+      "loss": 0.8286,
+      "step": 3311
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.145838858360391e-05,
+      "loss": 0.8381,
+      "step": 3312
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.140090137937158e-05,
+      "loss": 0.7998,
+      "step": 3313
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.134342916063838e-05,
+      "loss": 0.8481,
+      "step": 3314
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.12859719497169e-05,
+      "loss": 0.3369,
+      "step": 3315
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.122852976891414e-05,
+      "loss": 0.8257,
+      "step": 3316
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1171102640531e-05,
+      "loss": 0.7776,
+      "step": 3317
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.111369058686276e-05,
+      "loss": 0.8003,
+      "step": 3318
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.105629363019875e-05,
+      "loss": 0.8059,
+      "step": 3319
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.099891179282241e-05,
+      "loss": 0.75,
+      "step": 3320
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.094154509701133e-05,
+      "loss": 0.7511,
+      "step": 3321
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.088419356503732e-05,
+      "loss": 0.8748,
+      "step": 3322
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0826857219166124e-05,
+      "loss": 0.7039,
+      "step": 3323
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0769536081657717e-05,
+      "loss": 0.7795,
+      "step": 3324
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0712230174766194e-05,
+      "loss": 0.8167,
+      "step": 3325
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.065493952073961e-05,
+      "loss": 0.8333,
+      "step": 3326
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0597664141820176e-05,
+      "loss": 0.7678,
+      "step": 3327
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0540404060244214e-05,
+      "loss": 0.8357,
+      "step": 3328
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0483159298242e-05,
+      "loss": 0.8706,
+      "step": 3329
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.042592987803796e-05,
+      "loss": 0.7595,
+      "step": 3330
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.036871582185054e-05,
+      "loss": 0.8467,
+      "step": 3331
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.031151715189217e-05,
+      "loss": 0.7441,
+      "step": 3332
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.025433389036934e-05,
+      "loss": 0.7737,
+      "step": 3333
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.019716605948261e-05,
+      "loss": 0.7871,
+      "step": 3334
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.014001368142643e-05,
+      "loss": 0.8013,
+      "step": 3335
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0082876778389365e-05,
+      "loss": 0.7844,
+      "step": 3336
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.002575537255395e-05,
+      "loss": 0.7378,
+      "step": 3337
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.996864948609662e-05,
+      "loss": 0.7958,
+      "step": 3338
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.991155914118792e-05,
+      "loss": 0.7446,
+      "step": 3339
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.985448435999224e-05,
+      "loss": 0.3378,
+      "step": 3340
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.979742516466793e-05,
+      "loss": 0.7759,
+      "step": 3341
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9740381577367455e-05,
+      "loss": 0.7737,
+      "step": 3342
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.968335362023697e-05,
+      "loss": 0.8464,
+      "step": 3343
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9626341315416754e-05,
+      "loss": 0.7766,
+      "step": 3344
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9569344685041006e-05,
+      "loss": 0.7893,
+      "step": 3345
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9512363751237677e-05,
+      "loss": 0.7708,
+      "step": 3346
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.945539853612876e-05,
+      "loss": 0.3372,
+      "step": 3347
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9398449061830164e-05,
+      "loss": 0.689,
+      "step": 3348
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.934151535045156e-05,
+      "loss": 0.787,
+      "step": 3349
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.92845974240966e-05,
+      "loss": 0.8621,
+      "step": 3350
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9227695304862826e-05,
+      "loss": 0.7854,
+      "step": 3351
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.917080901484155e-05,
+      "loss": 0.7612,
+      "step": 3352
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9113938576118e-05,
+      "loss": 0.8184,
+      "step": 3353
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.905708401077128e-05,
+      "loss": 0.7969,
+      "step": 3354
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.900024534087421e-05,
+      "loss": 0.8013,
+      "step": 3355
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8943422588493544e-05,
+      "loss": 0.8311,
+      "step": 3356
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8886615775689904e-05,
+      "loss": 0.7776,
+      "step": 3357
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.882982492451757e-05,
+      "loss": 0.8118,
+      "step": 3358
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8773050057024716e-05,
+      "loss": 0.3308,
+      "step": 3359
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.871629119525335e-05,
+      "loss": 0.7847,
+      "step": 3360
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.865954836123915e-05,
+      "loss": 0.7532,
+      "step": 3361
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.860282157701167e-05,
+      "loss": 0.8181,
+      "step": 3362
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.854611086459423e-05,
+      "loss": 0.8198,
+      "step": 3363
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.848941624600381e-05,
+      "loss": 0.8171,
+      "step": 3364
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.843273774325131e-05,
+      "loss": 0.8049,
+      "step": 3365
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.837607537834119e-05,
+      "loss": 0.3507,
+      "step": 3366
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8319429173271725e-05,
+      "loss": 0.8308,
+      "step": 3367
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.826279915003503e-05,
+      "loss": 0.8101,
+      "step": 3368
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.820618533061672e-05,
+      "loss": 0.8157,
+      "step": 3369
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8149587736996246e-05,
+      "loss": 0.8152,
+      "step": 3370
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.809300639114683e-05,
+      "loss": 0.3632,
+      "step": 3371
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.803644131503516e-05,
+      "loss": 0.8,
+      "step": 3372
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.797989253062186e-05,
+      "loss": 0.8164,
+      "step": 3373
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.792336005986105e-05,
+      "loss": 0.8452,
+      "step": 3374
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7866843924700634e-05,
+      "loss": 0.8457,
+      "step": 3375
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7810344147082086e-05,
+      "loss": 0.7988,
+      "step": 3376
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.775386074894058e-05,
+      "loss": 0.8159,
+      "step": 3377
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7697393752204887e-05,
+      "loss": 0.3552,
+      "step": 3378
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7640943178797444e-05,
+      "loss": 0.8035,
+      "step": 3379
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.758450905063439e-05,
+      "loss": 0.8198,
+      "step": 3380
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7528091389625247e-05,
+      "loss": 0.7469,
+      "step": 3381
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.747169021767343e-05,
+      "loss": 0.8335,
+      "step": 3382
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7415305556675805e-05,
+      "loss": 0.7778,
+      "step": 3383
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7358937428522694e-05,
+      "loss": 0.803,
+      "step": 3384
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7302585855098315e-05,
+      "loss": 0.8511,
+      "step": 3385
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7246250858280214e-05,
+      "loss": 0.8264,
+      "step": 3386
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7189932459939575e-05,
+      "loss": 0.8389,
+      "step": 3387
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.713363068194115e-05,
+      "loss": 0.7954,
+      "step": 3388
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7077345546143235e-05,
+      "loss": 0.8499,
+      "step": 3389
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7021077074397654e-05,
+      "loss": 0.856,
+      "step": 3390
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.696482528854974e-05,
+      "loss": 0.7329,
+      "step": 3391
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6908590210438416e-05,
+      "loss": 0.7413,
+      "step": 3392
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6852371861896e-05,
+      "loss": 0.7815,
+      "step": 3393
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6796170264748525e-05,
+      "loss": 0.7517,
+      "step": 3394
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6739985440815266e-05,
+      "loss": 0.8113,
+      "step": 3395
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6683817411909114e-05,
+      "loss": 0.7874,
+      "step": 3396
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6627666199836526e-05,
+      "loss": 0.3403,
+      "step": 3397
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.657153182639721e-05,
+      "loss": 0.8228,
+      "step": 3398
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.651541431338454e-05,
+      "loss": 0.3475,
+      "step": 3399
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.645931368258527e-05,
+      "loss": 0.8679,
+      "step": 3400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.640322995577958e-05,
+      "loss": 0.7303,
+      "step": 3401
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.634716315474109e-05,
+      "loss": 0.8079,
+      "step": 3402
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.629111330123689e-05,
+      "loss": 0.8157,
+      "step": 3403
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.623508041702743e-05,
+      "loss": 0.8013,
+      "step": 3404
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.617906452386659e-05,
+      "loss": 0.8308,
+      "step": 3405
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.612306564350178e-05,
+      "loss": 0.3469,
+      "step": 3406
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6067083797673536e-05,
+      "loss": 0.3132,
+      "step": 3407
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.601111900811606e-05,
+      "loss": 0.8174,
+      "step": 3408
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.595517129655681e-05,
+      "loss": 0.3109,
+      "step": 3409
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.589924068471648e-05,
+      "loss": 0.7463,
+      "step": 3410
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5843327194309394e-05,
+      "loss": 0.7589,
+      "step": 3411
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.578743084704305e-05,
+      "loss": 0.8185,
+      "step": 3412
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.573155166461833e-05,
+      "loss": 0.7625,
+      "step": 3413
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.567568966872947e-05,
+      "loss": 0.8115,
+      "step": 3414
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5619844881064e-05,
+      "loss": 0.7498,
+      "step": 3415
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.556401732330281e-05,
+      "loss": 0.8384,
+      "step": 3416
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.550820701712007e-05,
+      "loss": 0.8308,
+      "step": 3417
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.545241398418326e-05,
+      "loss": 0.6696,
+      "step": 3418
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.539663824615312e-05,
+      "loss": 0.7742,
+      "step": 3419
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5340879824683834e-05,
+      "loss": 0.7761,
+      "step": 3420
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.528513874142262e-05,
+      "loss": 0.7264,
+      "step": 3421
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.522941501801008e-05,
+      "loss": 0.7625,
+      "step": 3422
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.51737086760802e-05,
+      "loss": 0.328,
+      "step": 3423
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5118019737259966e-05,
+      "loss": 0.8601,
+      "step": 3424
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5062348223169825e-05,
+      "loss": 0.7659,
+      "step": 3425
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5006694155423356e-05,
+      "loss": 0.7107,
+      "step": 3426
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.495105755562738e-05,
+      "loss": 0.77,
+      "step": 3427
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.489543844538194e-05,
+      "loss": 0.7213,
+      "step": 3428
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.48398368462803e-05,
+      "loss": 0.7942,
+      "step": 3429
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4784252779908905e-05,
+      "loss": 0.3656,
+      "step": 3430
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4728686267847354e-05,
+      "loss": 0.7949,
+      "step": 3431
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.467313733166862e-05,
+      "loss": 0.7893,
+      "step": 3432
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.461760599293855e-05,
+      "loss": 0.8054,
+      "step": 3433
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.456209227321642e-05,
+      "loss": 0.8577,
+      "step": 3434
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4506596194054584e-05,
+      "loss": 0.854,
+      "step": 3435
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.445111777699842e-05,
+      "loss": 0.864,
+      "step": 3436
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.439565704358667e-05,
+      "loss": 0.8621,
+      "step": 3437
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4340214015351046e-05,
+      "loss": 0.8489,
+      "step": 3438
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.428478871381646e-05,
+      "loss": 0.7759,
+      "step": 3439
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.422938116050091e-05,
+      "loss": 0.8081,
+      "step": 3440
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.417399137691552e-05,
+      "loss": 0.8557,
+      "step": 3441
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.411861938456453e-05,
+      "loss": 0.795,
+      "step": 3442
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4063265204945224e-05,
+      "loss": 0.7903,
+      "step": 3443
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.400792885954802e-05,
+      "loss": 0.8853,
+      "step": 3444
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.395261036985635e-05,
+      "loss": 0.7859,
+      "step": 3445
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.389730975734686e-05,
+      "loss": 0.7721,
+      "step": 3446
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.384202704348902e-05,
+      "loss": 0.7266,
+      "step": 3447
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.378676224974557e-05,
+      "loss": 0.8159,
+      "step": 3448
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.373151539757224e-05,
+      "loss": 0.749,
+      "step": 3449
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.36762865084176e-05,
+      "loss": 0.8149,
+      "step": 3450
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.362107560372358e-05,
+      "loss": 0.7676,
+      "step": 3451
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.356588270492487e-05,
+      "loss": 0.8389,
+      "step": 3452
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.351070783344926e-05,
+      "loss": 0.7065,
+      "step": 3453
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3455551010717545e-05,
+      "loss": 0.7841,
+      "step": 3454
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.340041225814349e-05,
+      "loss": 0.7964,
+      "step": 3455
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.334529159713388e-05,
+      "loss": 0.8413,
+      "step": 3456
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3290189049088416e-05,
+      "loss": 0.7651,
+      "step": 3457
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.323510463539989e-05,
+      "loss": 0.8074,
+      "step": 3458
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3180038377453826e-05,
+      "loss": 0.7885,
+      "step": 3459
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3124990296628976e-05,
+      "loss": 0.7859,
+      "step": 3460
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.306996041429687e-05,
+      "loss": 0.8069,
+      "step": 3461
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3014948751821915e-05,
+      "loss": 0.7283,
+      "step": 3462
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.295995533056162e-05,
+      "loss": 0.823,
+      "step": 3463
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.290498017186631e-05,
+      "loss": 0.8394,
+      "step": 3464
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.285002329707923e-05,
+      "loss": 0.8315,
+      "step": 3465
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2795084727536536e-05,
+      "loss": 0.8101,
+      "step": 3466
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.274016448456724e-05,
+      "loss": 0.8206,
+      "step": 3467
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.268526258949331e-05,
+      "loss": 0.3004,
+      "step": 3468
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2630379063629534e-05,
+      "loss": 0.8237,
+      "step": 3469
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2575513928283594e-05,
+      "loss": 0.8171,
+      "step": 3470
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.252066720475597e-05,
+      "loss": 0.8284,
+      "step": 3471
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.246583891434018e-05,
+      "loss": 0.7896,
+      "step": 3472
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2411029078322314e-05,
+      "loss": 0.761,
+      "step": 3473
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.235623771798151e-05,
+      "loss": 0.7817,
+      "step": 3474
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2301464854589696e-05,
+      "loss": 0.3381,
+      "step": 3475
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.224671050941146e-05,
+      "loss": 0.7438,
+      "step": 3476
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.219197470370443e-05,
+      "loss": 0.8096,
+      "step": 3477
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2137257458718894e-05,
+      "loss": 0.7522,
+      "step": 3478
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.208255879569799e-05,
+      "loss": 0.783,
+      "step": 3479
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.20278787358776e-05,
+      "loss": 0.7935,
+      "step": 3480
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.197321730048641e-05,
+      "loss": 0.8132,
+      "step": 3481
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.191857451074587e-05,
+      "loss": 0.7401,
+      "step": 3482
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1863950387870176e-05,
+      "loss": 0.8389,
+      "step": 3483
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1809344953066374e-05,
+      "loss": 0.7761,
+      "step": 3484
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.175475822753404e-05,
+      "loss": 0.8115,
+      "step": 3485
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.170019023246574e-05,
+      "loss": 0.8376,
+      "step": 3486
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1645640989046595e-05,
+      "loss": 0.8459,
+      "step": 3487
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.159111051845451e-05,
+      "loss": 0.7555,
+      "step": 3488
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.153659884186013e-05,
+      "loss": 0.7744,
+      "step": 3489
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1482105980426643e-05,
+      "loss": 0.764,
+      "step": 3490
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.142763195531017e-05,
+      "loss": 0.7816,
+      "step": 3491
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1373176787659386e-05,
+      "loss": 0.7616,
+      "step": 3492
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1318740498615624e-05,
+      "loss": 0.8574,
+      "step": 3493
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.126432310931295e-05,
+      "loss": 0.8237,
+      "step": 3494
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.120992464087807e-05,
+      "loss": 0.8091,
+      "step": 3495
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.115554511443033e-05,
+      "loss": 0.7441,
+      "step": 3496
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.11011845510817e-05,
+      "loss": 0.8086,
+      "step": 3497
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1046842971936936e-05,
+      "loss": 0.7969,
+      "step": 3498
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.099252039809317e-05,
+      "loss": 0.8,
+      "step": 3499
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0938216850640396e-05,
+      "loss": 0.8425,
+      "step": 3500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0883932350661134e-05,
+      "loss": 0.8228,
+      "step": 3501
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.082966691923037e-05,
+      "loss": 0.8223,
+      "step": 3502
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.077542057741592e-05,
+      "loss": 0.7205,
+      "step": 3503
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.072119334627806e-05,
+      "loss": 0.8088,
+      "step": 3504
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.066698524686966e-05,
+      "loss": 0.8049,
+      "step": 3505
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0612796300236175e-05,
+      "loss": 0.8667,
+      "step": 3506
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.055862652741562e-05,
+      "loss": 0.8,
+      "step": 3507
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0504475949438555e-05,
+      "loss": 0.7828,
+      "step": 3508
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.045034458732808e-05,
+      "loss": 0.8105,
+      "step": 3509
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.039623246209995e-05,
+      "loss": 0.7601,
+      "step": 3510
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0342139594762216e-05,
+      "loss": 0.8035,
+      "step": 3511
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.028806600631569e-05,
+      "loss": 0.7817,
+      "step": 3512
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.023401171775357e-05,
+      "loss": 0.3508,
+      "step": 3513
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0179976750061605e-05,
+      "loss": 0.8037,
+      "step": 3514
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.012596112421806e-05,
+      "loss": 0.7976,
+      "step": 3515
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.007196486119354e-05,
+      "loss": 0.8557,
+      "step": 3516
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.001798798195136e-05,
+      "loss": 0.7518,
+      "step": 3517
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.996403050744719e-05,
+      "loss": 0.7959,
+      "step": 3518
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.991009245862917e-05,
+      "loss": 0.7809,
+      "step": 3519
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.985617385643789e-05,
+      "loss": 0.7572,
+      "step": 3520
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.980227472180643e-05,
+      "loss": 0.7706,
+      "step": 3521
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.974839507566027e-05,
+      "loss": 0.7434,
+      "step": 3522
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.969453493891732e-05,
+      "loss": 0.3005,
+      "step": 3523
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.964069433248807e-05,
+      "loss": 0.8916,
+      "step": 3524
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.95868732772751e-05,
+      "loss": 0.8718,
+      "step": 3525
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.953307179417376e-05,
+      "loss": 0.7507,
+      "step": 3526
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.947928990407156e-05,
+      "loss": 0.7791,
+      "step": 3527
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.94255276278485e-05,
+      "loss": 0.7795,
+      "step": 3528
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.937178498637696e-05,
+      "loss": 0.7949,
+      "step": 3529
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9318062000521644e-05,
+      "loss": 0.7346,
+      "step": 3530
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.926435869113971e-05,
+      "loss": 0.7234,
+      "step": 3531
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9210675079080607e-05,
+      "loss": 0.7721,
+      "step": 3532
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.915701118518615e-05,
+      "loss": 0.783,
+      "step": 3533
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9103367030290547e-05,
+      "loss": 0.7852,
+      "step": 3534
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9049742635220243e-05,
+      "loss": 0.3293,
+      "step": 3535
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.899613802079419e-05,
+      "loss": 0.7898,
+      "step": 3536
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8942553207823395e-05,
+      "loss": 0.7262,
+      "step": 3537
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8888988217111444e-05,
+      "loss": 0.7957,
+      "step": 3538
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.883544306945407e-05,
+      "loss": 0.8149,
+      "step": 3539
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.878191778563934e-05,
+      "loss": 0.7578,
+      "step": 3540
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8728412386447654e-05,
+      "loss": 0.7876,
+      "step": 3541
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.867492689265154e-05,
+      "loss": 0.7762,
+      "step": 3542
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.862146132501602e-05,
+      "loss": 0.6918,
+      "step": 3543
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.856801570429822e-05,
+      "loss": 0.2961,
+      "step": 3544
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.851459005124759e-05,
+      "loss": 0.8079,
+      "step": 3545
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8461184386605785e-05,
+      "loss": 0.7942,
+      "step": 3546
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8407798731106746e-05,
+      "loss": 0.7671,
+      "step": 3547
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8354433105476604e-05,
+      "loss": 0.349,
+      "step": 3548
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.83010875304337e-05,
+      "loss": 0.7587,
+      "step": 3549
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8247762026688746e-05,
+      "loss": 0.7847,
+      "step": 3550
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8194456614944375e-05,
+      "loss": 0.8125,
+      "step": 3551
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.814117131589569e-05,
+      "loss": 0.7283,
+      "step": 3552
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8087906150229866e-05,
+      "loss": 0.8352,
+      "step": 3553
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8034661138626255e-05,
+      "loss": 0.7183,
+      "step": 3554
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.798143630175642e-05,
+      "loss": 0.8359,
+      "step": 3555
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7928231660284054e-05,
+      "loss": 0.8301,
+      "step": 3556
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7875047234865045e-05,
+      "loss": 0.843,
+      "step": 3557
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.782188304614741e-05,
+      "loss": 0.7881,
+      "step": 3558
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.776873911477132e-05,
+      "loss": 0.7815,
+      "step": 3559
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.771561546136908e-05,
+      "loss": 0.7986,
+      "step": 3560
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.766251210656509e-05,
+      "loss": 0.8237,
+      "step": 3561
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7609429070976006e-05,
+      "loss": 0.7786,
+      "step": 3562
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7556366375210345e-05,
+      "loss": 0.7876,
+      "step": 3563
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.750332403986901e-05,
+      "loss": 0.7343,
+      "step": 3564
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.745030208554473e-05,
+      "loss": 0.8125,
+      "step": 3565
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.739730053282255e-05,
+      "loss": 0.7614,
+      "step": 3566
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.734431940227951e-05,
+      "loss": 0.7654,
+      "step": 3567
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.729135871448459e-05,
+      "loss": 0.8274,
+      "step": 3568
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.723841848999907e-05,
+      "loss": 0.7825,
+      "step": 3569
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7185498749376125e-05,
+      "loss": 0.8169,
+      "step": 3570
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.713259951316103e-05,
+      "loss": 0.8445,
+      "step": 3571
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7079720801891055e-05,
+      "loss": 0.7751,
+      "step": 3572
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.702686263609559e-05,
+      "loss": 0.3225,
+      "step": 3573
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6974025036295954e-05,
+      "loss": 0.7805,
+      "step": 3574
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6921208023005505e-05,
+      "loss": 0.8311,
+      "step": 3575
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6868411616729734e-05,
+      "loss": 0.7386,
+      "step": 3576
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.681563583796587e-05,
+      "loss": 0.7859,
+      "step": 3577
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6762880707203404e-05,
+      "loss": 0.7897,
+      "step": 3578
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.671014624492365e-05,
+      "loss": 0.7422,
+      "step": 3579
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.665743247159995e-05,
+      "loss": 0.756,
+      "step": 3580
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.660473940769761e-05,
+      "loss": 0.781,
+      "step": 3581
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6552067073673876e-05,
+      "loss": 0.8604,
+      "step": 3582
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.649941548997797e-05,
+      "loss": 0.7399,
+      "step": 3583
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.644678467705101e-05,
+      "loss": 0.7784,
+      "step": 3584
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6394174655326216e-05,
+      "loss": 0.8074,
+      "step": 3585
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6341585445228485e-05,
+      "loss": 0.3293,
+      "step": 3586
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.628901706717476e-05,
+      "loss": 0.7361,
+      "step": 3587
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.623646954157399e-05,
+      "loss": 0.7925,
+      "step": 3588
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.618394288882681e-05,
+      "loss": 0.3269,
+      "step": 3589
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6131437129326025e-05,
+      "loss": 0.8069,
+      "step": 3590
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6078952283456024e-05,
+      "loss": 0.8174,
+      "step": 3591
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6026488371593325e-05,
+      "loss": 0.8528,
+      "step": 3592
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5974045414106216e-05,
+      "loss": 0.801,
+      "step": 3593
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.592162343135483e-05,
+      "loss": 0.7761,
+      "step": 3594
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.586922244369122e-05,
+      "loss": 0.8142,
+      "step": 3595
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.581684247145922e-05,
+      "loss": 0.7815,
+      "step": 3596
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.576448353499456e-05,
+      "loss": 0.7145,
+      "step": 3597
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5712145654624774e-05,
+      "loss": 0.811,
+      "step": 3598
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.565982885066923e-05,
+      "loss": 0.7922,
+      "step": 3599
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.560753314343912e-05,
+      "loss": 0.7656,
+      "step": 3600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.555525855323738e-05,
+      "loss": 0.7797,
+      "step": 3601
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5503005100358945e-05,
+      "loss": 0.3674,
+      "step": 3602
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5450772805090226e-05,
+      "loss": 0.8466,
+      "step": 3603
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.539856168770974e-05,
+      "loss": 0.8416,
+      "step": 3604
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5346371768487586e-05,
+      "loss": 0.8052,
+      "step": 3605
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.529420306768569e-05,
+      "loss": 0.7734,
+      "step": 3606
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.524205560555774e-05,
+      "loss": 0.7782,
+      "step": 3607
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.518992940234917e-05,
+      "loss": 0.7759,
+      "step": 3608
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5137824478297177e-05,
+      "loss": 0.7562,
+      "step": 3609
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.508574085363064e-05,
+      "loss": 0.7908,
+      "step": 3610
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5033678548570346e-05,
+      "loss": 0.825,
+      "step": 3611
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.498163758332853e-05,
+      "loss": 0.8208,
+      "step": 3612
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4929617978109315e-05,
+      "loss": 0.8083,
+      "step": 3613
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.48776197531086e-05,
+      "loss": 0.7915,
+      "step": 3614
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4825642928513744e-05,
+      "loss": 0.7809,
+      "step": 3615
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4773687524504084e-05,
+      "loss": 0.8428,
+      "step": 3616
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4721753561250354e-05,
+      "loss": 0.7969,
+      "step": 3617
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.466984105891521e-05,
+      "loss": 0.8184,
+      "step": 3618
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4617950037652846e-05,
+      "loss": 0.7548,
+      "step": 3619
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4566080517609134e-05,
+      "loss": 0.3434,
+      "step": 3620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4514232518921605e-05,
+      "loss": 0.8218,
+      "step": 3621
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4462406061719455e-05,
+      "loss": 0.8246,
+      "step": 3622
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.441060116612348e-05,
+      "loss": 0.7739,
+      "step": 3623
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.435881785224613e-05,
+      "loss": 0.8125,
+      "step": 3624
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.430705614019147e-05,
+      "loss": 0.8535,
+      "step": 3625
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.425531605005518e-05,
+      "loss": 0.7415,
+      "step": 3626
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.420359760192452e-05,
+      "loss": 0.7572,
+      "step": 3627
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.415190081587845e-05,
+      "loss": 0.3195,
+      "step": 3628
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4100225711987344e-05,
+      "loss": 0.8136,
+      "step": 3629
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.404857231031332e-05,
+      "loss": 0.694,
+      "step": 3630
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3996940630909985e-05,
+      "loss": 0.7788,
+      "step": 3631
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.394533069382255e-05,
+      "loss": 0.8401,
+      "step": 3632
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3893742519087754e-05,
+      "loss": 0.7618,
+      "step": 3633
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3842176126733916e-05,
+      "loss": 0.3618,
+      "step": 3634
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.379063153678087e-05,
+      "loss": 0.8555,
+      "step": 3635
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.373910876923997e-05,
+      "loss": 0.8228,
+      "step": 3636
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.368760784411423e-05,
+      "loss": 0.7651,
+      "step": 3637
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.363612878139799e-05,
+      "loss": 0.856,
+      "step": 3638
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.358467160107722e-05,
+      "loss": 0.8,
+      "step": 3639
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.353323632312938e-05,
+      "loss": 0.7671,
+      "step": 3640
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.348182296752336e-05,
+      "loss": 0.7249,
+      "step": 3641
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.343043155421971e-05,
+      "loss": 0.7617,
+      "step": 3642
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.337906210317021e-05,
+      "loss": 0.7888,
+      "step": 3643
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.332771463431836e-05,
+      "loss": 0.8015,
+      "step": 3644
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3276389167598975e-05,
+      "loss": 0.802,
+      "step": 3645
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.322508572293835e-05,
+      "loss": 0.7756,
+      "step": 3646
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3173804320254275e-05,
+      "loss": 0.8127,
+      "step": 3647
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.312254497945595e-05,
+      "loss": 0.7896,
+      "step": 3648
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.307130772044401e-05,
+      "loss": 0.7791,
+      "step": 3649
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.302009256311048e-05,
+      "loss": 0.7441,
+      "step": 3650
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.296889952733898e-05,
+      "loss": 0.8652,
+      "step": 3651
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.291772863300427e-05,
+      "loss": 0.7404,
+      "step": 3652
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.286657989997268e-05,
+      "loss": 0.7683,
+      "step": 3653
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.281545334810201e-05,
+      "loss": 0.7812,
+      "step": 3654
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.276434899724119e-05,
+      "loss": 0.3401,
+      "step": 3655
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2713266867230804e-05,
+      "loss": 0.8135,
+      "step": 3656
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2662206977902664e-05,
+      "loss": 0.7719,
+      "step": 3657
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.261116934907998e-05,
+      "loss": 0.7881,
+      "step": 3658
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2560154000577304e-05,
+      "loss": 0.7903,
+      "step": 3659
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.250916095220056e-05,
+      "loss": 0.7751,
+      "step": 3660
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2458190223747e-05,
+      "loss": 0.3473,
+      "step": 3661
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2407241835005176e-05,
+      "loss": 0.7725,
+      "step": 3662
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.235631580575513e-05,
+      "loss": 0.7881,
+      "step": 3663
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.230541215576798e-05,
+      "loss": 0.7939,
+      "step": 3664
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.225453090480631e-05,
+      "loss": 0.7469,
+      "step": 3665
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.220367207262398e-05,
+      "loss": 0.767,
+      "step": 3666
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2152835678966096e-05,
+      "loss": 0.7341,
+      "step": 3667
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.210202174356922e-05,
+      "loss": 0.843,
+      "step": 3668
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.20512302861609e-05,
+      "loss": 0.8132,
+      "step": 3669
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.200046132646027e-05,
+      "loss": 0.7761,
+      "step": 3670
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1949714884177525e-05,
+      "loss": 0.761,
+      "step": 3671
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1898990979014205e-05,
+      "loss": 0.7521,
+      "step": 3672
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.184828963066305e-05,
+      "loss": 0.6649,
+      "step": 3673
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.179761085880809e-05,
+      "loss": 0.751,
+      "step": 3674
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.174695468312456e-05,
+      "loss": 0.3344,
+      "step": 3675
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.16963211232789e-05,
+      "loss": 0.7927,
+      "step": 3676
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.16457101989289e-05,
+      "loss": 0.75,
+      "step": 3677
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.159512192972337e-05,
+      "loss": 0.7491,
+      "step": 3678
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.15445563353024e-05,
+      "loss": 0.8333,
+      "step": 3679
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1494013435297416e-05,
+      "loss": 0.8083,
+      "step": 3680
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1443493249330766e-05,
+      "loss": 0.7395,
+      "step": 3681
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.139299579701623e-05,
+      "loss": 0.7588,
+      "step": 3682
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.134252109795863e-05,
+      "loss": 0.7233,
+      "step": 3683
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.129206917175397e-05,
+      "loss": 0.8445,
+      "step": 3684
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.124164003798944e-05,
+      "loss": 0.8044,
+      "step": 3685
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.119123371624335e-05,
+      "loss": 0.7625,
+      "step": 3686
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.114085022608517e-05,
+      "loss": 0.7474,
+      "step": 3687
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.109048958707552e-05,
+      "loss": 0.8196,
+      "step": 3688
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.104015181876612e-05,
+      "loss": 0.73,
+      "step": 3689
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.09898369406998e-05,
+      "loss": 0.752,
+      "step": 3690
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.093954497241064e-05,
+      "loss": 0.8125,
+      "step": 3691
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.088927593342358e-05,
+      "loss": 0.8059,
+      "step": 3692
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0839029843254814e-05,
+      "loss": 0.8232,
+      "step": 3693
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.078880672141171e-05,
+      "loss": 0.8325,
+      "step": 3694
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.073860658739246e-05,
+      "loss": 0.7148,
+      "step": 3695
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0688429460686605e-05,
+      "loss": 0.7881,
+      "step": 3696
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0638275360774594e-05,
+      "loss": 0.8438,
+      "step": 3697
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.058814430712796e-05,
+      "loss": 0.74,
+      "step": 3698
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.053803631920933e-05,
+      "loss": 0.8328,
+      "step": 3699
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0487951416472326e-05,
+      "loss": 0.7649,
+      "step": 3700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0437889618361635e-05,
+      "loss": 0.7639,
+      "step": 3701
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0387850944312957e-05,
+      "loss": 0.7827,
+      "step": 3702
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.033783541375311e-05,
+      "loss": 0.8157,
+      "step": 3703
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.028784304609976e-05,
+      "loss": 0.8025,
+      "step": 3704
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0237873860761646e-05,
+      "loss": 0.8252,
+      "step": 3705
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.018792787713864e-05,
+      "loss": 0.7584,
+      "step": 3706
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.013800511462135e-05,
+      "loss": 0.7906,
+      "step": 3707
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0088105592591616e-05,
+      "loss": 0.7993,
+      "step": 3708
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.003822933042213e-05,
+      "loss": 0.7588,
+      "step": 3709
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.998837634747655e-05,
+      "loss": 0.7578,
+      "step": 3710
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9938546663109544e-05,
+      "loss": 0.7861,
+      "step": 3711
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9888740296666696e-05,
+      "loss": 0.8328,
+      "step": 3712
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.983895726748455e-05,
+      "loss": 0.7777,
+      "step": 3713
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9789197594890606e-05,
+      "loss": 0.7682,
+      "step": 3714
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.973946129820326e-05,
+      "loss": 0.8248,
+      "step": 3715
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9689748396731854e-05,
+      "loss": 0.8116,
+      "step": 3716
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.964005890977672e-05,
+      "loss": 0.8623,
+      "step": 3717
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.959039285662894e-05,
+      "loss": 0.7888,
+      "step": 3718
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.954075025657058e-05,
+      "loss": 0.7709,
+      "step": 3719
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.949113112887471e-05,
+      "loss": 0.8604,
+      "step": 3720
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.944153549280506e-05,
+      "loss": 0.7736,
+      "step": 3721
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.939196336761645e-05,
+      "loss": 0.8071,
+      "step": 3722
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.934241477255445e-05,
+      "loss": 0.8325,
+      "step": 3723
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.929288972685554e-05,
+      "loss": 0.8387,
+      "step": 3724
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9243388249747046e-05,
+      "loss": 0.8328,
+      "step": 3725
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.919391036044715e-05,
+      "loss": 0.8159,
+      "step": 3726
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.914445607816486e-05,
+      "loss": 0.7507,
+      "step": 3727
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9095025422100003e-05,
+      "loss": 0.7678,
+      "step": 3728
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9045618411443374e-05,
+      "loss": 0.785,
+      "step": 3729
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8996235065376354e-05,
+      "loss": 0.7827,
+      "step": 3730
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.894687540307127e-05,
+      "loss": 0.7507,
+      "step": 3731
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.889753944369135e-05,
+      "loss": 0.3478,
+      "step": 3732
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8848227206390364e-05,
+      "loss": 0.8014,
+      "step": 3733
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.879893871031314e-05,
+      "loss": 0.8052,
+      "step": 3734
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8749673974595105e-05,
+      "loss": 0.8245,
+      "step": 3735
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.870043301836256e-05,
+      "loss": 0.8252,
+      "step": 3736
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.865121586073251e-05,
+      "loss": 0.7776,
+      "step": 3737
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8602022520812764e-05,
+      "loss": 0.73,
+      "step": 3738
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.855285301770187e-05,
+      "loss": 0.7737,
+      "step": 3739
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.850370737048913e-05,
+      "loss": 0.7697,
+      "step": 3740
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.845458559825457e-05,
+      "loss": 0.8789,
+      "step": 3741
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.840548772006891e-05,
+      "loss": 0.7629,
+      "step": 3742
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.835641375499375e-05,
+      "loss": 0.8188,
+      "step": 3743
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.830736372208118e-05,
+      "loss": 0.8354,
+      "step": 3744
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.825833764037412e-05,
+      "loss": 0.8066,
+      "step": 3745
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.820933552890629e-05,
+      "loss": 0.7527,
+      "step": 3746
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.816035740670185e-05,
+      "loss": 0.3556,
+      "step": 3747
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8111403292775905e-05,
+      "loss": 0.7974,
+      "step": 3748
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8062473206134083e-05,
+      "loss": 0.8062,
+      "step": 3749
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.801356716577273e-05,
+      "loss": 0.7852,
+      "step": 3750
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.796468519067887e-05,
+      "loss": 0.8137,
+      "step": 3751
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7915827299830155e-05,
+      "loss": 0.7847,
+      "step": 3752
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.786699351219489e-05,
+      "loss": 0.7753,
+      "step": 3753
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7818183846732024e-05,
+      "loss": 0.8015,
+      "step": 3754
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.776939832239125e-05,
+      "loss": 0.7566,
+      "step": 3755
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.772063695811262e-05,
+      "loss": 0.8333,
+      "step": 3756
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.767189977282711e-05,
+      "loss": 0.8462,
+      "step": 3757
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.762318678545616e-05,
+      "loss": 0.8052,
+      "step": 3758
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7574498014911716e-05,
+      "loss": 0.8389,
+      "step": 3759
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.752583348009657e-05,
+      "loss": 0.7826,
+      "step": 3760
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7477193199903904e-05,
+      "loss": 0.325,
+      "step": 3761
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.742857719321756e-05,
+      "loss": 0.7955,
+      "step": 3762
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.737998547891195e-05,
+      "loss": 0.8015,
+      "step": 3763
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7331418075852053e-05,
+      "loss": 0.7917,
+      "step": 3764
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.728287500289339e-05,
+      "loss": 0.8162,
+      "step": 3765
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.723435627888208e-05,
+      "loss": 0.8149,
+      "step": 3766
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.718586192265473e-05,
+      "loss": 0.7804,
+      "step": 3767
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7137391953038514e-05,
+      "loss": 0.8235,
+      "step": 3768
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.708894638885122e-05,
+      "loss": 0.8362,
+      "step": 3769
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7040525248901005e-05,
+      "loss": 0.3303,
+      "step": 3770
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6992128551986616e-05,
+      "loss": 0.7864,
+      "step": 3771
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6943756316897404e-05,
+      "loss": 0.7798,
+      "step": 3772
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6895408562413025e-05,
+      "loss": 0.7625,
+      "step": 3773
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.684708530730382e-05,
+      "loss": 0.7681,
+      "step": 3774
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6798786570330525e-05,
+      "loss": 0.7933,
+      "step": 3775
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.675051237024436e-05,
+      "loss": 0.7412,
+      "step": 3776
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.670226272578704e-05,
+      "loss": 0.7869,
+      "step": 3777
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.665403765569073e-05,
+      "loss": 0.8413,
+      "step": 3778
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.660583717867807e-05,
+      "loss": 0.8447,
+      "step": 3779
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.655766131346211e-05,
+      "loss": 0.7822,
+      "step": 3780
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.650951007874648e-05,
+      "loss": 0.8062,
+      "step": 3781
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.646138349322501e-05,
+      "loss": 0.7612,
+      "step": 3782
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.641328157558219e-05,
+      "loss": 0.8435,
+      "step": 3783
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6365204344492867e-05,
+      "loss": 0.7539,
+      "step": 3784
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.631715181862215e-05,
+      "loss": 0.809,
+      "step": 3785
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.62691240166258e-05,
+      "loss": 0.7959,
+      "step": 3786
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6221120957149825e-05,
+      "loss": 0.7864,
+      "step": 3787
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.617314265883066e-05,
+      "loss": 0.8665,
+      "step": 3788
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6125189140295145e-05,
+      "loss": 0.7888,
+      "step": 3789
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.607726042016049e-05,
+      "loss": 0.8225,
+      "step": 3790
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.602935651703424e-05,
+      "loss": 0.863,
+      "step": 3791
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.598147744951438e-05,
+      "loss": 0.7775,
+      "step": 3792
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5933623236189195e-05,
+      "loss": 0.7878,
+      "step": 3793
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.58857938956373e-05,
+      "loss": 0.7419,
+      "step": 3794
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.58379894464278e-05,
+      "loss": 0.8379,
+      "step": 3795
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.57902099071199e-05,
+      "loss": 0.8293,
+      "step": 3796
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.574245529626334e-05,
+      "loss": 0.7756,
+      "step": 3797
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.569472563239814e-05,
+      "loss": 0.7899,
+      "step": 3798
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5647020934054464e-05,
+      "loss": 0.7639,
+      "step": 3799
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.559934121975304e-05,
+      "loss": 0.8069,
+      "step": 3800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.555168650800473e-05,
+      "loss": 0.8169,
+      "step": 3801
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5504056817310735e-05,
+      "loss": 0.8186,
+      "step": 3802
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5456452166162545e-05,
+      "loss": 0.8577,
+      "step": 3803
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.540887257304193e-05,
+      "loss": 0.7952,
+      "step": 3804
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5361318056420925e-05,
+      "loss": 0.8264,
+      "step": 3805
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.531378863476178e-05,
+      "loss": 0.7561,
+      "step": 3806
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.526628432651716e-05,
+      "loss": 0.7186,
+      "step": 3807
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5218805150129755e-05,
+      "loss": 0.7703,
+      "step": 3808
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.51713511240327e-05,
+      "loss": 0.7865,
+      "step": 3809
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5123922266649303e-05,
+      "loss": 0.3479,
+      "step": 3810
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.507651859639295e-05,
+      "loss": 0.7542,
+      "step": 3811
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.502914013166749e-05,
+      "loss": 0.7957,
+      "step": 3812
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4981786890866854e-05,
+      "loss": 0.7944,
+      "step": 3813
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.493445889237518e-05,
+      "loss": 0.7998,
+      "step": 3814
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.488715615456685e-05,
+      "loss": 0.3229,
+      "step": 3815
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.483987869580638e-05,
+      "loss": 0.7241,
+      "step": 3816
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4792626534448544e-05,
+      "loss": 0.7753,
+      "step": 3817
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.474539968883824e-05,
+      "loss": 0.7622,
+      "step": 3818
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.469819817731056e-05,
+      "loss": 0.7881,
+      "step": 3819
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4651022018190715e-05,
+      "loss": 0.8186,
+      "step": 3820
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.460387122979423e-05,
+      "loss": 0.3469,
+      "step": 3821
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.455674583042652e-05,
+      "loss": 0.7561,
+      "step": 3822
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.450964583838339e-05,
+      "loss": 0.3403,
+      "step": 3823
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.446257127195067e-05,
+      "loss": 0.7275,
+      "step": 3824
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4415522149404235e-05,
+      "loss": 0.7017,
+      "step": 3825
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4368498489010283e-05,
+      "loss": 0.7632,
+      "step": 3826
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.432150030902497e-05,
+      "loss": 0.8445,
+      "step": 3827
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.427452762769462e-05,
+      "loss": 0.7966,
+      "step": 3828
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.422758046325562e-05,
+      "loss": 0.8223,
+      "step": 3829
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.418065883393452e-05,
+      "loss": 0.7671,
+      "step": 3830
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4133762757947874e-05,
+      "loss": 0.7549,
+      "step": 3831
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.408689225350234e-05,
+      "loss": 0.8218,
+      "step": 3832
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4040047338794756e-05,
+      "loss": 0.8011,
+      "step": 3833
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.399322803201178e-05,
+      "loss": 0.8369,
+      "step": 3834
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.394643435133041e-05,
+      "loss": 0.8064,
+      "step": 3835
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.389966631491751e-05,
+      "loss": 0.7979,
+      "step": 3836
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.385292394093006e-05,
+      "loss": 0.8479,
+      "step": 3837
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.380620724751506e-05,
+      "loss": 0.801,
+      "step": 3838
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3759516252809477e-05,
+      "loss": 0.7764,
+      "step": 3839
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3712850974940435e-05,
+      "loss": 0.7949,
+      "step": 3840
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.366621143202498e-05,
+      "loss": 0.7239,
+      "step": 3841
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.361959764217017e-05,
+      "loss": 0.7479,
+      "step": 3842
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3573009623473126e-05,
+      "loss": 0.7849,
+      "step": 3843
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.352644739402089e-05,
+      "loss": 0.7152,
+      "step": 3844
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3479910971890515e-05,
+      "loss": 0.7844,
+      "step": 3845
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3433400375149026e-05,
+      "loss": 0.8616,
+      "step": 3846
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.338691562185353e-05,
+      "loss": 0.7787,
+      "step": 3847
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3340456730050884e-05,
+      "loss": 0.7495,
+      "step": 3848
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3294023717778125e-05,
+      "loss": 0.7976,
+      "step": 3849
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.324761660306215e-05,
+      "loss": 0.8191,
+      "step": 3850
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.320123540391968e-05,
+      "loss": 0.7802,
+      "step": 3851
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.315488013835762e-05,
+      "loss": 0.7584,
+      "step": 3852
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.310855082437263e-05,
+      "loss": 0.8386,
+      "step": 3853
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.306224747995136e-05,
+      "loss": 0.8191,
+      "step": 3854
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.301597012307034e-05,
+      "loss": 0.823,
+      "step": 3855
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.296971877169605e-05,
+      "loss": 0.7522,
+      "step": 3856
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.292349344378486e-05,
+      "loss": 0.8303,
+      "step": 3857
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.287729415728298e-05,
+      "loss": 0.8181,
+      "step": 3858
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.283112093012669e-05,
+      "loss": 0.7727,
+      "step": 3859
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.278497378024187e-05,
+      "loss": 0.8088,
+      "step": 3860
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2738852725544544e-05,
+      "loss": 0.7822,
+      "step": 3861
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.269275778394047e-05,
+      "loss": 0.8035,
+      "step": 3862
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.264668897332527e-05,
+      "loss": 0.7737,
+      "step": 3863
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.260064631158449e-05,
+      "loss": 0.7671,
+      "step": 3864
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2554629816593376e-05,
+      "loss": 0.7469,
+      "step": 3865
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.250863950621721e-05,
+      "loss": 0.7948,
+      "step": 3866
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2462675398310984e-05,
+      "loss": 0.7432,
+      "step": 3867
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2416737510719544e-05,
+      "loss": 0.782,
+      "step": 3868
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.237082586127757e-05,
+      "loss": 0.761,
+      "step": 3869
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.232494046780953e-05,
+      "loss": 0.8191,
+      "step": 3870
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.227908134812972e-05,
+      "loss": 0.7759,
+      "step": 3871
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.223324852004219e-05,
+      "loss": 0.7373,
+      "step": 3872
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.218744200134094e-05,
+      "loss": 0.8196,
+      "step": 3873
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.21416618098095e-05,
+      "loss": 0.769,
+      "step": 3874
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2095907963221396e-05,
+      "loss": 0.791,
+      "step": 3875
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.205018047933987e-05,
+      "loss": 0.3274,
+      "step": 3876
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.200447937591779e-05,
+      "loss": 0.7615,
+      "step": 3877
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.195880467069801e-05,
+      "loss": 0.7787,
+      "step": 3878
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.191315638141297e-05,
+      "loss": 0.3394,
+      "step": 3879
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1867534525784934e-05,
+      "loss": 0.7842,
+      "step": 3880
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.182193912152586e-05,
+      "loss": 0.3347,
+      "step": 3881
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.177637018633746e-05,
+      "loss": 0.7915,
+      "step": 3882
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.173082773791116e-05,
+      "loss": 0.8684,
+      "step": 3883
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.168531179392808e-05,
+      "loss": 0.7976,
+      "step": 3884
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.163982237205917e-05,
+      "loss": 0.7598,
+      "step": 3885
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.159435948996485e-05,
+      "loss": 0.8118,
+      "step": 3886
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.154892316529549e-05,
+      "loss": 0.7734,
+      "step": 3887
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.150351341569101e-05,
+      "loss": 0.7681,
+      "step": 3888
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1458130258781005e-05,
+      "loss": 0.3416,
+      "step": 3889
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.141277371218484e-05,
+      "loss": 0.8271,
+      "step": 3890
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1367443793511386e-05,
+      "loss": 0.8108,
+      "step": 3891
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.132214052035937e-05,
+      "loss": 0.348,
+      "step": 3892
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.127686391031705e-05,
+      "loss": 0.7533,
+      "step": 3893
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.123161398096237e-05,
+      "loss": 0.8197,
+      "step": 3894
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.11863907498629e-05,
+      "loss": 0.8035,
+      "step": 3895
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.114119423457588e-05,
+      "loss": 0.7976,
+      "step": 3896
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.109602445264812e-05,
+      "loss": 0.7515,
+      "step": 3897
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1050881421616076e-05,
+      "loss": 0.8569,
+      "step": 3898
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1005765159005904e-05,
+      "loss": 0.7942,
+      "step": 3899
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0960675682333186e-05,
+      "loss": 0.8035,
+      "step": 3900
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0915613009103295e-05,
+      "loss": 0.7701,
+      "step": 3901
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.087057715681107e-05,
+      "loss": 0.7903,
+      "step": 3902
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0825568142940995e-05,
+      "loss": 0.7639,
+      "step": 3903
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.078058598496711e-05,
+      "loss": 0.7491,
+      "step": 3904
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0735630700353044e-05,
+      "loss": 0.7798,
+      "step": 3905
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.069070230655198e-05,
+      "loss": 0.7659,
+      "step": 3906
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0645800821006665e-05,
+      "loss": 0.7326,
+      "step": 3907
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.060092626114941e-05,
+      "loss": 0.7275,
+      "step": 3908
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.055607864440206e-05,
+      "loss": 0.7772,
+      "step": 3909
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0511257988175978e-05,
+      "loss": 0.7869,
+      "step": 3910
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0466464309872166e-05,
+      "loss": 0.8169,
+      "step": 3911
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0421697626880962e-05,
+      "loss": 0.7957,
+      "step": 3912
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0376957956582452e-05,
+      "loss": 0.7606,
+      "step": 3913
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0332245316345996e-05,
+      "loss": 0.3403,
+      "step": 3914
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0287559723530667e-05,
+      "loss": 0.7551,
+      "step": 3915
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0242901195484953e-05,
+      "loss": 0.8462,
+      "step": 3916
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0198269749546737e-05,
+      "loss": 0.7445,
+      "step": 3917
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0153665403043586e-05,
+      "loss": 0.7617,
+      "step": 3918
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.01090881732924e-05,
+      "loss": 0.7777,
+      "step": 3919
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0064538077599603e-05,
+      "loss": 0.7858,
+      "step": 3920
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.002001513326107e-05,
+      "loss": 0.8179,
+      "step": 3921
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9975519357562155e-05,
+      "loss": 0.7974,
+      "step": 3922
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9931050767777625e-05,
+      "loss": 0.8516,
+      "step": 3923
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9886609381171705e-05,
+      "loss": 0.8196,
+      "step": 3924
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9842195214998158e-05,
+      "loss": 0.8027,
+      "step": 3925
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9797808286499973e-05,
+      "loss": 0.7878,
+      "step": 3926
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9753448612909775e-05,
+      "loss": 0.8348,
+      "step": 3927
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9709116211449483e-05,
+      "loss": 0.7346,
+      "step": 3928
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.966481109933047e-05,
+      "loss": 0.7551,
+      "step": 3929
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9620533293753495e-05,
+      "loss": 0.7607,
+      "step": 3930
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9576282811908728e-05,
+      "loss": 0.8269,
+      "step": 3931
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9532059670975732e-05,
+      "loss": 0.7668,
+      "step": 3932
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9487863888123456e-05,
+      "loss": 0.7986,
+      "step": 3933
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9443695480510224e-05,
+      "loss": 0.7579,
+      "step": 3934
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.939955446528374e-05,
+      "loss": 0.7993,
+      "step": 3935
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9355440859581018e-05,
+      "loss": 0.739,
+      "step": 3936
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.931135468052858e-05,
+      "loss": 0.7646,
+      "step": 3937
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9267295945242067e-05,
+      "loss": 0.3557,
+      "step": 3938
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9223264670826743e-05,
+      "loss": 0.775,
+      "step": 3939
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9179260874376913e-05,
+      "loss": 0.8459,
+      "step": 3940
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9135284572976486e-05,
+      "loss": 0.8051,
+      "step": 3941
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9091335783698515e-05,
+      "loss": 0.7463,
+      "step": 3942
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9047414523605465e-05,
+      "loss": 0.3217,
+      "step": 3943
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9003520809749053e-05,
+      "loss": 0.822,
+      "step": 3944
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8959654659170354e-05,
+      "loss": 0.8132,
+      "step": 3945
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8915816088899695e-05,
+      "loss": 0.7988,
+      "step": 3946
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8872005115956745e-05,
+      "loss": 0.7576,
+      "step": 3947
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8828221757350404e-05,
+      "loss": 0.792,
+      "step": 3948
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8784466030078905e-05,
+      "loss": 0.8174,
+      "step": 3949
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.874073795112967e-05,
+      "loss": 0.7681,
+      "step": 3950
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8697037537479564e-05,
+      "loss": 0.8041,
+      "step": 3951
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8653364806094453e-05,
+      "loss": 0.8125,
+      "step": 3952
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.86097197739297e-05,
+      "loss": 0.3467,
+      "step": 3953
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8566102457929755e-05,
+      "loss": 0.7778,
+      "step": 3954
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8522512875028395e-05,
+      "loss": 0.7449,
+      "step": 3955
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.847895104214856e-05,
+      "loss": 0.8357,
+      "step": 3956
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.843541697620249e-05,
+      "loss": 0.8237,
+      "step": 3957
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8391910694091583e-05,
+      "loss": 0.7964,
+      "step": 3958
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8348432212706445e-05,
+      "loss": 0.8267,
+      "step": 3959
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8304981548927022e-05,
+      "loss": 0.7927,
+      "step": 3960
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.826155871962227e-05,
+      "loss": 0.262,
+      "step": 3961
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8218163741650416e-05,
+      "loss": 0.8208,
+      "step": 3962
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8174796631858978e-05,
+      "loss": 0.8213,
+      "step": 3963
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.813145740708445e-05,
+      "loss": 0.8127,
+      "step": 3964
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8088146084152712e-05,
+      "loss": 0.7351,
+      "step": 3965
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8044862679878604e-05,
+      "loss": 0.8403,
+      "step": 3966
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.800160721106633e-05,
+      "loss": 0.7375,
+      "step": 3967
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7958379694509106e-05,
+      "loss": 0.7988,
+      "step": 3968
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.791518014698935e-05,
+      "loss": 0.7728,
+      "step": 3969
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.787200858527862e-05,
+      "loss": 0.7478,
+      "step": 3970
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7828865026137585e-05,
+      "loss": 0.8306,
+      "step": 3971
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.778574948631608e-05,
+      "loss": 0.8174,
+      "step": 3972
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.774266198255303e-05,
+      "loss": 0.7712,
+      "step": 3973
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7699602531576496e-05,
+      "loss": 0.7827,
+      "step": 3974
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.765657115010364e-05,
+      "loss": 0.8164,
+      "step": 3975
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7613567854840683e-05,
+      "loss": 0.8477,
+      "step": 3976
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7570592662483087e-05,
+      "loss": 0.8174,
+      "step": 3977
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7527645589715167e-05,
+      "loss": 0.792,
+      "step": 3978
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7484726653210558e-05,
+      "loss": 0.7583,
+      "step": 3979
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.744183586963185e-05,
+      "loss": 0.8245,
+      "step": 3980
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.739897325563069e-05,
+      "loss": 0.8235,
+      "step": 3981
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7356138827847855e-05,
+      "loss": 0.8108,
+      "step": 3982
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.731333260291311e-05,
+      "loss": 0.7365,
+      "step": 3983
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7270554597445342e-05,
+      "loss": 0.7795,
+      "step": 3984
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7227804828052384e-05,
+      "loss": 0.3428,
+      "step": 3985
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.718508331133128e-05,
+      "loss": 0.7889,
+      "step": 3986
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7142390063867896e-05,
+      "loss": 0.7965,
+      "step": 3987
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.709972510223725e-05,
+      "loss": 0.7521,
+      "step": 3988
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7057088443003343e-05,
+      "loss": 0.7317,
+      "step": 3989
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7014480102719176e-05,
+      "loss": 0.7311,
+      "step": 3990
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.697190009792685e-05,
+      "loss": 0.8467,
+      "step": 3991
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.692934844515729e-05,
+      "loss": 0.73,
+      "step": 3992
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6886825160930585e-05,
+      "loss": 0.8274,
+      "step": 3993
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6844330261755714e-05,
+      "loss": 0.7837,
+      "step": 3994
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6801863764130652e-05,
+      "loss": 0.7654,
+      "step": 3995
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.675942568454236e-05,
+      "loss": 0.7695,
+      "step": 3996
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.671701603946678e-05,
+      "loss": 0.7981,
+      "step": 3997
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.667463484536876e-05,
+      "loss": 0.7938,
+      "step": 3998
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6632282118702147e-05,
+      "loss": 0.8099,
+      "step": 3999
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6589957875909798e-05,
+      "loss": 0.7422,
+      "step": 4000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6547662133423345e-05,
+      "loss": 0.7444,
+      "step": 4001
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.650539490766346e-05,
+      "loss": 0.7493,
+      "step": 4002
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.646315621503983e-05,
+      "loss": 0.7522,
+      "step": 4003
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6420946071950847e-05,
+      "loss": 0.7585,
+      "step": 4004
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6378764494784024e-05,
+      "loss": 0.7894,
+      "step": 4005
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6336611499915686e-05,
+      "loss": 0.7874,
+      "step": 4006
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6294487103711065e-05,
+      "loss": 0.7877,
+      "step": 4007
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6252391322524293e-05,
+      "loss": 0.8115,
+      "step": 4008
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.621032417269843e-05,
+      "loss": 0.8267,
+      "step": 4009
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6168285670565373e-05,
+      "loss": 0.7656,
+      "step": 4010
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.612627583244589e-05,
+      "loss": 0.8042,
+      "step": 4011
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6084294674649735e-05,
+      "loss": 0.7045,
+      "step": 4012
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6042342213475347e-05,
+      "loss": 0.8347,
+      "step": 4013
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.600041846521014e-05,
+      "loss": 0.7849,
+      "step": 4014
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5958523446130378e-05,
+      "loss": 0.8137,
+      "step": 4015
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5916657172501103e-05,
+      "loss": 0.8064,
+      "step": 4016
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5874819660576334e-05,
+      "loss": 0.7654,
+      "step": 4017
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5833010926598722e-05,
+      "loss": 0.751,
+      "step": 4018
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.579123098679994e-05,
+      "loss": 0.7878,
+      "step": 4019
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5749479857400384e-05,
+      "loss": 0.8181,
+      "step": 4020
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5707757554609246e-05,
+      "loss": 0.8074,
+      "step": 4021
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5666064094624598e-05,
+      "loss": 0.8281,
+      "step": 4022
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5624399493633257e-05,
+      "loss": 0.7969,
+      "step": 4023
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5582763767810857e-05,
+      "loss": 0.8372,
+      "step": 4024
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5541156933321798e-05,
+      "loss": 0.7705,
+      "step": 4025
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5499579006319362e-05,
+      "loss": 0.8008,
+      "step": 4026
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5458030002945453e-05,
+      "loss": 0.8333,
+      "step": 4027
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5416509939330835e-05,
+      "loss": 0.803,
+      "step": 4028
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5375018831595087e-05,
+      "loss": 0.3419,
+      "step": 4029
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5333556695846383e-05,
+      "loss": 0.8081,
+      "step": 4030
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5292123548181845e-05,
+      "loss": 0.7728,
+      "step": 4031
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5250719404687216e-05,
+      "loss": 0.7682,
+      "step": 4032
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.520934428143701e-05,
+      "loss": 0.7947,
+      "step": 4033
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.516799819449447e-05,
+      "loss": 0.3488,
+      "step": 4034
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.512668115991156e-05,
+      "loss": 0.7654,
+      "step": 4035
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5085393193729e-05,
+      "loss": 0.7869,
+      "step": 4036
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5044134311976152e-05,
+      "loss": 0.8323,
+      "step": 4037
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5002904530671236e-05,
+      "loss": 0.7495,
+      "step": 4038
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4961703865820975e-05,
+      "loss": 0.8212,
+      "step": 4039
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.492053233342091e-05,
+      "loss": 0.7988,
+      "step": 4040
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4879389949455268e-05,
+      "loss": 0.7996,
+      "step": 4041
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.483827672989688e-05,
+      "loss": 0.7598,
+      "step": 4042
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4797192690707428e-05,
+      "loss": 0.3484,
+      "step": 4043
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4756137847837025e-05,
+      "loss": 0.7678,
+      "step": 4044
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4715112217224655e-05,
+      "loss": 0.7524,
+      "step": 4045
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.467411581479786e-05,
+      "loss": 0.8474,
+      "step": 4046
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.463314865647286e-05,
+      "loss": 0.7673,
+      "step": 4047
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.45922107581545e-05,
+      "loss": 0.8052,
+      "step": 4048
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4551302135736286e-05,
+      "loss": 0.8047,
+      "step": 4049
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4510422805100363e-05,
+      "loss": 0.7647,
+      "step": 4050
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4469572782117457e-05,
+      "loss": 0.7866,
+      "step": 4051
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4428752082647045e-05,
+      "loss": 0.7471,
+      "step": 4052
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.438796072253704e-05,
+      "loss": 0.7214,
+      "step": 4053
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4347198717624054e-05,
+      "loss": 0.7659,
+      "step": 4054
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4306466083733393e-05,
+      "loss": 0.8052,
+      "step": 4055
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.426576283667873e-05,
+      "loss": 0.7424,
+      "step": 4056
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.422508899226258e-05,
+      "loss": 0.8096,
+      "step": 4057
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4184444566275886e-05,
+      "loss": 0.8286,
+      "step": 4058
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4143829574498223e-05,
+      "loss": 0.7756,
+      "step": 4059
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4103244032697715e-05,
+      "loss": 0.3234,
+      "step": 4060
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.406268795663108e-05,
+      "loss": 0.781,
+      "step": 4061
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.402216136204357e-05,
+      "loss": 0.8202,
+      "step": 4062
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3981664264669023e-05,
+      "loss": 0.8135,
+      "step": 4063
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3941196680229794e-05,
+      "loss": 0.7476,
+      "step": 4064
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.390075862443677e-05,
+      "loss": 0.7827,
+      "step": 4065
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.386035011298947e-05,
+      "loss": 0.7922,
+      "step": 4066
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3819971161575803e-05,
+      "loss": 0.7888,
+      "step": 4067
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.377962178587225e-05,
+      "loss": 0.8171,
+      "step": 4068
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3739302001543918e-05,
+      "loss": 0.7466,
+      "step": 4069
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3699011824244232e-05,
+      "loss": 0.8716,
+      "step": 4070
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.365875126961531e-05,
+      "loss": 0.7661,
+      "step": 4071
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3618520353287643e-05,
+      "loss": 0.8274,
+      "step": 4072
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.357831909088026e-05,
+      "loss": 0.8279,
+      "step": 4073
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3538147498000695e-05,
+      "loss": 0.8145,
+      "step": 4074
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.349800559024492e-05,
+      "loss": 0.7959,
+      "step": 4075
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3457893383197416e-05,
+      "loss": 0.8286,
+      "step": 4076
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3417810892431103e-05,
+      "loss": 0.7856,
+      "step": 4077
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3377758133507456e-05,
+      "loss": 0.811,
+      "step": 4078
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3337735121976245e-05,
+      "loss": 0.8169,
+      "step": 4079
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3297741873375802e-05,
+      "loss": 0.7904,
+      "step": 4080
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3257778403232954e-05,
+      "loss": 0.7969,
+      "step": 4081
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.321784472706279e-05,
+      "loss": 0.7499,
+      "step": 4082
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3177940860369007e-05,
+      "loss": 0.8149,
+      "step": 4083
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3138066818643644e-05,
+      "loss": 0.7944,
+      "step": 4084
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3098222617367182e-05,
+      "loss": 0.7744,
+      "step": 4085
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.30584082720085e-05,
+      "loss": 0.8313,
+      "step": 4086
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3018623798024918e-05,
+      "loss": 0.7936,
+      "step": 4087
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.297886921086211e-05,
+      "loss": 0.7795,
+      "step": 4088
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.293914452595419e-05,
+      "loss": 0.7456,
+      "step": 4089
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2899449758723657e-05,
+      "loss": 0.8225,
+      "step": 4090
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.285978492458134e-05,
+      "loss": 0.7037,
+      "step": 4091
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.282015003892659e-05,
+      "loss": 0.8298,
+      "step": 4092
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2780545117146947e-05,
+      "loss": 0.7715,
+      "step": 4093
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2740970174618403e-05,
+      "loss": 0.813,
+      "step": 4094
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2701425226705408e-05,
+      "loss": 0.7609,
+      "step": 4095
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2661910288760547e-05,
+      "loss": 0.7825,
+      "step": 4096
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2622425376124967e-05,
+      "loss": 0.8054,
+      "step": 4097
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2582970504128042e-05,
+      "loss": 0.8513,
+      "step": 4098
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2543545688087518e-05,
+      "loss": 0.6781,
+      "step": 4099
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2504150943309453e-05,
+      "loss": 0.7791,
+      "step": 4100
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2464786285088268e-05,
+      "loss": 0.8184,
+      "step": 4101
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.242545172870665e-05,
+      "loss": 0.7585,
+      "step": 4102
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.238614728943561e-05,
+      "loss": 0.7764,
+      "step": 4103
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2346872982534584e-05,
+      "loss": 0.7239,
+      "step": 4104
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.230762882325108e-05,
+      "loss": 0.7195,
+      "step": 4105
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2268414826821115e-05,
+      "loss": 0.7775,
+      "step": 4106
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2229231008468932e-05,
+      "loss": 0.8403,
+      "step": 4107
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2190077383406936e-05,
+      "loss": 0.769,
+      "step": 4108
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2150953966835996e-05,
+      "loss": 0.7991,
+      "step": 4109
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.211186077394516e-05,
+      "loss": 0.7844,
+      "step": 4110
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2072797819911726e-05,
+      "loss": 0.8518,
+      "step": 4111
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2033765119901295e-05,
+      "loss": 0.735,
+      "step": 4112
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1994762689067704e-05,
+      "loss": 0.7367,
+      "step": 4113
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1955790542553033e-05,
+      "loss": 0.8083,
+      "step": 4114
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1916848695487614e-05,
+      "loss": 0.8196,
+      "step": 4115
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1877937162990014e-05,
+      "loss": 0.7614,
+      "step": 4116
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1839055960166997e-05,
+      "loss": 0.822,
+      "step": 4117
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.180020510211367e-05,
+      "loss": 0.8301,
+      "step": 4118
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1761384603913205e-05,
+      "loss": 0.7737,
+      "step": 4119
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1722594480637036e-05,
+      "loss": 0.8164,
+      "step": 4120
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.168383474734491e-05,
+      "loss": 0.7681,
+      "step": 4121
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1645105419084588e-05,
+      "loss": 0.7698,
+      "step": 4122
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1606406510892208e-05,
+      "loss": 0.8313,
+      "step": 4123
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1567738037791995e-05,
+      "loss": 0.3322,
+      "step": 4124
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.152910001479638e-05,
+      "loss": 0.7347,
+      "step": 4125
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1490492456905965e-05,
+      "loss": 0.7297,
+      "step": 4126
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1451915379109543e-05,
+      "loss": 0.7915,
+      "step": 4127
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.141336879638406e-05,
+      "loss": 0.8186,
+      "step": 4128
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1374852723694592e-05,
+      "loss": 0.7866,
+      "step": 4129
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1336367175994508e-05,
+      "loss": 0.791,
+      "step": 4130
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1297912168225088e-05,
+      "loss": 0.7451,
+      "step": 4131
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1259487715315997e-05,
+      "loss": 0.7347,
+      "step": 4132
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.12210938321849e-05,
+      "loss": 0.7362,
+      "step": 4133
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1182730533737573e-05,
+      "loss": 0.7731,
+      "step": 4134
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1144397834868035e-05,
+      "loss": 0.7664,
+      "step": 4135
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.110609575045833e-05,
+      "loss": 0.7825,
+      "step": 4136
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1067824295378658e-05,
+      "loss": 0.7993,
+      "step": 4137
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1029583484487314e-05,
+      "loss": 0.8137,
+      "step": 4138
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0991373332630682e-05,
+      "loss": 0.7991,
+      "step": 4139
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0953193854643273e-05,
+      "loss": 0.7998,
+      "step": 4140
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0915045065347672e-05,
+      "loss": 0.7797,
+      "step": 4141
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0876926979554546e-05,
+      "loss": 0.8218,
+      "step": 4142
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0838839612062634e-05,
+      "loss": 0.7776,
+      "step": 4143
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0800782977658838e-05,
+      "loss": 0.7815,
+      "step": 4144
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0762757091117936e-05,
+      "loss": 0.8364,
+      "step": 4145
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0724761967202987e-05,
+      "loss": 0.7798,
+      "step": 4146
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.068679762066499e-05,
+      "loss": 0.7523,
+      "step": 4147
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0648864066242933e-05,
+      "loss": 0.8448,
+      "step": 4148
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0610961318664013e-05,
+      "loss": 0.3409,
+      "step": 4149
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0573089392643364e-05,
+      "loss": 0.8013,
+      "step": 4150
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0535248302884146e-05,
+      "loss": 0.8296,
+      "step": 4151
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.04974380640776e-05,
+      "loss": 0.7114,
+      "step": 4152
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.045965869090295e-05,
+      "loss": 0.7861,
+      "step": 4153
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.042191019802745e-05,
+      "loss": 0.8042,
+      "step": 4154
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0384192600106335e-05,
+      "loss": 0.7996,
+      "step": 4155
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0346505911782954e-05,
+      "loss": 0.8184,
+      "step": 4156
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0308850147688486e-05,
+      "loss": 0.7458,
+      "step": 4157
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0271225322442256e-05,
+      "loss": 0.7853,
+      "step": 4158
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0233631450651523e-05,
+      "loss": 0.812,
+      "step": 4159
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.019606854691145e-05,
+      "loss": 0.7871,
+      "step": 4160
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0158536625805325e-05,
+      "loss": 0.8162,
+      "step": 4161
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.01210357019043e-05,
+      "loss": 0.7527,
+      "step": 4162
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0083565789767523e-05,
+      "loss": 0.8018,
+      "step": 4163
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.004612690394212e-05,
+      "loss": 0.7954,
+      "step": 4164
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0008719058963145e-05,
+      "loss": 0.7911,
+      "step": 4165
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.997134226935361e-05,
+      "loss": 0.7913,
+      "step": 4166
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.993399654962447e-05,
+      "loss": 0.8374,
+      "step": 4167
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9896681914274617e-05,
+      "loss": 0.8521,
+      "step": 4168
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9859398377790873e-05,
+      "loss": 0.8496,
+      "step": 4169
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.982214595464804e-05,
+      "loss": 0.8191,
+      "step": 4170
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.97849246593087e-05,
+      "loss": 0.8301,
+      "step": 4171
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9747734506223525e-05,
+      "loss": 0.821,
+      "step": 4172
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9710575509831007e-05,
+      "loss": 0.8274,
+      "step": 4173
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9673447684557468e-05,
+      "loss": 0.7419,
+      "step": 4174
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9636351044817292e-05,
+      "loss": 0.7482,
+      "step": 4175
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9599285605012642e-05,
+      "loss": 0.8086,
+      "step": 4176
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.956225137953359e-05,
+      "loss": 0.3358,
+      "step": 4177
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.952524838275811e-05,
+      "loss": 0.8403,
+      "step": 4178
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9488276629052026e-05,
+      "loss": 0.8157,
+      "step": 4179
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.945133613276907e-05,
+      "loss": 0.7957,
+      "step": 4180
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.941442690825076e-05,
+      "loss": 0.7974,
+      "step": 4181
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9377548969826633e-05,
+      "loss": 0.7993,
+      "step": 4182
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9340702331813842e-05,
+      "loss": 0.7197,
+      "step": 4183
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.930388700851762e-05,
+      "loss": 0.8503,
+      "step": 4184
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9267103014230935e-05,
+      "loss": 0.7886,
+      "step": 4185
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9230350363234516e-05,
+      "loss": 0.7103,
+      "step": 4186
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9193629069797102e-05,
+      "loss": 0.7435,
+      "step": 4187
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9156939148175125e-05,
+      "loss": 0.7529,
+      "step": 4188
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.912028061261287e-05,
+      "loss": 0.772,
+      "step": 4189
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9083653477342466e-05,
+      "loss": 0.74,
+      "step": 4190
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.904705775658381e-05,
+      "loss": 0.6899,
+      "step": 4191
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.901049346454462e-05,
+      "loss": 0.8416,
+      "step": 4192
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8973960615420416e-05,
+      "loss": 0.72,
+      "step": 4193
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8937459223394515e-05,
+      "loss": 0.3183,
+      "step": 4194
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8900989302637985e-05,
+      "loss": 0.7041,
+      "step": 4195
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8864550867309773e-05,
+      "loss": 0.7466,
+      "step": 4196
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8828143931556442e-05,
+      "loss": 0.7578,
+      "step": 4197
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8791768509512487e-05,
+      "loss": 0.8394,
+      "step": 4198
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.875542461530011e-05,
+      "loss": 0.7996,
+      "step": 4199
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.871911226302917e-05,
+      "loss": 0.7947,
+      "step": 4200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8682831466797467e-05,
+      "loss": 0.8284,
+      "step": 4201
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8646582240690412e-05,
+      "loss": 0.7688,
+      "step": 4202
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8610364598781226e-05,
+      "loss": 0.8196,
+      "step": 4203
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8574178555130817e-05,
+      "loss": 0.7294,
+      "step": 4204
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8538024123787868e-05,
+      "loss": 0.7742,
+      "step": 4205
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8501901318788774e-05,
+      "loss": 0.7531,
+      "step": 4206
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8465810154157624e-05,
+      "loss": 0.7194,
+      "step": 4207
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.842975064390633e-05,
+      "loss": 0.7891,
+      "step": 4208
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.839372280203433e-05,
+      "loss": 0.7379,
+      "step": 4209
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.835772664252895e-05,
+      "loss": 0.7751,
+      "step": 4210
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.832176217936511e-05,
+      "loss": 0.7888,
+      "step": 4211
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8285829426505453e-05,
+      "loss": 0.7396,
+      "step": 4212
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.824992839790035e-05,
+      "loss": 0.7869,
+      "step": 4213
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8214059107487726e-05,
+      "loss": 0.7737,
+      "step": 4214
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8178221569193343e-05,
+      "loss": 0.8008,
+      "step": 4215
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8142415796930568e-05,
+      "loss": 0.8137,
+      "step": 4216
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.810664180460041e-05,
+      "loss": 0.8152,
+      "step": 4217
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8070899606091583e-05,
+      "loss": 0.7974,
+      "step": 4218
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8035189215280423e-05,
+      "loss": 0.7849,
+      "step": 4219
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7999510646030948e-05,
+      "loss": 0.7877,
+      "step": 4220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7963863912194766e-05,
+      "loss": 0.7612,
+      "step": 4221
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7928249027611254e-05,
+      "loss": 0.7539,
+      "step": 4222
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.789266600610724e-05,
+      "loss": 0.7135,
+      "step": 4223
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7857114861497336e-05,
+      "loss": 0.7856,
+      "step": 4224
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.782159560758373e-05,
+      "loss": 0.799,
+      "step": 4225
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7786108258156155e-05,
+      "loss": 0.7833,
+      "step": 4226
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7750652826992075e-05,
+      "loss": 0.7515,
+      "step": 4227
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7715229327856498e-05,
+      "loss": 0.802,
+      "step": 4228
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.767983777450205e-05,
+      "loss": 0.7336,
+      "step": 4229
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7644478180668943e-05,
+      "loss": 0.793,
+      "step": 4230
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7609150560084985e-05,
+      "loss": 0.7979,
+      "step": 4231
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.757385492646558e-05,
+      "loss": 0.7433,
+      "step": 4232
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7538591293513685e-05,
+      "loss": 0.724,
+      "step": 4233
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.750335967491993e-05,
+      "loss": 0.8102,
+      "step": 4234
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.746816008436234e-05,
+      "loss": 0.7869,
+      "step": 4235
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7432992535506686e-05,
+      "loss": 0.75,
+      "step": 4236
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7397857042006194e-05,
+      "loss": 0.7758,
+      "step": 4237
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.736275361750167e-05,
+      "loss": 0.8079,
+      "step": 4238
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7327682275621503e-05,
+      "loss": 0.8328,
+      "step": 4239
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7292643029981526e-05,
+      "loss": 0.8173,
+      "step": 4240
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.725763589418523e-05,
+      "loss": 0.8027,
+      "step": 4241
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7222660881823593e-05,
+      "loss": 0.7592,
+      "step": 4242
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7187718006475117e-05,
+      "loss": 0.7898,
+      "step": 4243
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7152807281705806e-05,
+      "loss": 0.7428,
+      "step": 4244
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7117928721069232e-05,
+      "loss": 0.7776,
+      "step": 4245
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7083082338106438e-05,
+      "loss": 0.8125,
+      "step": 4246
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7048268146345968e-05,
+      "loss": 0.731,
+      "step": 4247
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.701348615930397e-05,
+      "loss": 0.7812,
+      "step": 4248
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6978736390483895e-05,
+      "loss": 0.7781,
+      "step": 4249
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6944018853376897e-05,
+      "loss": 0.6844,
+      "step": 4250
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6909333561461472e-05,
+      "loss": 0.7266,
+      "step": 4251
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6874680528203656e-05,
+      "loss": 0.8328,
+      "step": 4252
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.684005976705695e-05,
+      "loss": 0.7825,
+      "step": 4253
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6805471291462317e-05,
+      "loss": 0.7859,
+      "step": 4254
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6770915114848194e-05,
+      "loss": 0.8337,
+      "step": 4255
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.67363912506305e-05,
+      "loss": 0.7578,
+      "step": 4256
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6701899712212564e-05,
+      "loss": 0.8123,
+      "step": 4257
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.66674405129852e-05,
+      "loss": 0.8245,
+      "step": 4258
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6633013666326637e-05,
+      "loss": 0.8054,
+      "step": 4259
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6598619185602614e-05,
+      "loss": 0.7242,
+      "step": 4260
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.656425708416617e-05,
+      "loss": 0.3147,
+      "step": 4261
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6529927375357958e-05,
+      "loss": 0.7808,
+      "step": 4262
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.649563007250584e-05,
+      "loss": 0.7841,
+      "step": 4263
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6461365188925304e-05,
+      "loss": 0.7759,
+      "step": 4264
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6427132737919137e-05,
+      "loss": 0.7959,
+      "step": 4265
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6392932732777487e-05,
+      "loss": 0.7979,
+      "step": 4266
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6358765186778057e-05,
+      "loss": 0.8049,
+      "step": 4267
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6324630113185836e-05,
+      "loss": 0.7747,
+      "step": 4268
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6290527525253228e-05,
+      "loss": 0.7754,
+      "step": 4269
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6256457436220028e-05,
+      "loss": 0.8408,
+      "step": 4270
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6222419859313442e-05,
+      "loss": 0.8108,
+      "step": 4271
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6188414807747997e-05,
+      "loss": 0.8232,
+      "step": 4272
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6154442294725636e-05,
+      "loss": 0.7842,
+      "step": 4273
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6120502333435693e-05,
+      "loss": 0.2976,
+      "step": 4274
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6086594937054765e-05,
+      "loss": 0.75,
+      "step": 4275
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6052720118746922e-05,
+      "loss": 0.8025,
+      "step": 4276
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.601887789166352e-05,
+      "loss": 0.7607,
+      "step": 4277
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.598506826894328e-05,
+      "loss": 0.727,
+      "step": 4278
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5951291263712252e-05,
+      "loss": 0.7888,
+      "step": 4279
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5917546889083834e-05,
+      "loss": 0.7939,
+      "step": 4280
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5883835158158765e-05,
+      "loss": 0.7588,
+      "step": 4281
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.585015608402509e-05,
+      "loss": 0.8127,
+      "step": 4282
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5816509679758185e-05,
+      "loss": 0.8391,
+      "step": 4283
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.578289595842074e-05,
+      "loss": 0.7561,
+      "step": 4284
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5749314933062753e-05,
+      "loss": 0.8359,
+      "step": 4285
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5715766616721582e-05,
+      "loss": 0.7595,
+      "step": 4286
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5682251022421758e-05,
+      "loss": 0.7908,
+      "step": 4287
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5648768163175277e-05,
+      "loss": 0.7489,
+      "step": 4288
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.561531805198124e-05,
+      "loss": 0.32,
+      "step": 4289
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5581900701826225e-05,
+      "loss": 0.7666,
+      "step": 4290
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5548516125683978e-05,
+      "loss": 0.7837,
+      "step": 4291
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5515164336515465e-05,
+      "loss": 0.8071,
+      "step": 4292
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5481845347269076e-05,
+      "loss": 0.7856,
+      "step": 4293
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5448559170880374e-05,
+      "loss": 0.8024,
+      "step": 4294
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5415305820272197e-05,
+      "loss": 0.8091,
+      "step": 4295
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.538208530835463e-05,
+      "loss": 0.7733,
+      "step": 4296
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.534889764802503e-05,
+      "loss": 0.8044,
+      "step": 4297
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.531574285216799e-05,
+      "loss": 0.7625,
+      "step": 4298
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.528262093365531e-05,
+      "loss": 0.8259,
+      "step": 4299
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5249531905346138e-05,
+      "loss": 0.8308,
+      "step": 4300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5216475780086669e-05,
+      "loss": 0.8063,
+      "step": 4301
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5183452570710522e-05,
+      "loss": 0.7267,
+      "step": 4302
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5150462290038392e-05,
+      "loss": 0.3555,
+      "step": 4303
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.511750495087827e-05,
+      "loss": 0.8003,
+      "step": 4304
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5084580566025308e-05,
+      "loss": 0.8682,
+      "step": 4305
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5051689148261894e-05,
+      "loss": 0.8215,
+      "step": 4306
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5018830710357611e-05,
+      "loss": 0.6802,
+      "step": 4307
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4986005265069204e-05,
+      "loss": 0.8081,
+      "step": 4308
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4953212825140727e-05,
+      "loss": 0.7892,
+      "step": 4309
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4920453403303247e-05,
+      "loss": 0.7202,
+      "step": 4310
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.488772701227511e-05,
+      "loss": 0.3535,
+      "step": 4311
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4855033664761897e-05,
+      "loss": 0.6409,
+      "step": 4312
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.48223733734562e-05,
+      "loss": 0.7869,
+      "step": 4313
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4789746151037942e-05,
+      "loss": 0.7622,
+      "step": 4314
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4757152010174069e-05,
+      "loss": 0.335,
+      "step": 4315
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4724590963518803e-05,
+      "loss": 0.7651,
+      "step": 4316
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4692063023713443e-05,
+      "loss": 0.7526,
+      "step": 4317
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4659568203386465e-05,
+      "loss": 0.7637,
+      "step": 4318
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4627106515153455e-05,
+      "loss": 0.781,
+      "step": 4319
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4594677971617177e-05,
+      "loss": 0.8107,
+      "step": 4320
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4562282585367492e-05,
+      "loss": 0.7881,
+      "step": 4321
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.452992036898142e-05,
+      "loss": 0.7522,
+      "step": 4322
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4497591335023086e-05,
+      "loss": 0.7407,
+      "step": 4323
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4465295496043729e-05,
+      "loss": 0.7854,
+      "step": 4324
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4433032864581686e-05,
+      "loss": 0.8269,
+      "step": 4325
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4400803453162482e-05,
+      "loss": 0.7318,
+      "step": 4326
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4368607274298595e-05,
+      "loss": 0.8132,
+      "step": 4327
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4336444340489775e-05,
+      "loss": 0.3181,
+      "step": 4328
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4304314664222728e-05,
+      "loss": 0.7727,
+      "step": 4329
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4272218257971326e-05,
+      "loss": 0.8044,
+      "step": 4330
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4240155134196499e-05,
+      "loss": 0.7952,
+      "step": 4331
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4208125305346232e-05,
+      "loss": 0.7385,
+      "step": 4332
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4176128783855636e-05,
+      "loss": 0.7737,
+      "step": 4333
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4144165582146817e-05,
+      "loss": 0.7832,
+      "step": 4334
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4112235712629064e-05,
+      "loss": 0.761,
+      "step": 4335
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.40803391876986e-05,
+      "loss": 0.6969,
+      "step": 4336
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4048476019738755e-05,
+      "loss": 0.7847,
+      "step": 4337
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.401664622111991e-05,
+      "loss": 0.7537,
+      "step": 4338
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3984849804199485e-05,
+      "loss": 0.7639,
+      "step": 4339
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3953086781321988e-05,
+      "loss": 0.8201,
+      "step": 4340
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3921357164818848e-05,
+      "loss": 0.799,
+      "step": 4341
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3889660967008655e-05,
+      "loss": 0.7844,
+      "step": 4342
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3857998200196941e-05,
+      "loss": 0.7808,
+      "step": 4343
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3826368876676278e-05,
+      "loss": 0.7739,
+      "step": 4344
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3794773008726259e-05,
+      "loss": 0.7869,
+      "step": 4345
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3763210608613498e-05,
+      "loss": 0.7761,
+      "step": 4346
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3731681688591591e-05,
+      "loss": 0.7446,
+      "step": 4347
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.370018626090116e-05,
+      "loss": 0.75,
+      "step": 4348
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3668724337769823e-05,
+      "loss": 0.7341,
+      "step": 4349
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3637295931412152e-05,
+      "loss": 0.7869,
+      "step": 4350
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3605901054029746e-05,
+      "loss": 0.7729,
+      "step": 4351
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.357453971781123e-05,
+      "loss": 0.7578,
+      "step": 4352
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3543211934932065e-05,
+      "loss": 0.8312,
+      "step": 4353
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3511917717554846e-05,
+      "loss": 0.7786,
+      "step": 4354
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.348065707782904e-05,
+      "loss": 0.7054,
+      "step": 4355
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3449430027891097e-05,
+      "loss": 0.7255,
+      "step": 4356
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.341823657986445e-05,
+      "loss": 0.7964,
+      "step": 4357
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.338707674585945e-05,
+      "loss": 0.8162,
+      "step": 4358
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.335595053797344e-05,
+      "loss": 0.8071,
+      "step": 4359
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.332485796829065e-05,
+      "loss": 0.7834,
+      "step": 4360
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.329379904888235e-05,
+      "loss": 0.8198,
+      "step": 4361
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3262773791806615e-05,
+      "loss": 0.7202,
+      "step": 4362
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3231782209108545e-05,
+      "loss": 0.8145,
+      "step": 4363
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3200824312820136e-05,
+      "loss": 0.7357,
+      "step": 4364
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3169900114960298e-05,
+      "loss": 0.7106,
+      "step": 4365
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3139009627534927e-05,
+      "loss": 0.7977,
+      "step": 4366
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3108152862536682e-05,
+      "loss": 0.8545,
+      "step": 4367
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3077329831945295e-05,
+      "loss": 0.7905,
+      "step": 4368
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3046540547727303e-05,
+      "loss": 0.7461,
+      "step": 4369
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3015785021836158e-05,
+      "loss": 0.7749,
+      "step": 4370
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2985063266212227e-05,
+      "loss": 0.7692,
+      "step": 4371
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.295437529278275e-05,
+      "loss": 0.7552,
+      "step": 4372
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2923721113461851e-05,
+      "loss": 0.8066,
+      "step": 4373
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.289310074015052e-05,
+      "loss": 0.7061,
+      "step": 4374
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2862514184736696e-05,
+      "loss": 0.7766,
+      "step": 4375
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2831961459095088e-05,
+      "loss": 0.7842,
+      "step": 4376
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2801442575087298e-05,
+      "loss": 0.7837,
+      "step": 4377
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2770957544561869e-05,
+      "loss": 0.7817,
+      "step": 4378
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2740506379354077e-05,
+      "loss": 0.7793,
+      "step": 4379
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2710089091286149e-05,
+      "loss": 0.7581,
+      "step": 4380
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.267970569216712e-05,
+      "loss": 0.7651,
+      "step": 4381
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2649356193792872e-05,
+      "loss": 0.8281,
+      "step": 4382
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2619040607946119e-05,
+      "loss": 0.7913,
+      "step": 4383
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2588758946396418e-05,
+      "loss": 0.3526,
+      "step": 4384
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2558511220900138e-05,
+      "loss": 0.7686,
+      "step": 4385
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2528297443200488e-05,
+      "loss": 0.7959,
+      "step": 4386
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2498117625027561e-05,
+      "loss": 0.7908,
+      "step": 4387
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.246797177809812e-05,
+      "loss": 0.7562,
+      "step": 4388
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2437859914115845e-05,
+      "loss": 0.7552,
+      "step": 4389
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2407782044771221e-05,
+      "loss": 0.7446,
+      "step": 4390
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.237773818174146e-05,
+      "loss": 0.7405,
+      "step": 4391
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2347728336690701e-05,
+      "loss": 0.8118,
+      "step": 4392
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2317752521269722e-05,
+      "loss": 0.7507,
+      "step": 4393
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2287810747116224e-05,
+      "loss": 0.7151,
+      "step": 4394
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2257903025854612e-05,
+      "loss": 0.7986,
+      "step": 4395
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2228029369096094e-05,
+      "loss": 0.3524,
+      "step": 4396
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2198189788438651e-05,
+      "loss": 0.7202,
+      "step": 4397
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2168384295467038e-05,
+      "loss": 0.7498,
+      "step": 4398
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2138612901752777e-05,
+      "loss": 0.7668,
+      "step": 4399
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2108875618854121e-05,
+      "loss": 0.7045,
+      "step": 4400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2079172458316168e-05,
+      "loss": 0.8176,
+      "step": 4401
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.204950343167065e-05,
+      "loss": 0.8164,
+      "step": 4402
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2019868550436098e-05,
+      "loss": 0.3388,
+      "step": 4403
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1990267826117873e-05,
+      "loss": 0.7415,
+      "step": 4404
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1960701270207886e-05,
+      "loss": 0.7749,
+      "step": 4405
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1931168894184974e-05,
+      "loss": 0.7306,
+      "step": 4406
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.19016707095146e-05,
+      "loss": 0.7487,
+      "step": 4407
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1872206727648972e-05,
+      "loss": 0.7981,
+      "step": 4408
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1842776960027013e-05,
+      "loss": 0.3087,
+      "step": 4409
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1813381418074388e-05,
+      "loss": 0.3427,
+      "step": 4410
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1784020113203454e-05,
+      "loss": 0.764,
+      "step": 4411
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.175469305681327e-05,
+      "loss": 0.8457,
+      "step": 4412
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1725400260289621e-05,
+      "loss": 0.7843,
+      "step": 4413
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1696141735004939e-05,
+      "loss": 0.7506,
+      "step": 4414
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1666917492318485e-05,
+      "loss": 0.761,
+      "step": 4415
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1637727543576027e-05,
+      "loss": 0.7091,
+      "step": 4416
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.160857190011012e-05,
+      "loss": 0.8002,
+      "step": 4417
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1579450573240058e-05,
+      "loss": 0.7839,
+      "step": 4418
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1550363574271639e-05,
+      "loss": 0.7869,
+      "step": 4419
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1521310914497519e-05,
+      "loss": 0.7493,
+      "step": 4420
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1492292605196907e-05,
+      "loss": 0.7908,
+      "step": 4421
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1463308657635719e-05,
+      "loss": 0.8169,
+      "step": 4422
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1434359083066515e-05,
+      "loss": 0.7795,
+      "step": 4423
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1405443892728528e-05,
+      "loss": 0.7944,
+      "step": 4424
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1376563097847614e-05,
+      "loss": 0.7615,
+      "step": 4425
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.134771670963628e-05,
+      "loss": 0.7622,
+      "step": 4426
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1318904739293745e-05,
+      "loss": 0.771,
+      "step": 4427
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1290127198005752e-05,
+      "loss": 0.8018,
+      "step": 4428
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1261384096944727e-05,
+      "loss": 0.7234,
+      "step": 4429
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1232675447269803e-05,
+      "loss": 0.7551,
+      "step": 4430
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1204001260126574e-05,
+      "loss": 0.8152,
+      "step": 4431
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1175361546647412e-05,
+      "loss": 0.776,
+      "step": 4432
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1146756317951224e-05,
+      "loss": 0.7993,
+      "step": 4433
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1118185585143537e-05,
+      "loss": 0.8242,
+      "step": 4434
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1089649359316501e-05,
+      "loss": 0.759,
+      "step": 4435
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1061147651548853e-05,
+      "loss": 0.8242,
+      "step": 4436
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1032680472905932e-05,
+      "loss": 0.8384,
+      "step": 4437
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1004247834439696e-05,
+      "loss": 0.7522,
+      "step": 4438
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.097584974718866e-05,
+      "loss": 0.3083,
+      "step": 4439
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0947486222177928e-05,
+      "loss": 0.7615,
+      "step": 4440
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0919157270419256e-05,
+      "loss": 0.3311,
+      "step": 4441
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0890862902910847e-05,
+      "loss": 0.7368,
+      "step": 4442
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0862603130637561e-05,
+      "loss": 0.7578,
+      "step": 4443
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0834377964570864e-05,
+      "loss": 0.8049,
+      "step": 4444
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0806187415668667e-05,
+      "loss": 0.7689,
+      "step": 4445
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0778031494875573e-05,
+      "loss": 0.8291,
+      "step": 4446
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0749910213122649e-05,
+      "loss": 0.7532,
+      "step": 4447
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.072182358132755e-05,
+      "loss": 0.8252,
+      "step": 4448
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0693771610394476e-05,
+      "loss": 0.7283,
+      "step": 4449
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0665754311214172e-05,
+      "loss": 0.7886,
+      "step": 4450
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0637771694663901e-05,
+      "loss": 0.7578,
+      "step": 4451
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0609823771607485e-05,
+      "loss": 0.803,
+      "step": 4452
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0581910552895302e-05,
+      "loss": 0.8477,
+      "step": 4453
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.055403204936416e-05,
+      "loss": 0.7466,
+      "step": 4454
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0526188271837511e-05,
+      "loss": 0.8123,
+      "step": 4455
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0498379231125278e-05,
+      "loss": 0.7988,
+      "step": 4456
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0470604938023808e-05,
+      "loss": 0.7147,
+      "step": 4457
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0442865403316116e-05,
+      "loss": 0.7856,
+      "step": 4458
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0415160637771603e-05,
+      "loss": 0.7488,
+      "step": 4459
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0387490652146237e-05,
+      "loss": 0.762,
+      "step": 4460
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0359855457182454e-05,
+      "loss": 0.7842,
+      "step": 4461
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0332255063609175e-05,
+      "loss": 0.7979,
+      "step": 4462
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0304689482141838e-05,
+      "loss": 0.79,
+      "step": 4463
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0277158723482339e-05,
+      "loss": 0.7632,
+      "step": 4464
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.024966279831907e-05,
+      "loss": 0.8223,
+      "step": 4465
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0222201717326885e-05,
+      "loss": 0.3219,
+      "step": 4466
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0194775491167163e-05,
+      "loss": 0.7981,
+      "step": 4467
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0167384130487667e-05,
+      "loss": 0.8192,
+      "step": 4468
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0140027645922657e-05,
+      "loss": 0.8669,
+      "step": 4469
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0112706048092923e-05,
+      "loss": 0.7808,
+      "step": 4470
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0085419347605574e-05,
+      "loss": 0.8054,
+      "step": 4471
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0058167555054298e-05,
+      "loss": 0.7107,
+      "step": 4472
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.003095068101917e-05,
+      "loss": 0.7448,
+      "step": 4473
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0003768736066722e-05,
+      "loss": 0.7783,
+      "step": 4474
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.97662173074989e-06,
+      "loss": 0.7949,
+      "step": 4475
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.949509675608115e-06,
+      "loss": 0.8013,
+      "step": 4476
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.922432581167207e-06,
+      "loss": 0.7205,
+      "step": 4477
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.895390457939413e-06,
+      "loss": 0.7418,
+      "step": 4478
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.86838331642348e-06,
+      "loss": 0.8577,
+      "step": 4479
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.84141116710442e-06,
+      "loss": 0.7704,
+      "step": 4480
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.814474020453822e-06,
+      "loss": 0.7988,
+      "step": 4481
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.787571886929603e-06,
+      "loss": 0.8398,
+      "step": 4482
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.760704776976049e-06,
+      "loss": 0.765,
+      "step": 4483
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.733872701023939e-06,
+      "loss": 0.7759,
+      "step": 4484
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.707075669490407e-06,
+      "loss": 0.8267,
+      "step": 4485
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.680313692778974e-06,
+      "loss": 0.8201,
+      "step": 4486
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.653586781279567e-06,
+      "loss": 0.7376,
+      "step": 4487
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.626894945368491e-06,
+      "loss": 0.7394,
+      "step": 4488
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.600238195408428e-06,
+      "loss": 0.821,
+      "step": 4489
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.573616541748464e-06,
+      "loss": 0.7483,
+      "step": 4490
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.547029994724022e-06,
+      "loss": 0.7825,
+      "step": 4491
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.520478564656899e-06,
+      "loss": 0.7587,
+      "step": 4492
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.493962261855349e-06,
+      "loss": 0.8071,
+      "step": 4493
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.467481096613828e-06,
+      "loss": 0.3455,
+      "step": 4494
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.441035079213267e-06,
+      "loss": 0.7484,
+      "step": 4495
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.414624219920953e-06,
+      "loss": 0.8197,
+      "step": 4496
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.388248528990428e-06,
+      "loss": 0.802,
+      "step": 4497
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.361908016661702e-06,
+      "loss": 0.7419,
+      "step": 4498
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.335602693161038e-06,
+      "loss": 0.8132,
+      "step": 4499
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.309332568701079e-06,
+      "loss": 0.769,
+      "step": 4500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.283097653480789e-06,
+      "loss": 0.7462,
+      "step": 4501
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.256897957685462e-06,
+      "loss": 0.7539,
+      "step": 4502
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.23073349148672e-06,
+      "loss": 0.7788,
+      "step": 4503
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.204604265042505e-06,
+      "loss": 0.7947,
+      "step": 4504
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.178510288497122e-06,
+      "loss": 0.7648,
+      "step": 4505
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.15245157198108e-06,
+      "loss": 0.8047,
+      "step": 4506
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.126428125611342e-06,
+      "loss": 0.7848,
+      "step": 4507
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.10043995949108e-06,
+      "loss": 0.3264,
+      "step": 4508
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.07448708370976e-06,
+      "loss": 0.7969,
+      "step": 4509
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.04856950834323e-06,
+      "loss": 0.7373,
+      "step": 4510
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.022687243453554e-06,
+      "loss": 0.7377,
+      "step": 4511
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.996840299089149e-06,
+      "loss": 0.7805,
+      "step": 4512
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.971028685284655e-06,
+      "loss": 0.8047,
+      "step": 4513
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.945252412061055e-06,
+      "loss": 0.7386,
+      "step": 4514
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.91951148942557e-06,
+      "loss": 0.8218,
+      "step": 4515
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.893805927371723e-06,
+      "loss": 0.7781,
+      "step": 4516
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.86813573587929e-06,
+      "loss": 0.7234,
+      "step": 4517
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.842500924914298e-06,
+      "loss": 0.7461,
+      "step": 4518
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.816901504429143e-06,
+      "loss": 0.7644,
+      "step": 4519
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.791337484362306e-06,
+      "loss": 0.7496,
+      "step": 4520
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.765808874638681e-06,
+      "loss": 0.708,
+      "step": 4521
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.740315685169365e-06,
+      "loss": 0.7319,
+      "step": 4522
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.714857925851615e-06,
+      "loss": 0.7576,
+      "step": 4523
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.689435606569085e-06,
+      "loss": 0.7776,
+      "step": 4524
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.664048737191566e-06,
+      "loss": 0.793,
+      "step": 4525
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.638697327575108e-06,
+      "loss": 0.7642,
+      "step": 4526
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.613381387562015e-06,
+      "loss": 0.7407,
+      "step": 4527
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.588100926980802e-06,
+      "loss": 0.7993,
+      "step": 4528
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.56285595564621e-06,
+      "loss": 0.745,
+      "step": 4529
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.537646483359186e-06,
+      "loss": 0.7762,
+      "step": 4530
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.512472519906978e-06,
+      "loss": 0.7452,
+      "step": 4531
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.487334075062913e-06,
+      "loss": 0.7571,
+      "step": 4532
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.462231158586653e-06,
+      "loss": 0.8105,
+      "step": 4533
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.437163780224011e-06,
+      "loss": 0.7367,
+      "step": 4534
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.41213194970696e-06,
+      "loss": 0.8618,
+      "step": 4535
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.387135676753755e-06,
+      "loss": 0.7753,
+      "step": 4536
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.362174971068803e-06,
+      "loss": 0.7225,
+      "step": 4537
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.33724984234272e-06,
+      "loss": 0.7915,
+      "step": 4538
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.312360300252286e-06,
+      "loss": 0.7771,
+      "step": 4539
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.287506354460484e-06,
+      "loss": 0.7949,
+      "step": 4540
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.26268801461646e-06,
+      "loss": 0.7825,
+      "step": 4541
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.237905290355564e-06,
+      "loss": 0.7937,
+      "step": 4542
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.213158191299297e-06,
+      "loss": 0.7776,
+      "step": 4543
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.18844672705531e-06,
+      "loss": 0.8062,
+      "step": 4544
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.163770907217505e-06,
+      "loss": 0.7307,
+      "step": 4545
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.13913074136582e-06,
+      "loss": 0.7646,
+      "step": 4546
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.114526239066455e-06,
+      "loss": 0.7107,
+      "step": 4547
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.08995740987173e-06,
+      "loss": 0.7983,
+      "step": 4548
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.065424263320054e-06,
+      "loss": 0.7571,
+      "step": 4549
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.040926808936111e-06,
+      "loss": 0.8142,
+      "step": 4550
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.016465056230615e-06,
+      "loss": 0.7915,
+      "step": 4551
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.99203901470047e-06,
+      "loss": 0.823,
+      "step": 4552
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.967648693828712e-06,
+      "loss": 0.731,
+      "step": 4553
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.943294103084487e-06,
+      "loss": 0.7266,
+      "step": 4554
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.918975251923099e-06,
+      "loss": 0.8115,
+      "step": 4555
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.894692149785954e-06,
+      "loss": 0.7729,
+      "step": 4556
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.87044480610062e-06,
+      "loss": 0.772,
+      "step": 4557
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.846233230280697e-06,
+      "loss": 0.814,
+      "step": 4558
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.822057431725993e-06,
+      "loss": 0.7683,
+      "step": 4559
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.797917419822375e-06,
+      "loss": 0.7664,
+      "step": 4560
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.773813203941826e-06,
+      "loss": 0.7852,
+      "step": 4561
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.749744793442448e-06,
+      "loss": 0.802,
+      "step": 4562
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.725712197668378e-06,
+      "loss": 0.709,
+      "step": 4563
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.70171542594995e-06,
+      "loss": 0.7339,
+      "step": 4564
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.677754487603517e-06,
+      "loss": 0.811,
+      "step": 4565
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.653829391931534e-06,
+      "loss": 0.8171,
+      "step": 4566
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.629940148222559e-06,
+      "loss": 0.7976,
+      "step": 4567
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.606086765751208e-06,
+      "loss": 0.7964,
+      "step": 4568
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.582269253778185e-06,
+      "loss": 0.3368,
+      "step": 4569
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.5584876215502695e-06,
+      "loss": 0.7795,
+      "step": 4570
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.534741878300333e-06,
+      "loss": 0.7898,
+      "step": 4571
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.511032033247256e-06,
+      "loss": 0.8081,
+      "step": 4572
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.487358095596031e-06,
+      "loss": 0.8264,
+      "step": 4573
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.463720074537728e-06,
+      "loss": 0.7793,
+      "step": 4574
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.440117979249362e-06,
+      "loss": 0.7545,
+      "step": 4575
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.416551818894157e-06,
+      "loss": 0.7615,
+      "step": 4576
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.393021602621264e-06,
+      "loss": 0.7864,
+      "step": 4577
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.3695273395659515e-06,
+      "loss": 0.7694,
+      "step": 4578
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.346069038849468e-06,
+      "loss": 0.8049,
+      "step": 4579
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.322646709579173e-06,
+      "loss": 0.7684,
+      "step": 4580
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.299260360848381e-06,
+      "loss": 0.7485,
+      "step": 4581
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.275910001736497e-06,
+      "loss": 0.7576,
+      "step": 4582
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.2525956413089565e-06,
+      "loss": 0.8342,
+      "step": 4583
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.2293172886171435e-06,
+      "loss": 0.7864,
+      "step": 4584
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.206074952698561e-06,
+      "loss": 0.73,
+      "step": 4585
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.1828686425766785e-06,
+      "loss": 0.7749,
+      "step": 4586
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.15969836726097e-06,
+      "loss": 0.8145,
+      "step": 4587
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.136564135746959e-06,
+      "loss": 0.8464,
+      "step": 4588
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.113465957016097e-06,
+      "loss": 0.8286,
+      "step": 4589
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.090403840035942e-06,
+      "loss": 0.7991,
+      "step": 4590
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.067377793759999e-06,
+      "loss": 0.8081,
+      "step": 4591
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.044387827127752e-06,
+      "loss": 0.7856,
+      "step": 4592
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.0214339490647045e-06,
+      "loss": 0.7349,
+      "step": 4593
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.99851616848235e-06,
+      "loss": 0.7383,
+      "step": 4594
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.975634494278149e-06,
+      "loss": 0.7727,
+      "step": 4595
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.952788935335541e-06,
+      "loss": 0.7837,
+      "step": 4596
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.92997950052402e-06,
+      "loss": 0.8169,
+      "step": 4597
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.907206198698913e-06,
+      "loss": 0.7615,
+      "step": 4598
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.884469038701646e-06,
+      "loss": 0.7111,
+      "step": 4599
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.861768029359594e-06,
+      "loss": 0.7769,
+      "step": 4600
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.839103179485995e-06,
+      "loss": 0.7751,
+      "step": 4601
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.8164744978801765e-06,
+      "loss": 0.7847,
+      "step": 4602
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.793881993327367e-06,
+      "loss": 0.3654,
+      "step": 4603
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.771325674598749e-06,
+      "loss": 0.7515,
+      "step": 4604
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.748805550451454e-06,
+      "loss": 0.7769,
+      "step": 4605
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.726321629628585e-06,
+      "loss": 0.7976,
+      "step": 4606
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.7038739208591605e-06,
+      "loss": 0.7784,
+      "step": 4607
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.681462432858154e-06,
+      "loss": 0.7815,
+      "step": 4608
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.659087174326506e-06,
+      "loss": 0.7739,
+      "step": 4609
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.636748153951e-06,
+      "loss": 0.6846,
+      "step": 4610
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.614445380404477e-06,
+      "loss": 0.7903,
+      "step": 4611
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.5921788623456216e-06,
+      "loss": 0.7742,
+      "step": 4612
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.569948608419041e-06,
+      "loss": 0.7812,
+      "step": 4613
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.547754627255331e-06,
+      "loss": 0.6932,
+      "step": 4614
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.5255969274708896e-06,
+      "loss": 0.7206,
+      "step": 4615
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.503475517668167e-06,
+      "loss": 0.7928,
+      "step": 4616
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.481390406435417e-06,
+      "loss": 0.7852,
+      "step": 4617
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.4593416023468575e-06,
+      "loss": 0.7405,
+      "step": 4618
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.437329113962576e-06,
+      "loss": 0.7266,
+      "step": 4619
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.4153529498286014e-06,
+      "loss": 0.7476,
+      "step": 4620
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.393413118476821e-06,
+      "loss": 0.7944,
+      "step": 4621
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.3715096284250205e-06,
+      "loss": 0.7537,
+      "step": 4622
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.349642488176944e-06,
+      "loss": 0.8088,
+      "step": 4623
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.327811706222098e-06,
+      "loss": 0.7468,
+      "step": 4624
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.3060172910359815e-06,
+      "loss": 0.7888,
+      "step": 4625
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.284259251079938e-06,
+      "loss": 0.77,
+      "step": 4626
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.262537594801177e-06,
+      "loss": 0.7981,
+      "step": 4627
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.240852330632796e-06,
+      "loss": 0.8088,
+      "step": 4628
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.219203466993761e-06,
+      "loss": 0.7566,
+      "step": 4629
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.197591012288917e-06,
+      "loss": 0.7559,
+      "step": 4630
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.17601497490895e-06,
+      "loss": 0.327,
+      "step": 4631
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.1544753632304165e-06,
+      "loss": 0.7954,
+      "step": 4632
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.132972185615749e-06,
+      "loss": 0.7605,
+      "step": 4633
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.111505450413202e-06,
+      "loss": 0.7878,
+      "step": 4634
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.090075165956943e-06,
+      "loss": 0.8472,
+      "step": 4635
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.068681340566895e-06,
+      "loss": 0.7593,
+      "step": 4636
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.0473239825489244e-06,
+      "loss": 0.3577,
+      "step": 4637
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.026003100194632e-06,
+      "loss": 0.7754,
+      "step": 4638
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.004718701781575e-06,
+      "loss": 0.7612,
+      "step": 4639
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.983470795573088e-06,
+      "loss": 0.7749,
+      "step": 4640
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.962259389818292e-06,
+      "loss": 0.7812,
+      "step": 4641
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.941084492752235e-06,
+      "loss": 0.7568,
+      "step": 4642
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.91994611259572e-06,
+      "loss": 0.7727,
+      "step": 4643
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.898844257555391e-06,
+      "loss": 0.748,
+      "step": 4644
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.877778935823719e-06,
+      "loss": 0.7429,
+      "step": 4645
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.856750155578983e-06,
+      "loss": 0.7354,
+      "step": 4646
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.835757924985286e-06,
+      "loss": 0.7437,
+      "step": 4647
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.8148022521925195e-06,
+      "loss": 0.7659,
+      "step": 4648
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.793883145336443e-06,
+      "loss": 0.7704,
+      "step": 4649
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.773000612538504e-06,
+      "loss": 0.7341,
+      "step": 4650
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.752154661906084e-06,
+      "loss": 0.7478,
+      "step": 4651
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.731345301532265e-06,
+      "loss": 0.7958,
+      "step": 4652
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.710572539495962e-06,
+      "loss": 0.7898,
+      "step": 4653
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.689836383861879e-06,
+      "loss": 0.8379,
+      "step": 4654
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.6691368426805114e-06,
+      "loss": 0.8046,
+      "step": 4655
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.648473923988129e-06,
+      "loss": 0.8228,
+      "step": 4656
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.627847635806771e-06,
+      "loss": 0.7977,
+      "step": 4657
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.607257986144321e-06,
+      "loss": 0.7683,
+      "step": 4658
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.5867049829943395e-06,
+      "loss": 0.8079,
+      "step": 4659
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.566188634336211e-06,
+      "loss": 0.7593,
+      "step": 4660
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.545708948135142e-06,
+      "loss": 0.7566,
+      "step": 4661
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.525265932341983e-06,
+      "loss": 0.3291,
+      "step": 4662
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.504859594893474e-06,
+      "loss": 0.8025,
+      "step": 4663
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.484489943712012e-06,
+      "loss": 0.7993,
+      "step": 4664
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.464156986705826e-06,
+      "loss": 0.7981,
+      "step": 4665
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.443860731768868e-06,
+      "loss": 0.7957,
+      "step": 4666
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.4236011867808355e-06,
+      "loss": 0.769,
+      "step": 4667
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.403378359607181e-06,
+      "loss": 0.8003,
+      "step": 4668
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.383192258099112e-06,
+      "loss": 0.8091,
+      "step": 4669
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.36304289009355e-06,
+      "loss": 0.7163,
+      "step": 4670
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.342930263413193e-06,
+      "loss": 0.7603,
+      "step": 4671
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.322854385866438e-06,
+      "loss": 0.7562,
+      "step": 4672
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.3028152652474514e-06,
+      "loss": 0.8313,
+      "step": 4673
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.282812909336077e-06,
+      "loss": 0.7097,
+      "step": 4674
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.262847325897968e-06,
+      "loss": 0.7324,
+      "step": 4675
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.242918522684393e-06,
+      "loss": 0.7766,
+      "step": 4676
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.223026507432449e-06,
+      "loss": 0.8008,
+      "step": 4677
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.203171287864872e-06,
+      "loss": 0.7158,
+      "step": 4678
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.183352871690162e-06,
+      "loss": 0.7524,
+      "step": 4679
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.163571266602485e-06,
+      "loss": 0.7294,
+      "step": 4680
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.143826480281777e-06,
+      "loss": 0.8212,
+      "step": 4681
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.124118520393606e-06,
+      "loss": 0.7896,
+      "step": 4682
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.104447394589296e-06,
+      "loss": 0.7261,
+      "step": 4683
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.0848131105058704e-06,
+      "loss": 0.7137,
+      "step": 4684
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.0652156757660215e-06,
+      "loss": 0.3505,
+      "step": 4685
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.045655097978131e-06,
+      "loss": 0.7944,
+      "step": 4686
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.026131384736321e-06,
+      "loss": 0.3222,
+      "step": 4687
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.006644543620342e-06,
+      "loss": 0.7357,
+      "step": 4688
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.987194582195687e-06,
+      "loss": 0.803,
+      "step": 4689
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.967781508013458e-06,
+      "loss": 0.7467,
+      "step": 4690
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9484053286105055e-06,
+      "loss": 0.3115,
+      "step": 4691
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.929066051509346e-06,
+      "loss": 0.7791,
+      "step": 4692
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.909763684218116e-06,
+      "loss": 0.7321,
+      "step": 4693
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.89049823423069e-06,
+      "loss": 0.7837,
+      "step": 4694
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.87126970902656e-06,
+      "loss": 0.7516,
+      "step": 4695
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.852078116070902e-06,
+      "loss": 0.764,
+      "step": 4696
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.832923462814565e-06,
+      "loss": 0.7484,
+      "step": 4697
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.813805756694034e-06,
+      "loss": 0.7488,
+      "step": 4698
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.794725005131462e-06,
+      "loss": 0.7523,
+      "step": 4699
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.775681215534656e-06,
+      "loss": 0.7932,
+      "step": 4700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.756674395297089e-06,
+      "loss": 0.7593,
+      "step": 4701
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.737704551797817e-06,
+      "loss": 0.7988,
+      "step": 4702
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.718771692401636e-06,
+      "loss": 0.3347,
+      "step": 4703
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.699875824458899e-06,
+      "loss": 0.8103,
+      "step": 4704
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6810169553056615e-06,
+      "loss": 0.8154,
+      "step": 4705
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.662195092263566e-06,
+      "loss": 0.7192,
+      "step": 4706
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.643410242639911e-06,
+      "loss": 0.3193,
+      "step": 4707
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.624662413727621e-06,
+      "loss": 0.7522,
+      "step": 4708
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.605951612805237e-06,
+      "loss": 0.7646,
+      "step": 4709
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.587277847136984e-06,
+      "loss": 0.751,
+      "step": 4710
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5686411239726055e-06,
+      "loss": 0.3192,
+      "step": 4711
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.550041450547548e-06,
+      "loss": 0.8076,
+      "step": 4712
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.531478834082836e-06,
+      "loss": 0.7906,
+      "step": 4713
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.512953281785104e-06,
+      "loss": 0.8567,
+      "step": 4714
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.494464800846654e-06,
+      "loss": 0.7441,
+      "step": 4715
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.476013398445289e-06,
+      "loss": 0.7891,
+      "step": 4716
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.457599081744523e-06,
+      "loss": 0.7875,
+      "step": 4717
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.439221857893416e-06,
+      "loss": 0.3007,
+      "step": 4718
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.420881734026639e-06,
+      "loss": 0.8203,
+      "step": 4719
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.402578717264449e-06,
+      "loss": 0.7461,
+      "step": 4720
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.384312814712721e-06,
+      "loss": 0.8293,
+      "step": 4721
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.366084033462914e-06,
+      "loss": 0.7151,
+      "step": 4722
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.347892380592034e-06,
+      "loss": 0.6963,
+      "step": 4723
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.329737863162752e-06,
+      "loss": 0.8589,
+      "step": 4724
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.311620488223256e-06,
+      "loss": 0.7609,
+      "step": 4725
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.293540262807317e-06,
+      "loss": 0.7478,
+      "step": 4726
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.27549719393433e-06,
+      "loss": 0.7583,
+      "step": 4727
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.257491288609217e-06,
+      "loss": 0.7488,
+      "step": 4728
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.239522553822495e-06,
+      "loss": 0.8088,
+      "step": 4729
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.221590996550251e-06,
+      "loss": 0.7671,
+      "step": 4730
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.203696623754138e-06,
+      "loss": 0.7493,
+      "step": 4731
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.185839442381357e-06,
+      "loss": 0.8271,
+      "step": 4732
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.168019459364669e-06,
+      "loss": 0.7935,
+      "step": 4733
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.150236681622433e-06,
+      "loss": 0.7269,
+      "step": 4734
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.132491116058501e-06,
+      "loss": 0.7533,
+      "step": 4735
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.114782769562364e-06,
+      "loss": 0.7957,
+      "step": 4736
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.097111649008967e-06,
+      "loss": 0.7844,
+      "step": 4737
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.079477761258854e-06,
+      "loss": 0.8202,
+      "step": 4738
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.061881113158117e-06,
+      "loss": 0.3165,
+      "step": 4739
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.044321711538368e-06,
+      "loss": 0.7317,
+      "step": 4740
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0267995632168094e-06,
+      "loss": 0.8085,
+      "step": 4741
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.00931467499609e-06,
+      "loss": 0.7573,
+      "step": 4742
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9918670536644775e-06,
+      "loss": 0.7765,
+      "step": 4743
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.974456705995733e-06,
+      "loss": 0.6978,
+      "step": 4744
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.957083638749148e-06,
+      "loss": 0.7759,
+      "step": 4745
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.939747858669551e-06,
+      "loss": 0.7434,
+      "step": 4746
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.922449372487291e-06,
+      "loss": 0.7998,
+      "step": 4747
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.905188186918229e-06,
+      "loss": 0.7941,
+      "step": 4748
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.887964308663739e-06,
+      "loss": 0.7115,
+      "step": 4749
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8707777444107695e-06,
+      "loss": 0.7767,
+      "step": 4750
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.853628500831685e-06,
+      "loss": 0.8064,
+      "step": 4751
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.836516584584426e-06,
+      "loss": 0.8076,
+      "step": 4752
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.819442002312457e-06,
+      "loss": 0.7815,
+      "step": 4753
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8024047606446735e-06,
+      "loss": 0.3605,
+      "step": 4754
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7854048661955522e-06,
+      "loss": 0.7485,
+      "step": 4755
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7684423255650357e-06,
+      "loss": 0.7461,
+      "step": 4756
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7515171453385458e-06,
+      "loss": 0.7854,
+      "step": 4757
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7346293320870363e-06,
+      "loss": 0.8237,
+      "step": 4758
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.717778892366941e-06,
+      "loss": 0.7278,
+      "step": 4759
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.700965832720171e-06,
+      "loss": 0.7532,
+      "step": 4760
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.684190159674117e-06,
+      "loss": 0.8022,
+      "step": 4761
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6674518797417235e-06,
+      "loss": 0.7511,
+      "step": 4762
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.650750999421315e-06,
+      "loss": 0.8369,
+      "step": 4763
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.634087525196794e-06,
+      "loss": 0.7317,
+      "step": 4764
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6174614635374637e-06,
+      "loss": 0.7472,
+      "step": 4765
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6008728208981156e-06,
+      "loss": 0.7646,
+      "step": 4766
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5843216037190876e-06,
+      "loss": 0.7837,
+      "step": 4767
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.567807818426083e-06,
+      "loss": 0.8169,
+      "step": 4768
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5513314714303526e-06,
+      "loss": 0.7822,
+      "step": 4769
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.534892569128567e-06,
+      "loss": 0.714,
+      "step": 4770
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.518491117902878e-06,
+      "loss": 0.7319,
+      "step": 4771
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5021271241208907e-06,
+      "loss": 0.8315,
+      "step": 4772
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.48580059413568e-06,
+      "loss": 0.8074,
+      "step": 4773
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.469511534285752e-06,
+      "loss": 0.7098,
+      "step": 4774
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4532599508950824e-06,
+      "loss": 0.7993,
+      "step": 4775
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.437045850273113e-06,
+      "loss": 0.7542,
+      "step": 4776
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.420869238714708e-06,
+      "loss": 0.7675,
+      "step": 4777
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4047301225001547e-06,
+      "loss": 0.7708,
+      "step": 4778
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.388628507895275e-06,
+      "loss": 0.7292,
+      "step": 4779
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3725644011512124e-06,
+      "loss": 0.7742,
+      "step": 4780
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3565378085046337e-06,
+      "loss": 0.738,
+      "step": 4781
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.340548736177618e-06,
+      "loss": 0.7732,
+      "step": 4782
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.324597190377665e-06,
+      "loss": 0.7723,
+      "step": 4783
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.308683177297711e-06,
+      "loss": 0.7573,
+      "step": 4784
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2928067031161247e-06,
+      "loss": 0.7134,
+      "step": 4785
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2769677739966973e-06,
+      "loss": 0.7183,
+      "step": 4786
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2611663960886664e-06,
+      "loss": 0.7651,
+      "step": 4787
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.245402575526646e-06,
+      "loss": 0.7722,
+      "step": 4788
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.229676318430697e-06,
+      "loss": 0.7524,
+      "step": 4789
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2139876309063234e-06,
+      "loss": 0.7358,
+      "step": 4790
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.198336519044376e-06,
+      "loss": 0.7051,
+      "step": 4791
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1827229889211606e-06,
+      "loss": 0.8203,
+      "step": 4792
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1671470465984177e-06,
+      "loss": 0.7444,
+      "step": 4793
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.151608698123232e-06,
+      "loss": 0.7588,
+      "step": 4794
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1361079495281444e-06,
+      "loss": 0.7803,
+      "step": 4795
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1206448068310633e-06,
+      "loss": 0.79,
+      "step": 4796
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1052192760353314e-06,
+      "loss": 0.7769,
+      "step": 4797
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0898313631296583e-06,
+      "loss": 0.7432,
+      "step": 4798
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.074481074088165e-06,
+      "loss": 0.7822,
+      "step": 4799
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0591684148703614e-06,
+      "loss": 0.8225,
+      "step": 4800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0438933914211486e-06,
+      "loss": 0.7916,
+      "step": 4801
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.028656009670827e-06,
+      "loss": 0.8271,
+      "step": 4802
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0134562755350537e-06,
+      "loss": 0.7156,
+      "step": 4803
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.998294194914897e-06,
+      "loss": 0.7883,
+      "step": 4804
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9831697736968146e-06,
+      "loss": 0.7769,
+      "step": 4805
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9680830177525986e-06,
+      "loss": 0.7465,
+      "step": 4806
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.953033932939464e-06,
+      "loss": 0.7715,
+      "step": 4807
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9380225250999816e-06,
+      "loss": 0.8057,
+      "step": 4808
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9230488000621004e-06,
+      "loss": 0.3185,
+      "step": 4809
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.908112763639137e-06,
+      "loss": 0.8359,
+      "step": 4810
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.893214421629764e-06,
+      "loss": 0.7834,
+      "step": 4811
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8783537798180437e-06,
+      "loss": 0.7416,
+      "step": 4812
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.863530843973372e-06,
+      "loss": 0.7313,
+      "step": 4813
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.848745619850546e-06,
+      "loss": 0.7388,
+      "step": 4814
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.833998113189662e-06,
+      "loss": 0.7828,
+      "step": 4815
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.819288329716263e-06,
+      "loss": 0.7798,
+      "step": 4816
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8046162751411475e-06,
+      "loss": 0.7898,
+      "step": 4817
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7899819551605257e-06,
+      "loss": 0.7639,
+      "step": 4818
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7753853754559635e-06,
+      "loss": 0.7728,
+      "step": 4819
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.760826541694328e-06,
+      "loss": 0.7827,
+      "step": 4820
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.746305459527876e-06,
+      "loss": 0.7937,
+      "step": 4821
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7318221345941863e-06,
+      "loss": 0.7727,
+      "step": 4822
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7173765725161837e-06,
+      "loss": 0.8175,
+      "step": 4823
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7029687789021373e-06,
+      "loss": 0.3259,
+      "step": 4824
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6885987593456507e-06,
+      "loss": 0.757,
+      "step": 4825
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.67426651942565e-06,
+      "loss": 0.7632,
+      "step": 4826
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6599720647064062e-06,
+      "loss": 0.7896,
+      "step": 4827
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.645715400737536e-06,
+      "loss": 0.7876,
+      "step": 4828
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.631496533053934e-06,
+      "loss": 0.8101,
+      "step": 4829
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.617315467175885e-06,
+      "loss": 0.7378,
+      "step": 4830
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6031722086089615e-06,
+      "loss": 0.7374,
+      "step": 4831
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.589066762844039e-06,
+      "loss": 0.7668,
+      "step": 4832
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.57499913535737e-06,
+      "loss": 0.8062,
+      "step": 4833
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5609693316104744e-06,
+      "loss": 0.6937,
+      "step": 4834
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5469773570502065e-06,
+      "loss": 0.7837,
+      "step": 4835
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5330232171087433e-06,
+      "loss": 0.7537,
+      "step": 4836
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5191069172035396e-06,
+      "loss": 0.7612,
+      "step": 4837
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5052284627374077e-06,
+      "loss": 0.8152,
+      "step": 4838
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.491387859098426e-06,
+      "loss": 0.755,
+      "step": 4839
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4775851116599967e-06,
+      "loss": 0.77,
+      "step": 4840
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4638202257808107e-06,
+      "loss": 0.7656,
+      "step": 4841
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4500932068049043e-06,
+      "loss": 0.8064,
+      "step": 4842
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.436404060061548e-06,
+      "loss": 0.3203,
+      "step": 4843
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4227527908653458e-06,
+      "loss": 0.7363,
+      "step": 4844
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.409139404516203e-06,
+      "loss": 0.8271,
+      "step": 4845
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3955639062992696e-06,
+      "loss": 0.8127,
+      "step": 4846
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.382026301485074e-06,
+      "loss": 0.7266,
+      "step": 4847
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3685265953293343e-06,
+      "loss": 0.7786,
+      "step": 4848
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3550647930731364e-06,
+      "loss": 0.785,
+      "step": 4849
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3416408999427875e-06,
+      "loss": 0.7524,
+      "step": 4850
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3282549211499304e-06,
+      "loss": 0.7482,
+      "step": 4851
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3149068618914415e-06,
+      "loss": 0.8003,
+      "step": 4852
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3015967273494864e-06,
+      "loss": 0.3594,
+      "step": 4853
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.288324522691565e-06,
+      "loss": 0.856,
+      "step": 4854
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2750902530703666e-06,
+      "loss": 0.7183,
+      "step": 4855
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2618939236238923e-06,
+      "loss": 0.792,
+      "step": 4856
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.248735539475433e-06,
+      "loss": 0.8008,
+      "step": 4857
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2356151057334906e-06,
+      "loss": 0.7932,
+      "step": 4858
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2225326274919133e-06,
+      "loss": 0.7955,
+      "step": 4859
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.209488109829727e-06,
+      "loss": 0.7915,
+      "step": 4860
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.196481557811303e-06,
+      "loss": 0.7262,
+      "step": 4861
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1835129764861906e-06,
+      "loss": 0.7334,
+      "step": 4862
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1705823708892734e-06,
+      "loss": 0.8,
+      "step": 4863
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1576897460406474e-06,
+      "loss": 0.7725,
+      "step": 4864
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.144835106945664e-06,
+      "loss": 0.7323,
+      "step": 4865
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1320184585949533e-06,
+      "loss": 0.7437,
+      "step": 4866
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.119239805964357e-06,
+      "loss": 0.7925,
+      "step": 4867
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.106499154015018e-06,
+      "loss": 0.76,
+      "step": 4868
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0937965076932574e-06,
+      "loss": 0.825,
+      "step": 4869
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0811318719307193e-06,
+      "loss": 0.7781,
+      "step": 4870
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.068505251644237e-06,
+      "loss": 0.7937,
+      "step": 4871
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0559166517358785e-06,
+      "loss": 0.7815,
+      "step": 4872
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0433660770930007e-06,
+      "loss": 0.7634,
+      "step": 4873
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.030853532588162e-06,
+      "loss": 0.7856,
+      "step": 4874
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0183790230791534e-06,
+      "loss": 0.7695,
+      "step": 4875
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0059425534090125e-06,
+      "loss": 0.8491,
+      "step": 4876
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.993544128406e-06,
+      "loss": 0.8118,
+      "step": 4877
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.981183752883631e-06,
+      "loss": 0.8308,
+      "step": 4878
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9688614316406007e-06,
+      "loss": 0.7734,
+      "step": 4879
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9565771694608936e-06,
+      "loss": 0.6978,
+      "step": 4880
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.944330971113639e-06,
+      "loss": 0.363,
+      "step": 4881
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9321228413532787e-06,
+      "loss": 0.8005,
+      "step": 4882
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.91995278491941e-06,
+      "loss": 0.7277,
+      "step": 4883
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.907820806536842e-06,
+      "loss": 0.8083,
+      "step": 4884
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8957269109156628e-06,
+      "loss": 0.8164,
+      "step": 4885
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8836711027511277e-06,
+      "loss": 0.7203,
+      "step": 4886
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8716533867237152e-06,
+      "loss": 0.3091,
+      "step": 4887
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.859673767499115e-06,
+      "loss": 0.7646,
+      "step": 4888
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8477322497282178e-06,
+      "loss": 0.6858,
+      "step": 4889
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.83582883804716e-06,
+      "loss": 0.8034,
+      "step": 4890
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8239635370772223e-06,
+      "loss": 0.3502,
+      "step": 4891
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8121363514249535e-06,
+      "loss": 0.3442,
+      "step": 4892
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8003472856820469e-06,
+      "loss": 0.7848,
+      "step": 4893
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7885963444254528e-06,
+      "loss": 0.7715,
+      "step": 4894
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.776883532217255e-06,
+      "loss": 0.7678,
+      "step": 4895
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7652088536048051e-06,
+      "loss": 0.7974,
+      "step": 4896
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7535723131206106e-06,
+      "loss": 0.3055,
+      "step": 4897
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7419739152823467e-06,
+      "loss": 0.7295,
+      "step": 4898
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7304136645929447e-06,
+      "loss": 0.7551,
+      "step": 4899
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7188915655404813e-06,
+      "loss": 0.7927,
+      "step": 4900
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7074076225982228e-06,
+      "loss": 0.8125,
+      "step": 4901
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6959618402246357e-06,
+      "loss": 0.7336,
+      "step": 4902
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.684554222863377e-06,
+      "loss": 0.804,
+      "step": 4903
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6731847749432705e-06,
+      "loss": 0.7582,
+      "step": 4904
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6618535008783075e-06,
+      "loss": 0.769,
+      "step": 4905
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.650560405067725e-06,
+      "loss": 0.7876,
+      "step": 4906
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6393054918958373e-06,
+      "loss": 0.8491,
+      "step": 4907
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6280887657322275e-06,
+      "loss": 0.7571,
+      "step": 4908
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.616910230931612e-06,
+      "loss": 0.802,
+      "step": 4909
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6057698918338526e-06,
+      "loss": 0.7867,
+      "step": 4910
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5946677527640564e-06,
+      "loss": 0.8186,
+      "step": 4911
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5836038180324198e-06,
+      "loss": 0.7908,
+      "step": 4912
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5725780919343625e-06,
+      "loss": 0.7471,
+      "step": 4913
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.561590578750438e-06,
+      "loss": 0.7795,
+      "step": 4914
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.55064128274639e-06,
+      "loss": 0.7302,
+      "step": 4915
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5397302081731068e-06,
+      "loss": 0.8416,
+      "step": 4916
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5288573592666444e-06,
+      "loss": 0.7063,
+      "step": 4917
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.518022740248215e-06,
+      "loss": 0.7438,
+      "step": 4918
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5072263553241872e-06,
+      "loss": 0.7585,
+      "step": 4919
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4964682086861082e-06,
+      "loss": 0.7733,
+      "step": 4920
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4857483045106257e-06,
+      "loss": 0.7578,
+      "step": 4921
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4750666469596108e-06,
+      "loss": 0.7932,
+      "step": 4922
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.464423240180035e-06,
+      "loss": 0.824,
+      "step": 4923
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4538180883040265e-06,
+      "loss": 0.7715,
+      "step": 4924
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4432511954488914e-06,
+      "loss": 0.8032,
+      "step": 4925
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4327225657170484e-06,
+      "loss": 0.7703,
+      "step": 4926
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4222322031960722e-06,
+      "loss": 0.7767,
+      "step": 4927
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4117801119586938e-06,
+      "loss": 0.7961,
+      "step": 4928
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.401366296062756e-06,
+      "loss": 0.3162,
+      "step": 4929
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3909907595512805e-06,
+      "loss": 0.8164,
+      "step": 4930
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3806535064524006e-06,
+      "loss": 0.79,
+      "step": 4931
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.370354540779395e-06,
+      "loss": 0.7949,
+      "step": 4932
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.360093866530665e-06,
+      "loss": 0.7827,
+      "step": 4933
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.34987148768978e-06,
+      "loss": 0.7805,
+      "step": 4934
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3396874082253986e-06,
+      "loss": 0.7886,
+      "step": 4935
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3295416320913357e-06,
+      "loss": 0.7864,
+      "step": 4936
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3194341632265516e-06,
+      "loss": 0.7627,
+      "step": 4937
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3093650055550854e-06,
+      "loss": 0.7413,
+      "step": 4938
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2993341629861433e-06,
+      "loss": 0.7388,
+      "step": 4939
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2893416394140323e-06,
+      "loss": 0.8047,
+      "step": 4940
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.279387438718216e-06,
+      "loss": 0.7896,
+      "step": 4941
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.269471564763247e-06,
+      "loss": 0.7996,
+      "step": 4942
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2595940213988022e-06,
+      "loss": 0.7534,
+      "step": 4943
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2497548124597024e-06,
+      "loss": 0.7705,
+      "step": 4944
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2399539417658369e-06,
+      "loss": 0.8572,
+      "step": 4945
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2301914131222726e-06,
+      "loss": 0.7443,
+      "step": 4946
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2204672303191334e-06,
+      "loss": 0.7747,
+      "step": 4947
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2107813971317105e-06,
+      "loss": 0.7909,
+      "step": 4948
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2011339173203628e-06,
+      "loss": 0.8037,
+      "step": 4949
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1915247946305496e-06,
+      "loss": 0.7764,
+      "step": 4950
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.181954032792909e-06,
+      "loss": 0.7407,
+      "step": 4951
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1724216355231022e-06,
+      "loss": 0.7969,
+      "step": 4952
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1629276065219575e-06,
+      "loss": 0.7339,
+      "step": 4953
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.153471949475382e-06,
+      "loss": 0.7917,
+      "step": 4954
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1440546680543728e-06,
+      "loss": 0.7761,
+      "step": 4955
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1346757659150497e-06,
+      "loss": 0.8241,
+      "step": 4956
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1253352466986334e-06,
+      "loss": 0.7612,
+      "step": 4957
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1160331140314339e-06,
+      "loss": 0.8105,
+      "step": 4958
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1067693715248406e-06,
+      "loss": 0.7444,
+      "step": 4959
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0975440227753764e-06,
+      "loss": 0.7988,
+      "step": 4960
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0883570713646319e-06,
+      "loss": 0.7896,
+      "step": 4961
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0792085208593094e-06,
+      "loss": 0.6929,
+      "step": 4962
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0700983748111793e-06,
+      "loss": 0.7942,
+      "step": 4963
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0610266367571009e-06,
+      "loss": 0.7446,
+      "step": 4964
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0519933102190683e-06,
+      "loss": 0.7535,
+      "step": 4965
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0429983987041092e-06,
+      "loss": 0.719,
+      "step": 4966
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0340419057043526e-06,
+      "loss": 0.7585,
+      "step": 4967
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0251238346970393e-06,
+      "loss": 0.7617,
+      "step": 4968
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0162441891444441e-06,
+      "loss": 0.7574,
+      "step": 4969
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.007402972493976e-06,
+      "loss": 0.7583,
+      "step": 4970
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.986001881780782e-07,
+      "loss": 0.812,
+      "step": 4971
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.898358396143171e-07,
+      "loss": 0.7446,
+      "step": 4972
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.811099302052929e-07,
+      "loss": 0.7612,
+      "step": 4973
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.72422463338718e-07,
+      "loss": 0.793,
+      "step": 4974
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.637734423873613e-07,
+      "loss": 0.3362,
+      "step": 4975
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.551628707090809e-07,
+      "loss": 0.8008,
+      "step": 4976
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.465907516467698e-07,
+      "loss": 0.824,
+      "step": 4977
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.380570885284545e-07,
+      "loss": 0.7659,
+      "step": 4978
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.295618846671738e-07,
+      "loss": 0.7849,
+      "step": 4979
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.211051433610674e-07,
+      "loss": 0.7959,
+      "step": 4980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.126868678933198e-07,
+      "loss": 0.8083,
+      "step": 4981
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.043070615322169e-07,
+      "loss": 0.812,
+      "step": 4982
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.959657275310674e-07,
+      "loss": 0.7434,
+      "step": 4983
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.876628691282918e-07,
+      "loss": 0.7783,
+      "step": 4984
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.793984895473117e-07,
+      "loss": 0.7729,
+      "step": 4985
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.711725919966718e-07,
+      "loss": 0.8236,
+      "step": 4986
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.629851796699284e-07,
+      "loss": 0.3015,
+      "step": 4987
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.548362557457279e-07,
+      "loss": 0.8218,
+      "step": 4988
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.467258233877728e-07,
+      "loss": 0.6875,
+      "step": 4989
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.386538857447779e-07,
+      "loss": 0.7365,
+      "step": 4990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.306204459505806e-07,
+      "loss": 0.8003,
+      "step": 4991
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.226255071240308e-07,
+      "loss": 0.7781,
+      "step": 4992
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.146690723690342e-07,
+      "loss": 0.7781,
+      "step": 4993
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.067511447745535e-07,
+      "loss": 0.7944,
+      "step": 4994
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.988717274146074e-07,
+      "loss": 0.7506,
+      "step": 4995
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.910308233482488e-07,
+      "loss": 0.3518,
+      "step": 4996
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.832284356195762e-07,
+      "loss": 0.7903,
+      "step": 4997
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.754645672577776e-07,
+      "loss": 0.7714,
+      "step": 4998
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.677392212770196e-07,
+      "loss": 0.7996,
+      "step": 4999
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.600524006765808e-07,
+      "loss": 0.7844,
+      "step": 5000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.524041084407185e-07,
+      "loss": 0.7544,
+      "step": 5001
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.447943475387797e-07,
+      "loss": 0.7932,
+      "step": 5002
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.372231209251346e-07,
+      "loss": 0.7565,
+      "step": 5003
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.296904315391873e-07,
+      "loss": 0.7271,
+      "step": 5004
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.221962823053874e-07,
+      "loss": 0.7595,
+      "step": 5005
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.147406761332298e-07,
+      "loss": 0.7903,
+      "step": 5006
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.073236159172325e-07,
+      "loss": 0.7686,
+      "step": 5007
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.999451045369587e-07,
+      "loss": 0.7595,
+      "step": 5008
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.926051448569948e-07,
+      "loss": 0.8274,
+      "step": 5009
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.853037397269724e-07,
+      "loss": 0.7412,
+      "step": 5010
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.78040891981524e-07,
+      "loss": 0.8953,
+      "step": 5011
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.70816604440383e-07,
+      "loss": 0.7743,
+      "step": 5012
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.63630879908217e-07,
+      "loss": 0.7891,
+      "step": 5013
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.564837211748054e-07,
+      "loss": 0.7927,
+      "step": 5014
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.493751310149176e-07,
+      "loss": 0.3232,
+      "step": 5015
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.42305112188335e-07,
+      "loss": 0.7947,
+      "step": 5016
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.352736674398951e-07,
+      "loss": 0.7703,
+      "step": 5017
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.282807994994478e-07,
+      "loss": 0.8503,
+      "step": 5018
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.213265110818655e-07,
+      "loss": 0.7615,
+      "step": 5019
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.144108048870333e-07,
+      "loss": 0.7607,
+      "step": 5020
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.075336835998812e-07,
+      "loss": 0.7568,
+      "step": 5021
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.00695149890329e-07,
+      "loss": 0.7855,
+      "step": 5022
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.938952064133418e-07,
+      "loss": 0.7756,
+      "step": 5023
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.871338558088857e-07,
+      "loss": 0.759,
+      "step": 5024
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.804111007019497e-07,
+      "loss": 0.7369,
+      "step": 5025
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.737269437025461e-07,
+      "loss": 0.825,
+      "step": 5026
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.67081387405688e-07,
+      "loss": 0.7999,
+      "step": 5027
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.604744343914115e-07,
+      "loss": 0.7811,
+      "step": 5028
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.539060872247537e-07,
+      "loss": 0.739,
+      "step": 5029
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.47376348455797e-07,
+      "loss": 0.7898,
+      "step": 5030
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.408852206195913e-07,
+      "loss": 0.738,
+      "step": 5031
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.344327062362098e-07,
+      "loss": 0.7876,
+      "step": 5032
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.280188078107595e-07,
+      "loss": 0.2898,
+      "step": 5033
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.216435278333376e-07,
+      "loss": 0.7687,
+      "step": 5034
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.153068687790197e-07,
+      "loss": 0.729,
+      "step": 5035
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.09008833107949e-07,
+      "loss": 0.7676,
+      "step": 5036
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.027494232652142e-07,
+      "loss": 0.7593,
+      "step": 5037
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9652864168096e-07,
+      "loss": 0.7854,
+      "step": 5038
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.90346490770277e-07,
+      "loss": 0.7485,
+      "step": 5039
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.84202972933312e-07,
+      "loss": 0.7298,
+      "step": 5040
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.78098090555179e-07,
+      "loss": 0.7773,
+      "step": 5041
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.720318460060047e-07,
+      "loss": 0.7908,
+      "step": 5042
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6600424164091605e-07,
+      "loss": 0.7733,
+      "step": 5043
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6001527980004124e-07,
+      "loss": 0.7854,
+      "step": 5044
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.54064962808487e-07,
+      "loss": 0.7905,
+      "step": 5045
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.481532929763943e-07,
+      "loss": 0.8137,
+      "step": 5046
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.422802725988606e-07,
+      "loss": 0.7915,
+      "step": 5047
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.364459039559843e-07,
+      "loss": 0.801,
+      "step": 5048
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.306501893128978e-07,
+      "loss": 0.7812,
+      "step": 5049
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2489313091967906e-07,
+      "loss": 0.7605,
+      "step": 5050
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.191747310114069e-07,
+      "loss": 0.793,
+      "step": 5051
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.134949918081832e-07,
+      "loss": 0.7595,
+      "step": 5052
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0785391551506623e-07,
+      "loss": 0.8259,
+      "step": 5053
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0225150432211535e-07,
+      "loss": 0.7874,
+      "step": 5054
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9668776040437947e-07,
+      "loss": 0.7982,
+      "step": 5055
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9116268592189755e-07,
+      "loss": 0.7437,
+      "step": 5056
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8567628301969803e-07,
+      "loss": 0.7675,
+      "step": 5057
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8022855382777724e-07,
+      "loss": 0.8167,
+      "step": 5058
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7481950046115433e-07,
+      "loss": 0.7798,
+      "step": 5059
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.69449125019794e-07,
+      "loss": 0.741,
+      "step": 5060
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.641174295886618e-07,
+      "loss": 0.6854,
+      "step": 5061
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.588244162377019e-07,
+      "loss": 0.7747,
+      "step": 5062
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5357008702185944e-07,
+      "loss": 0.8237,
+      "step": 5063
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.483544439810249e-07,
+      "loss": 0.8171,
+      "step": 5064
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.431774891401118e-07,
+      "loss": 0.7871,
+      "step": 5065
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.380392245089792e-07,
+      "loss": 0.7646,
+      "step": 5066
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.329396520824757e-07,
+      "loss": 0.8236,
+      "step": 5067
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2787877384045096e-07,
+      "loss": 0.7371,
+      "step": 5068
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.228565917476889e-07,
+      "loss": 0.7847,
+      "step": 5069
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.178731077539743e-07,
+      "loss": 0.8145,
+      "step": 5070
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1292832379409274e-07,
+      "loss": 0.7772,
+      "step": 5071
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.080222417877421e-07,
+      "loss": 0.7791,
+      "step": 5072
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.031548636396764e-07,
+      "loss": 0.8108,
+      "step": 5073
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9832619123953965e-07,
+      "loss": 0.8098,
+      "step": 5074
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9353622646199897e-07,
+      "loss": 0.7391,
+      "step": 5075
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8878497116671123e-07,
+      "loss": 0.7959,
+      "step": 5076
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8407242719823424e-07,
+      "loss": 0.7771,
+      "step": 5077
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.793985963861712e-07,
+      "loss": 0.351,
+      "step": 5078
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.747634805450483e-07,
+      "loss": 0.7949,
+      "step": 5079
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7016708147439286e-07,
+      "loss": 0.7883,
+      "step": 5080
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.656094009586663e-07,
+      "loss": 0.7489,
+      "step": 5081
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6109044076733093e-07,
+      "loss": 0.7673,
+      "step": 5082
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5661020265479454e-07,
+      "loss": 0.7681,
+      "step": 5083
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.521686883604324e-07,
+      "loss": 0.8162,
+      "step": 5084
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4776589960862075e-07,
+      "loss": 0.8252,
+      "step": 5085
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.434018381086589e-07,
+      "loss": 0.8325,
+      "step": 5086
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3907650555481385e-07,
+      "loss": 0.7354,
+      "step": 5087
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3478990362634234e-07,
+      "loss": 0.7427,
+      "step": 5088
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3054203398743534e-07,
+      "loss": 0.7188,
+      "step": 5089
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2633289828729586e-07,
+      "loss": 0.7954,
+      "step": 5090
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.221624981600168e-07,
+      "loss": 0.7649,
+      "step": 5091
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1803083522471402e-07,
+      "loss": 0.772,
+      "step": 5092
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.139379110854267e-07,
+      "loss": 0.7266,
+      "step": 5093
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0988372733118379e-07,
+      "loss": 0.7632,
+      "step": 5094
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.058682855359595e-07,
+      "loss": 0.8076,
+      "step": 5095
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0189158725867352e-07,
+      "loss": 0.7664,
+      "step": 5096
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.979536340432131e-07,
+      "loss": 0.7479,
+      "step": 5097
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9405442741844416e-07,
+      "loss": 0.8477,
+      "step": 5098
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9019396889816688e-07,
+      "loss": 0.8186,
+      "step": 5099
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8637225998114906e-07,
+      "loss": 0.7863,
+      "step": 5100
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.825893021510927e-07,
+      "loss": 0.7959,
+      "step": 5101
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.788450968766897e-07,
+      "loss": 0.7582,
+      "step": 5102
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7513964561156615e-07,
+      "loss": 0.821,
+      "step": 5103
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7147294979429352e-07,
+      "loss": 0.814,
+      "step": 5104
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6784501084843308e-07,
+      "loss": 0.8638,
+      "step": 5105
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6425583018244706e-07,
+      "loss": 0.738,
+      "step": 5106
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.607054091897986e-07,
+      "loss": 0.7537,
+      "step": 5107
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.57193749248874e-07,
+      "loss": 0.832,
+      "step": 5108
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5372085172302708e-07,
+      "loss": 0.73,
+      "step": 5109
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5028671796055715e-07,
+      "loss": 0.804,
+      "step": 5110
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4689134929470882e-07,
+      "loss": 0.7308,
+      "step": 5111
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4353474704368319e-07,
+      "loss": 0.8572,
+      "step": 5112
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4021691251062673e-07,
+      "loss": 0.7903,
+      "step": 5113
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3693784698363133e-07,
+      "loss": 0.7272,
+      "step": 5114
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3369755173575637e-07,
+      "loss": 0.7817,
+      "step": 5115
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.304960280249845e-07,
+      "loss": 0.3546,
+      "step": 5116
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.273332770942659e-07,
+      "loss": 0.7861,
+      "step": 5117
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2420930017148502e-07,
+      "loss": 0.77,
+      "step": 5118
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.211240984694717e-07,
+      "loss": 0.7538,
+      "step": 5119
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1807767318602336e-07,
+      "loss": 0.7761,
+      "step": 5120
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1507002550386059e-07,
+      "loss": 0.8149,
+      "step": 5121
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1210115659063825e-07,
+      "loss": 0.7407,
+      "step": 5122
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0917106759900097e-07,
+      "loss": 0.7404,
+      "step": 5123
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.062797596664944e-07,
+      "loss": 0.7399,
+      "step": 5124
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.034272339156206e-07,
+      "loss": 0.6935,
+      "step": 5125
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0061349145383813e-07,
+      "loss": 0.7927,
+      "step": 5126
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.783853337353988e-08,
+      "loss": 0.8289,
+      "step": 5127
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.510236075205292e-08,
+      "loss": 0.7766,
+      "step": 5128
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.240497465164976e-08,
+      "loss": 0.822,
+      "step": 5129
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.974637611955938e-08,
+      "loss": 0.7966,
+      "step": 5130
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.712656618793391e-08,
+      "loss": 0.7544,
+      "step": 5131
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.454554587388197e-08,
+      "loss": 0.7823,
+      "step": 5132
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.200331617943536e-08,
+      "loss": 0.7822,
+      "step": 5133
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.949987809158233e-08,
+      "loss": 0.8135,
+      "step": 5134
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.703523258223432e-08,
+      "loss": 0.7562,
+      "step": 5135
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.460938060825929e-08,
+      "loss": 0.8303,
+      "step": 5136
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.222232311145937e-08,
+      "loss": 0.8352,
+      "step": 5137
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.987406101855998e-08,
+      "loss": 0.7671,
+      "step": 5138
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.756459524125402e-08,
+      "loss": 0.7756,
+      "step": 5139
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.529392667613543e-08,
+      "loss": 0.8792,
+      "step": 5140
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.306205620477678e-08,
+      "loss": 0.7581,
+      "step": 5141
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.086898469365165e-08,
+      "loss": 0.7618,
+      "step": 5142
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.8714712994190067e-08,
+      "loss": 0.8088,
+      "step": 5143
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.659924194276745e-08,
+      "loss": 0.8022,
+      "step": 5144
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.4522572360660164e-08,
+      "loss": 0.7935,
+      "step": 5145
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.2484705054123286e-08,
+      "loss": 0.8179,
+      "step": 5146
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.048564081431284e-08,
+      "loss": 0.722,
+      "step": 5147
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.8525380417330236e-08,
+      "loss": 0.7791,
+      "step": 5148
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6603924624244456e-08,
+      "loss": 0.7379,
+      "step": 5149
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.472127418099214e-08,
+      "loss": 0.7762,
+      "step": 5150
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.287742981851084e-08,
+      "loss": 0.8005,
+      "step": 5151
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1072392252639036e-08,
+      "loss": 0.7008,
+      "step": 5152
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.930616218414951e-08,
+      "loss": 0.7061,
+      "step": 5153
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7578740298749306e-08,
+      "loss": 0.7952,
+      "step": 5154
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5890127267090846e-08,
+      "loss": 0.7709,
+      "step": 5155
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.424032374476083e-08,
+      "loss": 0.7842,
+      "step": 5156
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.262933037224691e-08,
+      "loss": 0.7879,
+      "step": 5157
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.105714777501545e-08,
+      "loss": 0.7466,
+      "step": 5158
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9523776563422644e-08,
+      "loss": 0.7349,
+      "step": 5159
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8029217332781188e-08,
+      "loss": 0.7905,
+      "step": 5160
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.657347066333804e-08,
+      "loss": 0.7834,
+      "step": 5161
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5156537120263335e-08,
+      "loss": 0.6881,
+      "step": 5162
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3778417253650376e-08,
+      "loss": 0.7864,
+      "step": 5163
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.243911159853784e-08,
+      "loss": 0.8149,
+      "step": 5164
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.113862067488759e-08,
+      "loss": 0.3363,
+      "step": 5165
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9876944987606838e-08,
+      "loss": 0.7488,
+      "step": 5166
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8654085026503787e-08,
+      "loss": 0.8252,
+      "step": 5167
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.747004126635421e-08,
+      "loss": 0.7893,
+      "step": 5168
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6324814166823744e-08,
+      "loss": 0.7454,
+      "step": 5169
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5218404172545608e-08,
+      "loss": 0.7368,
+      "step": 5170
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.415081171305399e-08,
+      "loss": 0.7542,
+      "step": 5171
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3122037202828452e-08,
+      "loss": 0.7573,
+      "step": 5172
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2132081041282827e-08,
+      "loss": 0.7577,
+      "step": 5173
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1180943612754124e-08,
+      "loss": 0.8025,
+      "step": 5174
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0268625286491418e-08,
+      "loss": 0.746,
+      "step": 5175
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.39512641668916e-09,
+      "loss": 0.7583,
+      "step": 5176
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.560447342487176e-09,
+      "loss": 0.7324,
+      "step": 5177
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.76458838791516e-09,
+      "loss": 0.7781,
+      "step": 5178
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.0075498619703865e-09,
+      "loss": 0.7657,
+      "step": 5179
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.28933205855109e-09,
+      "loss": 0.7903,
+      "step": 5180
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.609935256500886e-09,
+      "loss": 0.7367,
+      "step": 5181
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.9693597195865634e-09,
+      "loss": 0.7969,
+      "step": 5182
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.367605696486976e-09,
+      "loss": 0.8192,
+      "step": 5183
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.8046734208374565e-09,
+      "loss": 0.8357,
+      "step": 5184
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2805631111743062e-09,
+      "loss": 0.783,
+      "step": 5185
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7952749710014047e-09,
+      "loss": 0.7573,
+      "step": 5186
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.348809188690293e-09,
+      "loss": 0.8369,
+      "step": 5187
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.941165937602296e-09,
+      "loss": 0.7666,
+      "step": 5188
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5723453759886041e-09,
+      "loss": 0.77,
+      "step": 5189
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2423476470346806e-09,
+      "loss": 0.7549,
+      "step": 5190
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.511728788602625e-10,
+      "loss": 0.8169,
+      "step": 5191
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.988211845082582e-10,
+      "loss": 0.3004,
+      "step": 5192
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.852926619447473e-10,
+      "loss": 0.8069,
+      "step": 5193
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1058739408118544e-10,
+      "loss": 0.8689,
+      "step": 5194
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7470544874109706e-10,
+      "loss": 0.7932,
+      "step": 5195
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.764687866007592e-11,
+      "loss": 0.7651,
+      "step": 5196
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.941172155239812e-11,
+      "loss": 0.8051,
+      "step": 5197
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.5033,
+      "step": 5198
+    },
+    {
+      "epoch": 1.0,
+      "step": 5198,
+      "total_flos": 7567157260779520.0,
+      "train_loss": 0.7976963913592067,
+      "train_runtime": 39477.0228,
+      "train_samples_per_second": 16.853,
+      "train_steps_per_second": 0.132
+    }
+  ],
+  "max_steps": 5198,
+  "num_train_epochs": 1,
+  "total_flos": 7567157260779520.0,
+  "trial_name": null,
+  "trial_params": null
+}